public inbox for [email protected]help / color / mirror / Atom feed
Fix error handling in be_tls_open_server() 37+ messages / 5 participants [nested] [flat]
* Fix error handling in be_tls_open_server() @ 2023-08-01 14:44 Sergey Shinderuk <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Sergey Shinderuk @ 2023-08-01 14:44 UTC (permalink / raw) To: pgsql-hackers Hi, A static analyzer reported a possible pfree(NULL) in be_tls_open_server(). Here is a fix. Also handle an error from X509_NAME_print_ex(). AFAICS, the error "SSL certificate's distinguished name contains embedded null" could not be reached at all, because XN_FLAG_RFC2253 passed to X509_NAME_print_ex() ensures that null bytes are escaped. Best regards, -- Sergey Shinderuk https://postgrespro.com/ diff --git a/src/backend/libpq/be-secure-openssl.c b/src/backend/libpq/be-secure-openssl.c index 658b09988d6..31b6a6eacdf 100644 --- a/src/backend/libpq/be-secure-openssl.c +++ b/src/backend/libpq/be-secure-openssl.c @@ -620,8 +620,11 @@ aloop: bio = BIO_new(BIO_s_mem()); if (!bio) { - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } @@ -632,12 +635,15 @@ aloop: * which make regular expression matching a bit easier. Also note that * it prints the Subject fields in reverse order. */ - X509_NAME_print_ex(bio, x509name, 0, XN_FLAG_RFC2253); - if (BIO_get_mem_ptr(bio, &bio_buf) <= 0) + if (X509_NAME_print_ex(bio, x509name, 0, XN_FLAG_RFC2253) == -1 || + BIO_get_mem_ptr(bio, &bio_buf) <= 0) { BIO_free(bio); - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } peer_dn = MemoryContextAlloc(TopMemoryContext, bio_buf->length + 1); @@ -651,8 +657,11 @@ aloop: (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("SSL certificate's distinguished name contains embedded null"))); pfree(peer_dn); - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } Attachments: [text/plain] pfree-peer-cn-fix.diff (1.4K, ../../[email protected]/2-pfree-peer-cn-fix.diff) download | inline diff: diff --git a/src/backend/libpq/be-secure-openssl.c b/src/backend/libpq/be-secure-openssl.c index 658b09988d6..31b6a6eacdf 100644 --- a/src/backend/libpq/be-secure-openssl.c +++ b/src/backend/libpq/be-secure-openssl.c @@ -620,8 +620,11 @@ aloop: bio = BIO_new(BIO_s_mem()); if (!bio) { - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } @@ -632,12 +635,15 @@ aloop: * which make regular expression matching a bit easier. Also note that * it prints the Subject fields in reverse order. */ - X509_NAME_print_ex(bio, x509name, 0, XN_FLAG_RFC2253); - if (BIO_get_mem_ptr(bio, &bio_buf) <= 0) + if (X509_NAME_print_ex(bio, x509name, 0, XN_FLAG_RFC2253) == -1 || + BIO_get_mem_ptr(bio, &bio_buf) <= 0) { BIO_free(bio); - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } peer_dn = MemoryContextAlloc(TopMemoryContext, bio_buf->length + 1); @@ -651,8 +657,11 @@ aloop: (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("SSL certificate's distinguished name contains embedded null"))); pfree(peer_dn); - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-23 13:23 Daniel Gustafsson <[email protected]> parent: Sergey Shinderuk <[email protected]> 0 siblings, 2 replies; 37+ messages in thread From: Daniel Gustafsson @ 2023-08-23 13:23 UTC (permalink / raw) To: Sergey Shinderuk <[email protected]>; +Cc: pgsql-hackers > On 1 Aug 2023, at 16:44, Sergey Shinderuk <[email protected]> wrote: > A static analyzer reported a possible pfree(NULL) in be_tls_open_server(). This has the smell of a theoretical problem, I can't really imagine a certificate where which would produce this. Have you been able to trigger it? Wouldn't a better fix be to error out on len == -1 as in the attached, maybe with a "Shouldn't happen" comment? -- Daniel Gustafsson Attachments: [application/octet-stream] X509_NAME_get_text_by_NID.diff (1.8K, ../../[email protected]/2-X509_NAME_get_text_by_NID.diff) download | inline diff: diff --git a/src/backend/libpq/be-secure-openssl.c b/src/backend/libpq/be-secure-openssl.c index 658b09988d..de3ebb344f 100644 --- a/src/backend/libpq/be-secure-openssl.c +++ b/src/backend/libpq/be-secure-openssl.c @@ -584,39 +584,38 @@ aloop: char *peer_dn; BIO *bio = NULL; BUF_MEM *bio_buf = NULL; + char *peer_cn; len = X509_NAME_get_text_by_NID(x509name, NID_commonName, NULL, 0); - if (len != -1) - { - char *peer_cn; - - peer_cn = MemoryContextAlloc(TopMemoryContext, len + 1); - r = X509_NAME_get_text_by_NID(x509name, NID_commonName, peer_cn, - len + 1); - peer_cn[len] = '\0'; - if (r != len) - { - /* shouldn't happen */ - pfree(peer_cn); - return -1; - } + if (len == -1) + return -1; - /* - * Reject embedded NULLs in certificate common name to prevent - * attacks like CVE-2009-4034. - */ - if (len != strlen(peer_cn)) - { - ereport(COMMERROR, - (errcode(ERRCODE_PROTOCOL_VIOLATION), - errmsg("SSL certificate's common name contains embedded null"))); - pfree(peer_cn); - return -1; - } + peer_cn = MemoryContextAlloc(TopMemoryContext, len + 1); + r = X509_NAME_get_text_by_NID(x509name, NID_commonName, peer_cn, + len + 1); + peer_cn[len] = '\0'; + if (r != len) + { + /* shouldn't happen */ + pfree(peer_cn); + return -1; + } - port->peer_cn = peer_cn; + /* + * Reject embedded NULLs in certificate common name to prevent + * attacks like CVE-2009-4034. + */ + if (len != strlen(peer_cn)) + { + ereport(COMMERROR, + (errcode(ERRCODE_PROTOCOL_VIOLATION), + errmsg("SSL certificate's common name contains embedded null"))); + pfree(peer_cn); + return -1; } + port->peer_cn = peer_cn; + bio = BIO_new(BIO_s_mem()); if (!bio) { ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-23 21:47 Jacob Champion <[email protected]> parent: Daniel Gustafsson <[email protected]> 1 sibling, 1 reply; 37+ messages in thread From: Jacob Champion @ 2023-08-23 21:47 UTC (permalink / raw) To: Daniel Gustafsson <[email protected]>; +Cc: Sergey Shinderuk <[email protected]>; pgsql-hackers On Wed, Aug 23, 2023 at 6:23 AM Daniel Gustafsson <[email protected]> wrote: > This has the smell of a theoretical problem, I can't really imagine a > certificate where which would produce this. Have you been able to trigger it? > > Wouldn't a better fix be to error out on len == -1 as in the attached, maybe > with a "Shouldn't happen" comment? Using "cert clientname=DN" in the HBA should let you issue Subjects without Common Names. Or, if you're using a certificate to authorize the connection rather than authenticate the user (for example "scram-sha-256 clientcert=verify-ca" in your HBA), then the certs you distribute could even be SAN-only with a completely empty Subject. --Jacob ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-23 21:54 Daniel Gustafsson <[email protected]> parent: Jacob Champion <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Daniel Gustafsson @ 2023-08-23 21:54 UTC (permalink / raw) To: Jacob Champion <[email protected]>; +Cc: Sergey Shinderuk <[email protected]>; pgsql-hackers > On 23 Aug 2023, at 23:47, Jacob Champion <[email protected]> wrote: > > On Wed, Aug 23, 2023 at 6:23 AM Daniel Gustafsson <[email protected]> wrote: >> This has the smell of a theoretical problem, I can't really imagine a >> certificate where which would produce this. Have you been able to trigger it? >> >> Wouldn't a better fix be to error out on len == -1 as in the attached, maybe >> with a "Shouldn't happen" comment? > > Using "cert clientname=DN" in the HBA should let you issue Subjects > without Common Names. Or, if you're using a certificate to authorize > the connection rather than authenticate the user (for example > "scram-sha-256 clientcert=verify-ca" in your HBA), then the certs you > distribute could even be SAN-only with a completely empty Subject. That's a good point, didn't think about those. In those cases the original patch by Sergey seems along the right lines. -- Daniel Gustafsson ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-24 08:11 Sergey Shinderuk <[email protected]> parent: Daniel Gustafsson <[email protected]> 1 sibling, 1 reply; 37+ messages in thread From: Sergey Shinderuk @ 2023-08-24 08:11 UTC (permalink / raw) To: Daniel Gustafsson <[email protected]>; +Cc: pgsql-hackers On 23.08.2023 16:23, Daniel Gustafsson wrote: >> On 1 Aug 2023, at 16:44, Sergey Shinderuk <[email protected]> wrote: > >> A static analyzer reported a possible pfree(NULL) in be_tls_open_server(). > > This has the smell of a theoretical problem, I can't really imagine a > certificate where which would produce this. Have you been able to trigger it? I triggered a crash by generating a certificate without a CN and forcing malloc to return NULL when called from X509_NAME_print_ex or BIO_get_mem_ptr with gdb. Initially I tried to trigger a crash by generating a certificate without a CN and with a DN contaning the null byte. But as I said, the error condition "SSL certificate's distinguished name contains embedded null" isn't really reachable, because XN_FLAG_RFC2253 escapes null bytes. -- Sergey Shinderuk https://postgrespro.com/ ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-24 08:38 Daniel Gustafsson <[email protected]> parent: Sergey Shinderuk <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Daniel Gustafsson @ 2023-08-24 08:38 UTC (permalink / raw) To: Sergey Shinderuk <[email protected]>; +Cc: pgsql-hackers > On 24 Aug 2023, at 10:11, Sergey Shinderuk <[email protected]> wrote: > > On 23.08.2023 16:23, Daniel Gustafsson wrote: >>> On 1 Aug 2023, at 16:44, Sergey Shinderuk <[email protected]> wrote: >>> A static analyzer reported a possible pfree(NULL) in be_tls_open_server(). >> This has the smell of a theoretical problem, I can't really imagine a >> certificate where which would produce this. Have you been able to trigger it? > > I triggered a crash by generating a certificate without a CN and forcing malloc to return NULL when called from X509_NAME_print_ex or BIO_get_mem_ptr with gdb. Can you extend the patch with that certificate and a test using it? The certificates are generated from config files kept in the repo in src/test/ssl in order to be reproducible. -- Daniel Gustafsson ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-24 09:13 Sergey Shinderuk <[email protected]> parent: Daniel Gustafsson <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Sergey Shinderuk @ 2023-08-24 09:13 UTC (permalink / raw) To: Daniel Gustafsson <[email protected]>; +Cc: pgsql-hackers On 24.08.2023 11:38, Daniel Gustafsson wrote: >> On 24 Aug 2023, at 10:11, Sergey Shinderuk <[email protected]> wrote: >> I triggered a crash by generating a certificate without a CN and forcing malloc to return NULL when called from X509_NAME_print_ex or BIO_get_mem_ptr with gdb. > > Can you extend the patch with that certificate and a test using it? The > certificates are generated from config files kept in the repo in src/test/ssl > in order to be reproducible. But I need to force malloc to fail at the right time during the test. I don't think I can make a stable and portable test from this. -- Sergey Shinderuk https://postgrespro.com/ ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-25 01:25 Michael Paquier <[email protected]> parent: Sergey Shinderuk <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Michael Paquier @ 2023-08-25 01:25 UTC (permalink / raw) To: Sergey Shinderuk <[email protected]>; +Cc: Daniel Gustafsson <[email protected]>; pgsql-hackers On Thu, Aug 24, 2023 at 12:13:18PM +0300, Sergey Shinderuk wrote: > But I need to force malloc to fail at the right time during the test. I > don't think I can make a stable and portable test from this. LD_PRELOAD is the only thing I can think about, but that's very fancy. Even with that, having a certificate with a NULL peer_cn could prove to be useful in the SSL suite to stress more patterns around it? -- Michael Attachments: [application/pgp-signature] signature.asc (833B, ../../ZOgDDH%[email protected]/2-signature.asc) download ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-28 15:12 Jacob Champion <[email protected]> parent: Michael Paquier <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Jacob Champion @ 2023-08-28 15:12 UTC (permalink / raw) To: Michael Paquier <[email protected]>; +Cc: Sergey Shinderuk <[email protected]>; Daniel Gustafsson <[email protected]>; pgsql-hackers On Thu, Aug 24, 2023 at 6:25 PM Michael Paquier <[email protected]> wrote: > LD_PRELOAD is the only thing I can think about, but that's very fancy. > Even with that, having a certificate with a NULL peer_cn could prove > to be useful in the SSL suite to stress more patterns around it? +1. Last we tried it, OpenSSL didn't want to create a certificate with an embedded null, but maybe things have changed? --Jacob ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-08-28 17:39 Sergey Shinderuk <[email protected]> parent: Jacob Champion <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Sergey Shinderuk @ 2023-08-28 17:39 UTC (permalink / raw) To: Jacob Champion <[email protected]>; Michael Paquier <[email protected]>; +Cc: Daniel Gustafsson <[email protected]>; pgsql-hackers On 28.08.2023 18:12, Jacob Champion wrote: > On Thu, Aug 24, 2023 at 6:25 PM Michael Paquier <[email protected]> wrote: >> LD_PRELOAD is the only thing I can think about, but that's very fancy. >> Even with that, having a certificate with a NULL peer_cn could prove >> to be useful in the SSL suite to stress more patterns around it? > > +1. Last we tried it, OpenSSL didn't want to create a certificate with > an embedded null, but maybe things have changed? > To embed a null byte into the Subject, I first generated a regular certificate request in the DER (binary) format, then manually inserted null into the file and recomputed the checksum. Like this: https://security.stackexchange.com/a/58845 I'll try to add a client certificate lacking a CN to the SSL test suite. -- Sergey Shinderuk https://postgrespro.com/ ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-09-18 12:35 Daniel Gustafsson <[email protected]> parent: Sergey Shinderuk <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Daniel Gustafsson @ 2023-09-18 12:35 UTC (permalink / raw) To: Sergey Shinderuk <[email protected]>; +Cc: Jacob Champion <[email protected]>; Michael Paquier <[email protected]>; pgsql-hackers > On 28 Aug 2023, at 19:39, Sergey Shinderuk <[email protected]> wrote: > I'll try to add a client certificate lacking a CN to the SSL test suite. Certificates can be regenerated with the buildsystem, which ideally would apply to this cert as well, but if that's not feasible we can perhaps accept a static one with build information detailed in the README. Having such a cert could for sure be interesting for testing. Awaiting resolution on this, I propose we go ahead with the original patch from this thread. Any objections to that? -- Daniel Gustafsson ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-09-19 00:54 Michael Paquier <[email protected]> parent: Daniel Gustafsson <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Michael Paquier @ 2023-09-19 00:54 UTC (permalink / raw) To: Daniel Gustafsson <[email protected]>; +Cc: Sergey Shinderuk <[email protected]>; Jacob Champion <[email protected]>; pgsql-hackers On Mon, Sep 18, 2023 at 02:35:28PM +0200, Daniel Gustafsson wrote: > Certificates can be regenerated with the buildsystem, which ideally would apply > to this cert as well, but if that's not feasible we can perhaps accept a static > one with build information detailed in the README. Having such a cert could > for sure be interesting for testing. WFM, but I'd prefer something that would be generated with the makefile rules. These are so handy when it comes to regenerate all these certs.. > Awaiting resolution on this, I propose we go ahead with the original patch from > this thread. Any objections to that? I was wondering for a few seconds if you talked about the one posted on [1], which would break the case where X509_NAME_get_text_by_NID() fails if there's a valid bio, but you mean the one at the top of the thread in [2], of course :) One doubt that I have is if we shouldn't let X509_NAME_print_ex() be as it is now, and not force a failure on the bio if this calls fails. [1]: https://www.postgresql.org/message-id/[email protected] [2]: https://www.postgresql.org/message-id/[email protected] -- Michael Attachments: [application/pgp-signature] signature.asc (833B, ../../[email protected]/2-signature.asc) download ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-09-19 08:06 Sergey Shinderuk <[email protected]> parent: Michael Paquier <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Sergey Shinderuk @ 2023-09-19 08:06 UTC (permalink / raw) To: Michael Paquier <[email protected]>; Daniel Gustafsson <[email protected]>; +Cc: Jacob Champion <[email protected]>; pgsql-hackers On 19.09.2023 03:54, Michael Paquier wrote: > One doubt that I have is if we shouldn't let X509_NAME_print_ex() be > as it is now, and not force a failure on the bio if this calls fails. If malloc fails inside X509_NAME_print_ex, then we will be left with empty port->peer_dn. Here is a gdb session showing this: (gdb) b X509_NAME_print_ex Breakpoint 1 at 0x7f539f6c0cf0 (gdb) c Continuing. Breakpoint 1, 0x00007f539f6c0cf0 in X509_NAME_print_ex () from /lib/x86_64-linux-gnu/libcrypto.so.3 (gdb) bt #0 0x00007f539f6c0cf0 in X509_NAME_print_ex () from /lib/x86_64-linux-gnu/libcrypto.so.3 #1 0x000056026d2fbe8d in be_tls_open_server (port=port@entry=0x56026ed5d730) at be-secure-openssl.c:635 #2 0x000056026d2eefa5 in secure_open_server (port=port@entry=0x56026ed5d730) at be-secure.c:118 #3 0x000056026d3dc412 in ProcessStartupPacket (port=port@entry=0x56026ed5d730, ssl_done=ssl_done@entry=false, gss_done=gss_done@entry=false) at postmaster.c:2065 #4 0x000056026d3dcd8e in BackendInitialize (port=port@entry=0x56026ed5d730) at postmaster.c:4377 #5 0x000056026d3def6a in BackendStartup (port=port@entry=0x56026ed5d730) at postmaster.c:4155 #6 0x000056026d3df115 in ServerLoop () at postmaster.c:1781 #7 0x000056026d3e0645 in PostmasterMain (argc=argc@entry=20, argv=argv@entry=0x56026ec5a0d0) at postmaster.c:1465 #8 0x000056026d2fcd7c in main (argc=20, argv=0x56026ec5a0d0) at main.c:198 (gdb) b malloc Breakpoint 2 at 0x7f539eca5120: file ./malloc/malloc.c, line 3287. (gdb) c Continuing. Breakpoint 2, __GI___libc_malloc (bytes=4) at ./malloc/malloc.c:3287 3287 ./malloc/malloc.c: No such file or directory. (gdb) bt #0 __GI___libc_malloc (bytes=4) at ./malloc/malloc.c:3287 #1 0x00007f539f6f6e09 in BUF_MEM_grow_clean () from /lib/x86_64-linux-gnu/libcrypto.so.3 #2 0x00007f539f6e4fb8 in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3 #3 0x00007f539f6d22fb in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3 #4 0x00007f539f6d5c06 in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3 #5 0x00007f539f6d5d37 in BIO_write () from /lib/x86_64-linux-gnu/libcrypto.so.3 #6 0x00007f539f6bdb41 in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3 #7 0x00007f539f6c0b7d in ?? () from /lib/x86_64-linux-gnu/libcrypto.so.3 #8 0x000056026d2fbe8d in be_tls_open_server (port=port@entry=0x56026ed5d730) at be-secure-openssl.c:635 #9 0x000056026d2eefa5 in secure_open_server (port=port@entry=0x56026ed5d730) at be-secure.c:118 #10 0x000056026d3dc412 in ProcessStartupPacket (port=port@entry=0x56026ed5d730, ssl_done=ssl_done@entry=false, gss_done=gss_done@entry=false) at postmaster.c:2065 #11 0x000056026d3dcd8e in BackendInitialize (port=port@entry=0x56026ed5d730) at postmaster.c:4377 #12 0x000056026d3def6a in BackendStartup (port=port@entry=0x56026ed5d730) at postmaster.c:4155 #13 0x000056026d3df115 in ServerLoop () at postmaster.c:1781 #14 0x000056026d3e0645 in PostmasterMain (argc=argc@entry=20, argv=argv@entry=0x56026ec5a0d0) at postmaster.c:1465 #15 0x000056026d2fcd7c in main (argc=20, argv=0x56026ec5a0d0) at main.c:198 (gdb) return 0 Make __GI___libc_malloc return now? (y or n) y #0 0x00007f539f6f6e09 in BUF_MEM_grow_clean () from /lib/x86_64-linux-gnu/libcrypto.so.3 (gdb) delete Delete all breakpoints? (y or n) y (gdb) c Continuing. And in the server log we see: DEBUG: SSL connection from DN:"" CN:"ssltestuser" While in the normal case we get: DEBUG: SSL connection from DN:"CN=ssltestuser" CN:"ssltestuser"\ Probably we shouldn't ignore the error from X509_NAME_print_ex? -- Sergey Shinderuk https://postgrespro.com/ ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-09-20 08:42 Daniel Gustafsson <[email protected]> parent: Sergey Shinderuk <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Daniel Gustafsson @ 2023-09-20 08:42 UTC (permalink / raw) To: Sergey Shinderuk <[email protected]>; +Cc: Michael Paquier <[email protected]>; Jacob Champion <[email protected]>; pgsql-hackers > On 19 Sep 2023, at 10:06, Sergey Shinderuk <[email protected]> wrote: > > On 19.09.2023 03:54, Michael Paquier wrote: >> One doubt that I have is if we shouldn't let X509_NAME_print_ex() be >> as it is now, and not force a failure on the bio if this calls fails. > > If malloc fails inside X509_NAME_print_ex, then we will be left with empty port->peer_dn. Looking at the OpenSSL code, there a other (albeit esoteric) errors that return -1 as well. I agree that we should handle this error. X509_NAME_print_ex is not documented to return -1 in OpenSSL 1.0.2 but reading the code it's clear that it does, so checking for -1 is safe for all supported OpenSSL versions (supported by us that is). Attached is a v2 on top of HEAD with commit message etc, which I propose to backpatch to v15 where it was introduced. -- Daniel Gustafsson Attachments: [application/octet-stream] v2-0001-Avoid-potential-pfree-on-NULL-on-OpenSSL-errors.patch (2.1K, ../../[email protected]/2-v2-0001-Avoid-potential-pfree-on-NULL-on-OpenSSL-errors.patch) download | inline diff: From 168c8f4b62bdbbf905af9d0c7f3f7af68cf8c6e9 Mon Sep 17 00:00:00 2001 From: Daniel Gustafsson <[email protected]> Date: Wed, 20 Sep 2023 10:31:10 +0200 Subject: [PATCH v2] Avoid potential pfree on NULL on OpenSSL errors Guard against the pointer being NULL before pfreeing upon en error returned from OpenSSL. Also handle errors from X509_NAME_print_ex which also can return -1 on memory allocation errors. Author: Sergey Shinderuk <[email protected]> Discussion: https://postgr.es/m/[email protected] --- src/backend/libpq/be-secure-openssl.c | 25 +++++++++++++++++-------- 1 file changed, 17 insertions(+), 8 deletions(-) diff --git a/src/backend/libpq/be-secure-openssl.c b/src/backend/libpq/be-secure-openssl.c index 658b09988d..31b6a6eacd 100644 --- a/src/backend/libpq/be-secure-openssl.c +++ b/src/backend/libpq/be-secure-openssl.c @@ -620,8 +620,11 @@ aloop: bio = BIO_new(BIO_s_mem()); if (!bio) { - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } @@ -632,12 +635,15 @@ aloop: * which make regular expression matching a bit easier. Also note that * it prints the Subject fields in reverse order. */ - X509_NAME_print_ex(bio, x509name, 0, XN_FLAG_RFC2253); - if (BIO_get_mem_ptr(bio, &bio_buf) <= 0) + if (X509_NAME_print_ex(bio, x509name, 0, XN_FLAG_RFC2253) == -1 || + BIO_get_mem_ptr(bio, &bio_buf) <= 0) { BIO_free(bio); - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } peer_dn = MemoryContextAlloc(TopMemoryContext, bio_buf->length + 1); @@ -651,8 +657,11 @@ aloop: (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("SSL certificate's distinguished name contains embedded null"))); pfree(peer_dn); - pfree(port->peer_cn); - port->peer_cn = NULL; + if (port->peer_cn != NULL) + { + pfree(port->peer_cn); + port->peer_cn = NULL; + } return -1; } -- 2.32.1 (Apple Git-133) ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-09-20 08:55 Sergey Shinderuk <[email protected]> parent: Daniel Gustafsson <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Sergey Shinderuk @ 2023-09-20 08:55 UTC (permalink / raw) To: Daniel Gustafsson <[email protected]>; +Cc: Michael Paquier <[email protected]>; Jacob Champion <[email protected]>; pgsql-hackers On 20.09.2023 11:42, Daniel Gustafsson wrote: > Attached is a v2 on top of HEAD with commit message etc, which I propose to > backpatch to v15 where it was introduced. > There is a typo: upon en error. Otherwise, looks good to me. Thank you. -- Sergey Shinderuk https://postgrespro.com/ ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-09-20 08:58 Daniel Gustafsson <[email protected]> parent: Sergey Shinderuk <[email protected]> 0 siblings, 1 reply; 37+ messages in thread From: Daniel Gustafsson @ 2023-09-20 08:58 UTC (permalink / raw) To: Sergey Shinderuk <[email protected]>; +Cc: Michael Paquier <[email protected]>; Jacob Champion <[email protected]>; pgsql-hackers > On 20 Sep 2023, at 10:55, Sergey Shinderuk <[email protected]> wrote: > > On 20.09.2023 11:42, Daniel Gustafsson wrote: >> Attached is a v2 on top of HEAD with commit message etc, which I propose to >> backpatch to v15 where it was introduced. > There is a typo: upon en error. Otherwise, looks good to me. Thank you. Thanks, will fix before pushing. -- Daniel Gustafsson ^ permalink raw reply [nested|flat] 37+ messages in thread
* Re: Fix error handling in be_tls_open_server() @ 2023-09-22 09:43 Daniel Gustafsson <[email protected]> parent: Daniel Gustafsson <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Daniel Gustafsson @ 2023-09-22 09:43 UTC (permalink / raw) To: Sergey Shinderuk <[email protected]>; +Cc: Michael Paquier <[email protected]>; Jacob Champion <[email protected]>; pgsql-hackers > On 20 Sep 2023, at 10:58, Daniel Gustafsson <[email protected]> wrote: >> On 20 Sep 2023, at 10:55, Sergey Shinderuk <[email protected]> wrote: >> There is a typo: upon en error. Otherwise, looks good to me. Thank you. > > Thanks, will fix before pushing. I've applied this down to v15 now and the buildfarm have green builds on all three branches, thanks for the submission! -- Daniel Gustafsson ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v7 3/3] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 5d9db167e59..8b6a7652fcf 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1795,7 +1795,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index b55221d44cd..da75dfa6ab8 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 3aac459e483..540923452fb 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16848,6 +16848,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16886,6 +16887,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16898,6 +16900,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 7debb14bb5d..15b4663eb77 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -165,14 +224,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -215,14 +314,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -660,6 +782,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -675,6 +839,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -754,11 +942,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -778,10 +974,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -804,6 +1073,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -839,6 +1110,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -853,9 +1147,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -910,7 +1221,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1162,3 +1483,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index bc9864bd8d9..f8cf3755ce2 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 915d0bc9084..7a7f8023eb3 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3780,6 +3781,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3951,6 +3953,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 5d0fe79f7e3..332dffde400 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -665,7 +665,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -677,6 +677,9 @@ extern void pgstat_report_vacuum(RelFileLocator locator, PgStat_Counter livetupl extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 432509277c9..bd8fcb16dcf 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2248,6 +2248,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --KMYjwlsACqdYx2P8-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v11 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 43de42ce39e..88ef00609db 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1793,7 +1793,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 60a4617a585..3a446f32921 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index df1ba112b35..6244d04ab12 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16896,6 +16896,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16934,6 +16935,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16946,6 +16948,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 6b634c9fff1..692d084acb2 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3775,6 +3776,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3946,6 +3948,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index e9e4d32c3b8..a9a4fa9f8f2 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -669,7 +669,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -681,6 +681,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 241945734ec..05b1380c08c 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2279,6 +2279,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --zLA8DXafBEP889wv-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v8 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 8dea58ad96b..b71925a22c3 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1795,7 +1795,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 2120c85ccb4..6155b12afab 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 6b1a00ed477..9de70f321ed 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16884,6 +16884,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16922,6 +16923,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16934,6 +16936,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index ee6f2eb2bdb..83103189fb5 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -165,14 +224,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -215,14 +314,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -660,6 +782,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -675,6 +839,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -754,11 +942,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -778,10 +974,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -804,6 +1073,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -839,6 +1110,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -853,9 +1147,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -910,7 +1221,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1162,3 +1483,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index bc9864bd8d9..f8cf3755ce2 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 2d0cb7bcfd4..c98e5c51d63 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3780,6 +3781,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3951,6 +3953,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index c9e451094f3..33d530740d0 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -670,7 +670,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -682,6 +682,9 @@ extern void pgstat_report_vacuum(RelFileLocator locator, PgStat_Counter livetupl extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 3451538565e..0ca3eae8026 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2268,6 +2268,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --BHD/2LBqtxMXWADJ-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v9 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 8dea58ad96b..b71925a22c3 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1795,7 +1795,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 2120c85ccb4..6155b12afab 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 6b1a00ed477..9de70f321ed 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16884,6 +16884,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16922,6 +16923,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16934,6 +16936,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 48bf93cae6e..b6ba8fdf440 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index bc9864bd8d9..f8cf3755ce2 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 2d0cb7bcfd4..c98e5c51d63 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3780,6 +3781,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3951,6 +3953,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 3102f86aa24..8750c025bbe 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -670,7 +670,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -682,6 +682,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 3451538565e..0ca3eae8026 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2268,6 +2268,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --tV2Nti+lXhoeT6I5-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v10 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 8dea58ad96b..b71925a22c3 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1795,7 +1795,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 2120c85ccb4..6155b12afab 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 6b1a00ed477..9de70f321ed 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16884,6 +16884,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16922,6 +16923,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16934,6 +16936,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index a267e93f8be..123fb50d98f 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index bc9864bd8d9..f8cf3755ce2 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 2d0cb7bcfd4..c98e5c51d63 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3780,6 +3781,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3951,6 +3953,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 3102f86aa24..8750c025bbe 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -670,7 +670,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -682,6 +682,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 04845d5e680..3184e9c464b 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2269,6 +2269,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --xJsrX25x3tm7Tn81-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v12 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 43de42ce39e..88ef00609db 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1793,7 +1793,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 60a4617a585..3a446f32921 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 85242dcc245..8d10d9d1e0f 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16883,6 +16883,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16921,6 +16922,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16933,6 +16935,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index a1c88c6b1b6..dc5948f446c 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3775,6 +3776,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3946,6 +3948,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 689c624f373..d0d5473f8a5 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -673,7 +673,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -685,6 +685,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 3250564d4ff..fbcbe5da26a 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2281,6 +2281,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --sdA5zuaL94ts0rne-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v13 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 8b3c60d91f9..7c7d0fb8e7a 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1796,7 +1796,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 09066db0956..c491daceb0f 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1206,6 +1206,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 67e42e5df29..1715af2c0d8 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -17099,6 +17099,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -17137,6 +17138,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -17149,6 +17151,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 3a4f19e8d58..60491962668 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -86,6 +86,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3771,6 +3772,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3942,6 +3944,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 689c624f373..d0d5473f8a5 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -673,7 +673,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -685,6 +685,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 52f8603a7be..8b84e3a0bcf 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2298,6 +2298,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --USlF2Vr25us5UpvY-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v14 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index d8219b18c48..079e504b96f 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1796,7 +1796,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 09066db0956..c491daceb0f 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1206,6 +1206,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index c69c12dc014..13630846116 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -17099,6 +17099,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -17137,6 +17138,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -17149,6 +17151,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index e19f0d3e51c..bf49830dba2 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -87,6 +87,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3772,6 +3773,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3943,6 +3945,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index ef740aff26e..4d3130ae99e 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -701,7 +701,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -713,6 +713,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index dbbec84b222..bae04e95989 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2310,6 +2310,7 @@ PgStat_LockEntry PgStat_PendingDroppedStatsItem PgStat_PendingIO PgStat_PendingLock +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --96iU9DKeu+W+YlSa-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v15 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 1ccfa687f05..e292199ca7c 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1796,7 +1796,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 09066db0956..c491daceb0f 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1206,6 +1206,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 8b4ebc6f226..49de04d9697 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -17102,6 +17102,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -17140,6 +17141,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -17152,6 +17154,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index e19f0d3e51c..bf49830dba2 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -87,6 +87,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3772,6 +3773,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3943,6 +3945,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index ef740aff26e..4d3130ae99e 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -701,7 +701,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -713,6 +713,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index c9da1f91cb9..53c94382a8d 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2314,6 +2314,7 @@ PgStat_LockEntry PgStat_PendingDroppedStatsItem PgStat_PendingIO PgStat_PendingLock +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --Og8T0Lf2JVsuKOzn-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v14 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index d8219b18c48..079e504b96f 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1796,7 +1796,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 09066db0956..c491daceb0f 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1206,6 +1206,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index c69c12dc014..13630846116 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -17099,6 +17099,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -17137,6 +17138,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -17149,6 +17151,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index e19f0d3e51c..bf49830dba2 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -87,6 +87,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3772,6 +3773,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3943,6 +3945,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index ef740aff26e..4d3130ae99e 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -701,7 +701,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -713,6 +713,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index dbbec84b222..bae04e95989 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2310,6 +2310,7 @@ PgStat_LockEntry PgStat_PendingDroppedStatsItem PgStat_PendingIO PgStat_PendingLock +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --96iU9DKeu+W+YlSa-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v15 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 1ccfa687f05..e292199ca7c 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1796,7 +1796,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 09066db0956..c491daceb0f 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1206,6 +1206,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 8b4ebc6f226..49de04d9697 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -17102,6 +17102,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -17140,6 +17141,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -17152,6 +17154,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index e19f0d3e51c..bf49830dba2 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -87,6 +87,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3772,6 +3773,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3943,6 +3945,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index ef740aff26e..4d3130ae99e 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -701,7 +701,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -713,6 +713,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index c9da1f91cb9..53c94382a8d 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2314,6 +2314,7 @@ PgStat_LockEntry PgStat_PendingDroppedStatsItem PgStat_PendingIO PgStat_PendingLock +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --Og8T0Lf2JVsuKOzn-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v7 3/3] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 5d9db167e59..8b6a7652fcf 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1795,7 +1795,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index b55221d44cd..da75dfa6ab8 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 3aac459e483..540923452fb 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16848,6 +16848,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16886,6 +16887,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16898,6 +16900,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 7debb14bb5d..15b4663eb77 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -165,14 +224,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -215,14 +314,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -660,6 +782,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -675,6 +839,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -754,11 +942,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -778,10 +974,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -804,6 +1073,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -839,6 +1110,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -853,9 +1147,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -910,7 +1221,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1162,3 +1483,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index bc9864bd8d9..f8cf3755ce2 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 915d0bc9084..7a7f8023eb3 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3780,6 +3781,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3951,6 +3953,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 5d0fe79f7e3..332dffde400 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -665,7 +665,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -677,6 +677,9 @@ extern void pgstat_report_vacuum(RelFileLocator locator, PgStat_Counter livetupl extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 432509277c9..bd8fcb16dcf 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2248,6 +2248,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --KMYjwlsACqdYx2P8-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v8 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 8dea58ad96b..b71925a22c3 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1795,7 +1795,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 2120c85ccb4..6155b12afab 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 6b1a00ed477..9de70f321ed 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16884,6 +16884,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16922,6 +16923,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16934,6 +16936,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index ee6f2eb2bdb..83103189fb5 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -165,14 +224,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -215,14 +314,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -660,6 +782,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -675,6 +839,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -754,11 +942,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -778,10 +974,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -804,6 +1073,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -839,6 +1110,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -853,9 +1147,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -910,7 +1221,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1162,3 +1483,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index bc9864bd8d9..f8cf3755ce2 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 2d0cb7bcfd4..c98e5c51d63 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3780,6 +3781,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3951,6 +3953,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index c9e451094f3..33d530740d0 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -670,7 +670,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -682,6 +682,9 @@ extern void pgstat_report_vacuum(RelFileLocator locator, PgStat_Counter livetupl extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 3451538565e..0ca3eae8026 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2268,6 +2268,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --BHD/2LBqtxMXWADJ-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v9 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 8dea58ad96b..b71925a22c3 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1795,7 +1795,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 2120c85ccb4..6155b12afab 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 6b1a00ed477..9de70f321ed 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16884,6 +16884,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16922,6 +16923,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16934,6 +16936,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 48bf93cae6e..b6ba8fdf440 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index bc9864bd8d9..f8cf3755ce2 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 2d0cb7bcfd4..c98e5c51d63 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3780,6 +3781,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3951,6 +3953,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 3102f86aa24..8750c025bbe 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -670,7 +670,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -682,6 +682,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 3451538565e..0ca3eae8026 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2268,6 +2268,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --tV2Nti+lXhoeT6I5-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v10 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 8dea58ad96b..b71925a22c3 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1795,7 +1795,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 2120c85ccb4..6155b12afab 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 6b1a00ed477..9de70f321ed 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16884,6 +16884,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16922,6 +16923,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16934,6 +16936,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index a267e93f8be..123fb50d98f 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index bc9864bd8d9..f8cf3755ce2 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 2d0cb7bcfd4..c98e5c51d63 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3780,6 +3781,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3951,6 +3953,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 3102f86aa24..8750c025bbe 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -670,7 +670,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -682,6 +682,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 04845d5e680..3184e9c464b 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2269,6 +2269,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --xJsrX25x3tm7Tn81-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v11 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 43de42ce39e..88ef00609db 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1793,7 +1793,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 60a4617a585..3a446f32921 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index df1ba112b35..6244d04ab12 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16896,6 +16896,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16934,6 +16935,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16946,6 +16948,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 6b634c9fff1..692d084acb2 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3775,6 +3776,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3946,6 +3948,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index e9e4d32c3b8..a9a4fa9f8f2 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -669,7 +669,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -681,6 +681,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 241945734ec..05b1380c08c 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2279,6 +2279,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --zLA8DXafBEP889wv-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v12 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 43de42ce39e..88ef00609db 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1793,7 +1793,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 60a4617a585..3a446f32921 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1196,6 +1196,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 85242dcc245..8d10d9d1e0f 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -16883,6 +16883,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -16921,6 +16922,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -16933,6 +16935,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index a1c88c6b1b6..dc5948f446c 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -85,6 +85,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3775,6 +3776,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3946,6 +3948,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 689c624f373..d0d5473f8a5 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -673,7 +673,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -685,6 +685,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 3250564d4ff..fbcbe5da26a 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2281,6 +2281,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --sdA5zuaL94ts0rne-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v13 2/2] handle relation statistics correctly during rewrites @ 2025-11-04 13:52 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2025-11-04 13:52 UTC (permalink / raw) Now that PGSTAT_KIND_RELATION is keyed by refilenode, we need to handle rewrites. To do so, this patch: - Adds PgStat_PendingRewrite, a new struct to track rewrite operations within a transaction, storing the old locator, new locator, and original locator (for rewrite chains). This allows stats to be copied from the original location to the final location at commit time. - Adds a new function, pgstat_mark_rewrite(), called when a table rewrite begins. It records the rewrite operation in a local list and detects rewrite chains by checking if the old_locator matches any existing new_locator, preserving the chain's original_locator. - Modifies pgstat_copy_relation_stats(), to accept RelFileLocators instead of Relations, with a new increment parameter to accumulate stats (needed for rewrite chains with DML between rewrites). - Ensures that AtEOXact_PgStat_Relations(), AtPrepare_PgStat_Relations(), pgstat_twophase_postcommit()/postabort() pgstat_drop_relation() handle the PgStat_PendingRewrite list correctly. Note that due to the new flush call in pgstat_twophase_postcommit() we can not call GetCurrentTransactionStopTimestamp() in pgstat_relation_flush_cb(). So, adding a check to handle this special case and call GetCurrentTimestamp() instead. Note that we'd call GetCurrentTimestamp() only if there is a rewrite, so that the GetCurrentTimestamp() extra cost should be negligible. Another solution could be to trigger the flush from FinishPreparedTransaction() but that's not worth the extra complexity. The new pending_rewrites list is traversed in multiple places. The overhead should be negligible in comparison to a rewrite and the list should not contain a lot of rewrites in practice. The pending_rewrites list is traversed in multiple places. In typical usage, the list will contain only a few entries so the traversal cost is negligible ( furthermore in comparison to a rewrite). --- src/backend/catalog/index.c | 2 +- src/backend/commands/cluster.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 391 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 424 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.9% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 8b3c60d91f9..7c7d0fb8e7a 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1796,7 +1796,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 09066db0956..c491daceb0f 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -1206,6 +1206,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 67e42e5df29..1715af2c0d8 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -17099,6 +17099,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -17137,6 +17138,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -17149,6 +17151,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 89bf0cbed56..4d929972fab 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,70 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src)); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src)); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +194,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +223,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +313,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -666,6 +788,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -681,6 +845,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -760,11 +948,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -784,10 +980,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -810,6 +1079,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -845,6 +1116,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -859,9 +1153,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -916,7 +1227,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1167,3 +1488,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 3a4f19e8d58..60491962668 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -86,6 +86,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3771,6 +3772,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3942,6 +3944,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 689c624f373..d0d5473f8a5 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -673,7 +673,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -685,6 +685,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 52f8603a7be..8b84e3a0bcf 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2298,6 +2298,7 @@ PgStat_KindInfo PgStat_LocalState PgStat_PendingDroppedStatsItem PgStat_PendingIO +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --USlF2Vr25us5UpvY-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v16 2/2] handle relation statistics correctly during rewrites @ 2026-05-18 15:20 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2026-05-18 15:20 UTC (permalink / raw) This patch ensures that relation statistics are preserved correctly when a relation is rewritten (e.g., during CLUSTER, VACUUM FULL, ALTER TABLE operations that require a table rewrite, or TRUNCATE). Since patch 0001 keys PGSTAT_KIND_RELATION by the physical RelFileLocator rather than the relation OID, a rewrite (which assigns a new relfilenode) would lose the accumulated statistics if not handled specially. The approach is: 1. When a rewrite occurs (RelationSetNewRelfilenumber or swap_relation_files), record the old and new locators via pgstat_mark_rewrite(). 2. During pgstat_assoc_relation(), if the relation's new locator matches a pending rewrite, continue using the stats entry associated with the old locator. This ensures stats accumulated after the rewrite are still tracked in the original entry. 3. At transaction commit (AtEOXact_PgStat_Relations), process pending rewrites: flush any pending stats for the old locator, copy them to the new locator's stats entry, and drop the old entry. 4. At subtransaction abort, remove rewrites from the aborted nesting level. 5. For two-phase transactions, the rewrite information is recorded in the TwoPhasePgStatRecord and replayed at commit/abort time. This also handles chained rewrites (multiple rewrites of the same relation in a single transaction) by tracking the original locator through the chain. --- src/backend/catalog/index.c | 2 +- src/backend/commands/repack.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 392 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 425 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.8% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 9407c357f27..335ff1e50f6 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1813,7 +1813,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 351a3cc32e8..eda4135bcd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -1633,6 +1633,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 92b0f38c353..b553b9a4970 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -17129,6 +17129,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -17167,6 +17168,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -17179,6 +17181,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 7fcda7f7518..18d1b9fc59d 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,71 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src), NULL); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src), + NULL); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +195,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +224,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +314,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -668,6 +791,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -683,6 +848,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -762,11 +951,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -786,10 +983,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -812,6 +1082,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -847,6 +1119,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -861,9 +1156,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -918,7 +1230,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1169,3 +1491,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 0572ab424e7..d84058db102 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -87,6 +87,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3782,6 +3783,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3953,6 +3955,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 941a39bd16e..5344ece9ecf 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -697,7 +697,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -709,6 +709,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 8cf40c87043..1544f54e918 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2349,6 +2349,7 @@ PgStat_LockEntry PgStat_PendingDroppedStatsItem PgStat_PendingIO PgStat_PendingLock +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --Uc+t0mng6H5VV2zk-- ^ permalink raw reply [nested|flat] 37+ messages in thread
* [PATCH v16 2/2] handle relation statistics correctly during rewrites @ 2026-05-18 15:20 Bertrand Drouvot <[email protected]> 0 siblings, 0 replies; 37+ messages in thread From: Bertrand Drouvot @ 2026-05-18 15:20 UTC (permalink / raw) This patch ensures that relation statistics are preserved correctly when a relation is rewritten (e.g., during CLUSTER, VACUUM FULL, ALTER TABLE operations that require a table rewrite, or TRUNCATE). Since patch 0001 keys PGSTAT_KIND_RELATION by the physical RelFileLocator rather than the relation OID, a rewrite (which assigns a new relfilenode) would lose the accumulated statistics if not handled specially. The approach is: 1. When a rewrite occurs (RelationSetNewRelfilenumber or swap_relation_files), record the old and new locators via pgstat_mark_rewrite(). 2. During pgstat_assoc_relation(), if the relation's new locator matches a pending rewrite, continue using the stats entry associated with the old locator. This ensures stats accumulated after the rewrite are still tracked in the original entry. 3. At transaction commit (AtEOXact_PgStat_Relations), process pending rewrites: flush any pending stats for the old locator, copy them to the new locator's stats entry, and drop the old entry. 4. At subtransaction abort, remove rewrites from the aborted nesting level. 5. For two-phase transactions, the rewrite information is recorded in the TwoPhasePgStatRecord and replayed at commit/abort time. This also handles chained rewrites (multiple rewrites of the same relation in a single transaction) by tracking the original locator through the chain. --- src/backend/catalog/index.c | 2 +- src/backend/commands/repack.c | 5 + src/backend/commands/tablecmds.c | 6 + src/backend/utils/activity/pgstat_relation.c | 392 ++++++++++++++++++- src/backend/utils/activity/pgstat_xact.c | 25 +- src/backend/utils/cache/relcache.c | 6 + src/include/pgstat.h | 5 +- src/tools/pgindent/typedefs.list | 1 + 8 files changed, 425 insertions(+), 17 deletions(-) 92.8% src/backend/utils/activity/ 4.8% src/backend/ diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 9407c357f27..335ff1e50f6 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -1813,7 +1813,7 @@ index_concurrently_swap(Oid newIndexId, Oid oldIndexId, const char *oldName) changeDependenciesOn(RelationRelationId, oldIndexId, newIndexId); /* copy over statistics from old to new index */ - pgstat_copy_relation_stats(newClassRel, oldClassRel); + pgstat_copy_relation_stats(newClassRel->rd_locator, oldClassRel->rd_locator, false); /* Copy data of pg_statistic from the old index to the new one */ CopyStatistics(oldIndexId, newIndexId); diff --git a/src/backend/commands/repack.c b/src/backend/commands/repack.c index 351a3cc32e8..eda4135bcd1 100644 --- a/src/backend/commands/repack.c +++ b/src/backend/commands/repack.c @@ -1633,6 +1633,11 @@ swap_relation_files(Oid r1, Oid r2, bool target_is_pg_class, rel1 = relation_open(r1, NoLock); rel2 = relation_open(r2, NoLock); + + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel1->rd_rel->relkind)) + pgstat_mark_rewrite(rel1->rd_locator, rel2->rd_locator); + rel2->rd_createSubid = rel1->rd_createSubid; rel2->rd_newRelfilelocatorSubid = rel1->rd_newRelfilelocatorSubid; rel2->rd_firstRelfilelocatorSubid = rel1->rd_firstRelfilelocatorSubid; diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index 92b0f38c353..b553b9a4970 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -17129,6 +17129,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) Oid reltoastrelid; RelFileNumber newrelfilenumber; RelFileLocator newrlocator; + RelFileLocator oldrlocator; List *reltoastidxids = NIL; ListCell *lc; @@ -17167,6 +17168,7 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) newrlocator = rel->rd_locator; newrlocator.relNumber = newrelfilenumber; newrlocator.spcOid = newTableSpace; + oldrlocator = rel->rd_locator; /* hand off to AM to actually create new rel storage and copy the data */ if (rel->rd_rel->relkind == RELKIND_INDEX) @@ -17179,6 +17181,10 @@ ATExecSetTableSpace(Oid tableOid, Oid newTableSpace, LOCKMODE lockmode) table_relation_copy_data(rel, &newrlocator); } + /* mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Update the pg_class row. * diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 7fcda7f7518..18d1b9fc59d 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -30,6 +30,19 @@ #include "utils/syscache.h" #include "utils/timestamp.h" +/* Pending rewrite operations for stats copying */ +typedef struct PgStat_PendingRewrite +{ + RelFileLocator old_locator; + RelFileLocator new_locator; + RelFileLocator original_locator; + int nest_level; /* Transaction nesting level where rewrite + * occurred */ + struct PgStat_PendingRewrite *next; +} PgStat_PendingRewrite; + +/* The pending rewrites list for current transaction */ +static PgStat_PendingRewrite *pending_rewrites = NULL; /* Record that's written to 2PC state file when pgstat state is persisted */ typedef struct TwoPhasePgStatRecord @@ -43,6 +56,8 @@ typedef struct TwoPhasePgStatRecord PgStat_Counter deleted_pre_truncdrop; RelFileLocator locator; /* table's rd_locator */ bool truncdropped; /* was the relation truncated/dropped? */ + RelFileLocator rewrite_old_locator; + int rewrite_nest_level; } TwoPhasePgStatRecord; @@ -54,27 +69,71 @@ static void restore_truncdrop_counters(PgStat_TableXactStatus *trans); /* - * Copy stats between relations. This is used for things like REINDEX + * Copy stats between RelFileLocator. This is used for things like REINDEX * CONCURRENTLY. */ void -pgstat_copy_relation_stats(Relation dst, Relation src) +pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment) { PgStat_StatTabEntry *srcstats; PgStatShared_Relation *dstshstats; PgStat_EntryRef *dst_ref; - srcstats = pgstat_fetch_stat_tabentry_ext(RelationGetRelid(src), NULL); + srcstats = (PgStat_StatTabEntry *) pgstat_fetch_entry(PGSTAT_KIND_RELATION, + src.dbOid, + RelFileLocatorToPgStatObjid(src), + NULL); if (!srcstats) return; dst_ref = pgstat_get_entry_ref_locked(PGSTAT_KIND_RELATION, - dst->rd_rel->relisshared ? InvalidOid : MyDatabaseId, - RelationGetRelid(dst), + dst.dbOid, + RelFileLocatorToPgStatObjid(dst), false); dstshstats = (PgStatShared_Relation *) dst_ref->shared_stats; - dstshstats->stats = *srcstats; + + if (!increment) + dstshstats->stats = *srcstats; + else + { + /* Increment those statistics */ +#define RELFSTAT_ACC(fld, stats_to_add) \ + (dstshstats->stats.fld += stats_to_add->fld) + RELFSTAT_ACC(numscans, srcstats); + RELFSTAT_ACC(tuples_returned, srcstats); + RELFSTAT_ACC(tuples_fetched, srcstats); + RELFSTAT_ACC(tuples_inserted, srcstats); + RELFSTAT_ACC(tuples_updated, srcstats); + RELFSTAT_ACC(tuples_deleted, srcstats); + RELFSTAT_ACC(tuples_hot_updated, srcstats); + RELFSTAT_ACC(tuples_newpage_updated, srcstats); + RELFSTAT_ACC(live_tuples, srcstats); + RELFSTAT_ACC(dead_tuples, srcstats); + RELFSTAT_ACC(mod_since_analyze, srcstats); + RELFSTAT_ACC(ins_since_vacuum, srcstats); + RELFSTAT_ACC(blocks_fetched, srcstats); + RELFSTAT_ACC(blocks_hit, srcstats); + RELFSTAT_ACC(vacuum_count, srcstats); + RELFSTAT_ACC(autovacuum_count, srcstats); + RELFSTAT_ACC(analyze_count, srcstats); + RELFSTAT_ACC(autoanalyze_count, srcstats); + RELFSTAT_ACC(total_vacuum_time, srcstats); + RELFSTAT_ACC(total_autovacuum_time, srcstats); + RELFSTAT_ACC(total_analyze_time, srcstats); + RELFSTAT_ACC(total_autoanalyze_time, srcstats); +#undef RELFSTAT_ACC + + /* Replace those statistics */ +#define RELFSTAT_REP(fld, stats_to_rep) \ + (dstshstats->stats.fld = stats_to_rep->fld) + RELFSTAT_REP(lastscan, srcstats); + RELFSTAT_REP(last_vacuum_time, srcstats); + RELFSTAT_REP(last_autovacuum_time, srcstats); + RELFSTAT_REP(last_analyze_time, srcstats); + RELFSTAT_REP(last_autoanalyze_time, srcstats); +#undef RELFSTAT_REP + } pgstat_unlock_entry(dst_ref); } @@ -136,6 +195,7 @@ void pgstat_assoc_relation(Relation rel) { RelFileLocator locator; + PgStat_TableStatus *pgstat_info; Assert(rel->pgstat_enabled); Assert(rel->pgstat_info == NULL); @@ -164,14 +224,54 @@ pgstat_assoc_relation(Relation rel) locator.relNumber = rel->rd_id; } + /* + * If this relation was rewritten during the current transaction we may be + * reopening it with its new RelFileLocator. In that case, continue using + * the stats entry associated with the old locator rather than creating a + * new one. This ensures all stats from before and after the rewrite are + * tracked in a single entry which will be properly copied to the new + * locator at transaction commit. + */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (locator.dbOid == rewrite->new_locator.dbOid && + locator.spcOid == rewrite->new_locator.spcOid && + locator.relNumber == rewrite->new_locator.relNumber) + { + pgstat_info = pgstat_prep_relation_pending(rewrite->old_locator); + goto found_entry; + } + } + } + /* Else find or make the PgStat_TableStatus entry, and update link */ - rel->pgstat_info = pgstat_prep_relation_pending(locator); + pgstat_info = pgstat_prep_relation_pending(locator); + +found_entry: + rel->pgstat_info = pgstat_info; + + /* + * For relations stats, we key by physical file location, not by relation + * OID. This means during operations like ALTER TYPE it's possible that + * the relation OID changes but the relfilenode stays the same (no actual + * rewrite needed). Unlink the old relation first. + */ + if (pgstat_info->relation != NULL && + pgstat_info->relation != rel) + { + pgstat_info->relation->pgstat_info = NULL; + pgstat_info->relation = NULL; + } /* don't allow link a stats to multiple relcache entries */ - Assert(rel->pgstat_info->relation == NULL); + Assert(pgstat_info->relation == NULL); /* mark this relation as the owner */ - rel->pgstat_info->relation = rel; + pgstat_info->relation = rel; } /* @@ -214,14 +314,37 @@ pgstat_drop_relation(Relation rel) { int nest_level = GetCurrentTransactionNestLevel(); PgStat_TableStatus *pgstat_info; + bool skip_transactional_drop = false; /* don't track stats for relations without storage */ if (!RELKIND_HAS_STORAGE(rel->rd_rel->relkind)) return; - pgstat_drop_transactional(PGSTAT_KIND_RELATION, - rel->rd_locator.dbOid, - RelFileLocatorToPgStatObjid(rel->rd_locator)); + /* Check if this drop is part of a pending rewrite */ + if (pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + if (rel->rd_locator.dbOid == rewrite->old_locator.dbOid && + rel->rd_locator.spcOid == rewrite->old_locator.spcOid && + rel->rd_locator.relNumber == rewrite->old_locator.relNumber) + { + skip_transactional_drop = true; + break; + } + } + } + + /* + * If it is part of a rewrite, drop its stats later, for example in + * AtEOXact_PgStat_Relations(), so skip it here. + */ + if (!skip_transactional_drop) + pgstat_drop_transactional(PGSTAT_KIND_RELATION, + rel->rd_locator.dbOid, + RelFileLocatorToPgStatObjid(rel->rd_locator)); if (!pgstat_should_count_relation(rel)) return; @@ -668,6 +791,48 @@ AtEOXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit) } tabstat->trans = NULL; } + + /* preserve the stats in case of rewrite */ + if (isCommit && pending_rewrites != NULL) + { + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *prev = NULL; + PgStat_PendingRewrite *current = pending_rewrites; + PgStat_PendingRewrite *next; + + /* reverse the rewrites list to process in chronological order */ + while (current != NULL) + { + next = current->next; + current->next = prev; + prev = current; + current = next; + } + + /* now process rewrites in chronological order */ + for (rewrite = prev; rewrite != NULL; rewrite = rewrite->next) + { + PgStat_EntryRef *old_entry_ref; + + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + pgstat_copy_relation_stats(rewrite->new_locator, + rewrite->old_locator, true); + + /* drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rewrite->old_locator.dbOid, + RelFileLocatorToPgStatObjid(rewrite->old_locator))) + pgstat_request_entry_refs_gc(); + } + } + + pending_rewrites = NULL; } /* @@ -683,6 +848,30 @@ AtEOSubXact_PgStat_Relations(PgStat_SubXactStatus *xact_state, bool isCommit, in PgStat_TableXactStatus *trans; PgStat_TableXactStatus *next_trans; + /* + * If we don't commit then remove the associated rewrites if any, to keep + * the rewrite chain in sync with what will be eventually committed. + */ + if (!isCommit) + { + PgStat_PendingRewrite **rewrite_ptr = &pending_rewrites; + + while (*rewrite_ptr != NULL) + { + if ((*rewrite_ptr)->nest_level >= nestDepth) + { + PgStat_PendingRewrite *to_remove = *rewrite_ptr; + + *rewrite_ptr = (*rewrite_ptr)->next; + pfree(to_remove); + } + else + { + rewrite_ptr = &((*rewrite_ptr)->next); + } + } + } + for (trans = xact_state->first; trans != NULL; trans = next_trans) { PgStat_TableStatus *tabstat; @@ -762,11 +951,19 @@ void AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) { PgStat_TableXactStatus *trans; + PgStat_PendingRewrite *rewrite; + /* + * For each tabstat, find its matching rewrite and remove it from the + * pending rewrites list. This way, after processing all tabstats, pending + * rewrites will only contain rewrite only transactions. + */ for (trans = xact_state->first; trans != NULL; trans = trans->next) { PgStat_TableStatus *tabstat PG_USED_FOR_ASSERTS_ONLY; TwoPhasePgStatRecord record; + PgStat_PendingRewrite **rewrite_ptr; + bool found_rewrite = false; Assert(trans->nest_level == 1); Assert(trans->upper == NULL); @@ -786,10 +983,83 @@ AtPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) record.locator = tabstat->locator; record.truncdropped = trans->truncdropped; + record.rewrite_nest_level = 0; + + /* + * Look for a matching rewrite and remove it from pending rewrites. We + * check three possible matches: + * + * The new_locator when stats have been added after the rewrite. The + * old_locator when stats have been added before the rewrite but not + * after. The original_locator when this tabstat is part of a rewrite + * chain. + */ + rewrite_ptr = &pending_rewrites; + while (*rewrite_ptr != NULL) + { + rewrite = *rewrite_ptr; + + if ((record.locator.dbOid == rewrite->new_locator.dbOid && + record.locator.spcOid == rewrite->new_locator.spcOid && + record.locator.relNumber == rewrite->new_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->old_locator.dbOid && + tabstat->locator.spcOid == rewrite->old_locator.spcOid && + tabstat->locator.relNumber == rewrite->old_locator.relNumber) || + (tabstat->locator.dbOid == rewrite->original_locator.dbOid && + tabstat->locator.spcOid == rewrite->original_locator.spcOid && + tabstat->locator.relNumber == rewrite->original_locator.relNumber)) + { + /* + * Found matching rewrite. Record the rewrite information and + * remove this rewrite from the list since it's now handled. + */ + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.locator = rewrite->new_locator; + found_rewrite = true; + + /* Remove from pending_rewrites list */ + *rewrite_ptr = rewrite->next; + pfree(rewrite); + break; + } + else + { + /* Move to next rewrite in the list */ + rewrite_ptr = &(rewrite->next); + } + } + + /* If no rewrite found, clear the rewrite fields */ + if (!found_rewrite) + { + memset(&record.rewrite_old_locator, 0, sizeof(RelFileLocator)); + } + + RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, + &record, sizeof(TwoPhasePgStatRecord)); + } + + /* + * Now process any rewrites still pending. These are rewrite only + * transactions. We need to preserve their stats even though there's no + * tabstat entry for them. + */ + for (rewrite = pending_rewrites; rewrite != NULL; rewrite = rewrite->next) + { + TwoPhasePgStatRecord record; + + memset(&record, 0, sizeof(TwoPhasePgStatRecord)); + record.locator = rewrite->new_locator; + record.rewrite_old_locator = rewrite->original_locator; + record.rewrite_nest_level = rewrite->nest_level; + record.truncdropped = false; RegisterTwoPhaseRecord(TWOPHASE_RM_PGSTAT_ID, 0, &record, sizeof(TwoPhasePgStatRecord)); } + + pending_rewrites = NULL; } /* @@ -812,6 +1082,8 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) tabstat = trans->parent; tabstat->trans = NULL; } + + pending_rewrites = NULL; } /* @@ -847,6 +1119,29 @@ pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, pgstat_info->counts.changed_tuples += rec->tuples_inserted + rec->tuples_updated + rec->tuples_deleted; + + if (rec->rewrite_nest_level > 0) + { + PgStat_EntryRef *old_entry_ref; + + /* Flush any pending stats for old locator first */ + old_entry_ref = pgstat_fetch_pending_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator)); + + if (old_entry_ref && old_entry_ref->pending) + pgstat_relation_flush_cb(old_entry_ref, false); + + /* Copy stats from old to new locator */ + pgstat_copy_relation_stats(rec->locator, rec->rewrite_old_locator, + true); + + /* Drop old locator's stats */ + if (!pgstat_drop_entry(PGSTAT_KIND_RELATION, + rec->rewrite_old_locator.dbOid, + RelFileLocatorToPgStatObjid(rec->rewrite_old_locator))) + pgstat_request_entry_refs_gc(); + } } /* @@ -861,9 +1156,26 @@ pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; PgStat_TableStatus *pgstat_info; + RelFileLocator target_locator; + + /* + * For aborted transactions with rewrites (like TRUNCATE), we need to + * restore stats to the old locator, not the new one. The new locator + * should be dropped since the rewrite is being rolled back. + */ + if (rec->rewrite_nest_level > 0) + { + /* Use the old locator */ + target_locator = rec->rewrite_old_locator; + } + else + { + /* No rewrite, use the original locator */ + target_locator = rec->locator; + } /* Find or create a tabstat entry for the target locator */ - pgstat_info = pgstat_prep_relation_pending(rec->locator); + pgstat_info = pgstat_prep_relation_pending(target_locator); /* Same math as in AtEOXact_PgStat, abort case */ if (rec->truncdropped) @@ -918,7 +1230,17 @@ pgstat_relation_flush_cb(PgStat_EntryRef *entry_ref, bool nowait) tabentry->numscans += lstats->counts.numscans; if (lstats->counts.numscans) { - TimestampTz t = GetCurrentTransactionStopTimestamp(); + TimestampTz t; + + /* + * Checking the transaction state due to the flush call in + * pgstat_twophase_postcommit() that would break the assertion on the + * state in GetCurrentTransactionStopTimestamp(). + */ + if (!IsTransactionState()) + t = GetCurrentTransactionStopTimestamp(); + else + t = GetCurrentTimestamp(); if (t > tabentry->lastscan) tabentry->lastscan = t; @@ -1169,3 +1491,45 @@ pgstat_reloid_to_relfilelocator(Oid reloid, RelFileLocator *locator) ReleaseSysCache(tuple); return result; } + +/* + * Mark that a relation rewrite has occurred, preserving the original locator + * so stats can be copied at transaction commit. + */ +void +pgstat_mark_rewrite(RelFileLocator old_locator, RelFileLocator new_locator) +{ + PgStat_PendingRewrite *rewrite; + PgStat_PendingRewrite *existing; + RelFileLocator original_locator = old_locator; + + for (existing = pending_rewrites; existing != NULL; existing = existing->next) + { + if (old_locator.dbOid == existing->new_locator.dbOid && + old_locator.spcOid == existing->new_locator.spcOid && + old_locator.relNumber == existing->new_locator.relNumber) + { + original_locator = existing->original_locator; + break; + } + } + + /* Allocate in TopTransactionContext memory context */ + rewrite = MemoryContextAlloc(TopTransactionContext, + sizeof(PgStat_PendingRewrite)); + + rewrite->old_locator = old_locator; + rewrite->new_locator = new_locator; + rewrite->original_locator = original_locator; + rewrite->nest_level = GetCurrentTransactionNestLevel(); + + /* Add to the list */ + rewrite->next = pending_rewrites; + pending_rewrites = rewrite; +} + +void +pgstat_clear_rewrite(void) +{ + pending_rewrites = NULL; +} diff --git a/src/backend/utils/activity/pgstat_xact.c b/src/backend/utils/activity/pgstat_xact.c index 5e2d69e6297..8ed8f5317f3 100644 --- a/src/backend/utils/activity/pgstat_xact.c +++ b/src/backend/utils/activity/pgstat_xact.c @@ -55,6 +55,8 @@ AtEOXact_PgStat(bool isCommit, bool parallel) } pgStatXactStack = NULL; + pgstat_clear_rewrite(); + /* Make sure any stats snapshot is thrown away */ pgstat_clear_snapshot(); } @@ -360,8 +362,29 @@ create_drop_transactional_internal(PgStat_Kind kind, Oid dboid, uint64 objid, bo void pgstat_create_transactional(PgStat_Kind kind, Oid dboid, uint64 objid) { - if (pgstat_get_entry_ref(kind, dboid, objid, false, NULL)) + PgStat_EntryRef *entry_ref; + + entry_ref = pgstat_get_entry_ref(kind, dboid, objid, false, NULL); + + if (entry_ref) { + /* + * For relations stats, we key by physical file location, not by + * relation OID. This means during operations like ALTER TYPE where + * the relation OID changes but the relfilenode stays the same (no + * actual rewrite needed), we'll find an existing entry. + * + * This is expected behavior, we want to preserve stats across the + * catalog change. Simply reset and recreate the entry for the new + * relation OID without warning. + */ + if (kind == PGSTAT_KIND_RELATION) + { + pgstat_reset(kind, dboid, objid); + create_drop_transactional_internal(kind, dboid, objid, true); + return; + } + ereport(WARNING, errmsg("resetting existing statistics for kind %s, db=%u, oid=%" PRIu64, (pgstat_get_kind_info(kind))->name, dboid, diff --git a/src/backend/utils/cache/relcache.c b/src/backend/utils/cache/relcache.c index 0572ab424e7..d84058db102 100644 --- a/src/backend/utils/cache/relcache.c +++ b/src/backend/utils/cache/relcache.c @@ -87,6 +87,7 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" +#include "utils/pgstat_internal.h" #include "utils/relmapper.h" #include "utils/resowner.h" #include "utils/snapmgr.h" @@ -3782,6 +3783,7 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) MultiXactId minmulti = InvalidMultiXactId; TransactionId freezeXid = InvalidTransactionId; RelFileLocator newrlocator; + RelFileLocator oldrlocator = relation->rd_locator; if (!IsBinaryUpgrade) { @@ -3953,6 +3955,10 @@ RelationSetNewRelfilenumber(Relation relation, char persistence) table_close(pg_class, RowExclusiveLock); + /* Mark that a rewrite happened */ + if (RELKIND_HAS_STORAGE(relation->rd_rel->relkind)) + pgstat_mark_rewrite(oldrlocator, newrlocator); + /* * Make the pg_class row change or relation map change visible. This will * cause the relcache entry to get updated, too. diff --git a/src/include/pgstat.h b/src/include/pgstat.h index 941a39bd16e..5344ece9ecf 100644 --- a/src/include/pgstat.h +++ b/src/include/pgstat.h @@ -697,7 +697,7 @@ extern PgStat_FunctionCounts *find_funcstat_entry(Oid func_id); extern void pgstat_create_relation(Relation rel); extern void pgstat_drop_relation(Relation rel); -extern void pgstat_copy_relation_stats(Relation dst, Relation src); +extern void pgstat_copy_relation_stats(RelFileLocator dst, RelFileLocator src, bool increment); extern void pgstat_init_relation(Relation rel); extern void pgstat_assoc_relation(Relation rel); @@ -709,6 +709,9 @@ extern void pgstat_report_vacuum(Relation rel, PgStat_Counter livetuples, extern void pgstat_report_analyze(Relation rel, PgStat_Counter livetuples, PgStat_Counter deadtuples, bool resetcounter, TimestampTz starttime); +extern void pgstat_mark_rewrite(RelFileLocator old_locator, + RelFileLocator new_locator); +extern void pgstat_clear_rewrite(void); /* * If stats are enabled, but pending data hasn't been prepared yet, call diff --git a/src/tools/pgindent/typedefs.list b/src/tools/pgindent/typedefs.list index 8cf40c87043..1544f54e918 100644 --- a/src/tools/pgindent/typedefs.list +++ b/src/tools/pgindent/typedefs.list @@ -2349,6 +2349,7 @@ PgStat_LockEntry PgStat_PendingDroppedStatsItem PgStat_PendingIO PgStat_PendingLock +PgStat_PendingRewrite PgStat_SLRUStats PgStat_ShmemControl PgStat_Snapshot -- 2.34.1 --Uc+t0mng6H5VV2zk-- ^ permalink raw reply [nested|flat] 37+ messages in thread
end of thread, other threads:[~2026-05-18 15:20 UTC | newest] Thread overview: 37+ messages (download: mbox mbox.gz follow: Atom feed) -- links below jump to the message on this page -- 2023-08-01 14:44 Fix error handling in be_tls_open_server() Sergey Shinderuk <[email protected]> 2023-08-23 13:23 ` Daniel Gustafsson <[email protected]> 2023-08-23 21:47 ` Jacob Champion <[email protected]> 2023-08-23 21:54 ` Daniel Gustafsson <[email protected]> 2023-08-24 08:11 ` Sergey Shinderuk <[email protected]> 2023-08-24 08:38 ` Daniel Gustafsson <[email protected]> 2023-08-24 09:13 ` Sergey Shinderuk <[email protected]> 2023-08-25 01:25 ` Michael Paquier <[email protected]> 2023-08-28 15:12 ` Jacob Champion <[email protected]> 2023-08-28 17:39 ` Sergey Shinderuk <[email protected]> 2023-09-18 12:35 ` Daniel Gustafsson <[email protected]> 2023-09-19 00:54 ` Michael Paquier <[email protected]> 2023-09-19 08:06 ` Sergey Shinderuk <[email protected]> 2023-09-20 08:42 ` Daniel Gustafsson <[email protected]> 2023-09-20 08:55 ` Sergey Shinderuk <[email protected]> 2023-09-20 08:58 ` Daniel Gustafsson <[email protected]> 2023-09-22 09:43 ` Daniel Gustafsson <[email protected]> 2025-11-04 13:52 [PATCH v14 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v9 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v11 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v7 3/3] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v11 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v15 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v15 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v8 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v9 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v13 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v12 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v10 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v13 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v7 3/3] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v10 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v12 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v14 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2025-11-04 13:52 [PATCH v8 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2026-05-18 15:20 [PATCH v16 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]> 2026-05-18 15:20 [PATCH v16 2/2] handle relation statistics correctly during rewrites Bertrand Drouvot <[email protected]>
This inbox is served by agora; see mirroring instructions for how to clone and mirror all data and code used for this inbox