pg.ddx.io  reorg/pg_repack GitHub issues and pull requests  
help / color / mirror / Atom feed
From: arthurprs (@arthurprs) <noreply+arthurprs@github.com>
To: reorg/pg_repack <noreply+reorg-pg_repack@github.com>
Subject: [reorg/pg_repack] PR #183: Add options for lock statement timeout
Date: Wed, 06 Jun 2018 11:29:37 +0000
Message-ID: <gh-reorg-pg_repack-183@github.com> (raw)

This change was needed to get pg_repack to get the locks in our reasonably busy servers.

inline patch 1: bin/pg_repack.c, doc/pg_repack.rst, doc/pg_repack_jp.rst (+20/-2 lines) download inline-1.patchdiff --git a/bin/pg_repack.c b/bin/pg_repack.c index 5bcee125..6382bb58 100644 --- a/bin/pg_repack.c +++ b/bin/pg_repack.c @@ -256,6 +256,7 @@ static unsigned int temp_obj_num = 0; /* temporary objects counter */ static bool no_kill_backend = false; /* abandon when timed-out */ static bool no_superuser_check = false; static SimpleStringList exclude_extension_list = {NULL, NULL}; /* don't repack tables of these extensions */ +static int lock_wait_max = 1; /* Max lock statement timeout, in seconds */ /* buffer should have at least 11 bytes */ static char * @@ -285,6 +286,7 @@ static pgut_option options[] = { 'b', 'D', "no-kill-backend", &no_kill_backend }, { 'b', 'k', "no-superuser-check", &no_superuser_check }, { 'l', 'C', "exclude-extension", &exclude_extension_list }, + { 'i', 'L', "lock-wait-max", &lock_wait_max }, { 0 }, }; @@ -308,6 +310,12 @@ main(int argc, char *argv[]) if (dryrun) elog(INFO, "Dry run enabled, not executing repack"); + if (wait_timeout < lock_wait_max) + { + ereport(ERROR, (errcode(EINVAL), + errmsg("wait-timeout needs to be >= lock-wait-max"))); + } + if (r_index.head || only_indexes) { if (r_index.head && table_list.head) @@ -1789,7 +1797,7 @@ lock_access_share(PGconn *conn, Oid relid, const char *target_name) break; /* wait for a while to lock the table. */ - wait_msec = Min(1000, i * 100); + wait_msec = Min(lock_wait_max * 1000, Max(lock_wait_max * 1000 * i / 10, 100)); printfStringInfo(&sql, "SET LOCAL statement_timeout = %d", wait_msec); pgut_command(conn, sql.data, 0, NULL); @@ -1930,7 +1938,7 @@ lock_exclusive(PGconn *conn, const char *relid, const char *lock_query, bool sta } /* wait for a while to lock the table. */ - wait_msec = Min(1000, i * 100); + wait_msec = Min(lock_wait_max * 1000, Max(lock_wait_max * 1000 * i / 10, 100)); snprintf(sql, lengthof(sql), "SET LOCAL statement_timeout = %d", wait_msec); pgut_command(conn, sql, 0, NULL); @@ -2351,6 +2359,7 @@ pgut_help(bool details) printf(" -i, --index=INDEX move only the specified index\n"); printf(" -x, --only-indexes move only indexes of the specified table\n"); printf(" -T, --wait-timeout=SECS timeout to cancel other backends on conflict\n"); + printf(" -L, --lock-wait-max=SECS lock statement max wait time\n"); printf(" -D, --no-kill-backend don't kill other backends when timed out\n"); printf(" -Z, --no-analyze don't analyze at end\n"); printf(" -k, --no-superuser-check skip superuser checks in client\n"); diff --git a/doc/pg_repack.rst b/doc/pg_repack.rst index 2aecb26e..d65a838e 100644 --- a/doc/pg_repack.rst +++ b/doc/pg_repack.rst @@ -119,6 +119,7 @@ Options: -i, --index=INDEX move only the specified index -x, --only-indexes move only indexes of the specified table -T, --wait-timeout=SECS timeout to cancel other backends on conflict + -L, --lock-wait-max=SECS lock statement max wait time -D, --no-kill-backend don't kill other backends when timed out -Z, --no-analyze don't analyze at end -k, --no-superuser-check skip superuser checks in client @@ -207,6 +208,13 @@ Reorg Options backends after twice this timeout has passed. The default is 60 seconds. +``-L SECS``, ``--lock-wait-max=SECS`` + When pg_repack needs to take table locks (shared or exclusive) it'll wait + up to this many seconds before canceling the lock request and trying again. + This is done to allow concurrent queries to make progress and not get + queued up behind the pg_repack lock request. Consider increasing this + timeout if pg_repack fails to acquire table locks. The default is 1 second. + ``-D``, ``--no-kill-backend`` Skip to repack table if the lock cannot be taken for duration specified ``--wait-timeout``, instead of cancelling conflicting queries. The default diff --git a/doc/pg_repack_jp.rst b/doc/pg_repack_jp.rst index b4ac5c5c..4b49af5f 100644 --- a/doc/pg_repack_jp.rst +++ b/doc/pg_repack_jp.rst @@ -207,6 +207,7 @@ pg_repackもしくはpg_reorgの古いバージョンからのアップグレー -i, --index=INDEX move only the specified index -x, --only-indexes move only indexes of the specified table -T, --wait-timeout=SECS timeout to cancel other backends on conflict + -L, --lock-wait-max=SECS lock statement max wait time -D, --no-kill-backend don't kill other backends when timed out -Z, --no-analyze don't analyze at end -k, --no-superuser-check skip superuser checks in client

view thread (15+ messages)  latest in thread

Message-ID: <gh-reorg-pg_repack-183@github.com>
Permalink:  ../gh-reorg-pg_repack-183@github.com/

 ·  · 


This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox