pg.ddx.io reorg/pg_repack GitHub issues and pull requests
help / color / mirror / Atom feed[reorg/pg_repack] PR #480: Fix NOTICE suppression for pg_regress with PGAPPNAME
6+ messages / 2 participants
[nested] [flat]
* [reorg/pg_repack] PR #480: Fix NOTICE suppression for pg_regress with PGAPPNAME
@ 2026-02-27 12:10 "bayandin (@bayandin)" <noreply+bayandin@github.com>
0 siblings, 0 replies; 6+ messages in thread
From: bayandin (@bayandin) @ 2026-02-27 12:10 UTC (permalink / raw)
To: reorg/pg_repack <noreply+reorg-pg_repack@github.com>
This PR fixes flaky regression tests caused by unsuppressed NOTICE messages
leaking into `make installcheck` output.
The NOTICE about waiting for transactions is meant to be suppressed when
running under pg_regress, but since PostgreSQL 10
(postgres/postgres@a4327296df7), pg_regress sets `PGAPPNAME` to
`pg_regress/<testname>` instead of just `pg_regress`. The existing
`strcmp` check only matched the exact string `"pg_regress"`, so the
suppression silently stopped working on PG 10+.
When a concurrent transaction happens to be present during the test run,
the NOTICE leaks into output and causes a spurious diff like:
```diff
@@ -380,6 +380,7 @@
INFO: repacking table "public.child_a_2"
INFO: repacking table "public.parent_a"
INFO: repacking table "public.partition_a_1"
+NOTICE: Waiting for 1 transactions to finish. First PID: 572
INFO: repacking table "public.partition_a_2"
```
- Match both `"pg_regress/<testname>"` (PG 10+) and bare `"pg_regress"` (PG <10) PGAPPNAME formats
- Update related comments to document both PGAPPNAME formats
Debugged and drafted the fix with assistance from Cursor IDE and Claude
claude-4.6-opus (Anthropic).
inline patch 1: bin/pg_repack.c (+6/-3 lines) download inline-1.patch
diff --git a/bin/pg_repack.c b/bin/pg_repack.c
index aa7476d..ac8453b 100644
--- a/bin/pg_repack.c
+++ b/bin/pg_repack.c
@@ -71,7 +71,7 @@ const char *PROGRAM_VERSION = "unknown";
* based on pg_backend_pid() and application_name), but that shouldn't hurt
* anything. Also, the test of application_name is not bulletproof -- for
* instance, the application name when running installcheck will be
- * pg_regress.
+ * pg_regress/<testname>, or just pg_regress on PostgreSQL <10.
*/
#define SQL_XID_SNAPSHOT_90200 \
"SELECT coalesce(array_agg(l.virtualtransaction), '{}') " \
@@ -1238,7 +1238,8 @@ repack_one_table(repack_table *table, const char *orderby)
int j;
/* appname will be "pg_repack" in normal use on 9.0+, or
- * "pg_regress" when run under `make installcheck`
+ * "pg_regress/<testname>" when run under `make installcheck`
+ * ("pg_regress" on PostgreSQL <10).
*/
const char *appname = getenv("PGAPPNAME");
@@ -1593,7 +1594,9 @@ repack_one_table(repack_table *table, const char *orderby)
* noise which would trip up pg_regress.
*/
- if (!appname || strcmp(appname, "pg_regress") != 0)
+ if (!appname ||
+ (strncmp(appname, "pg_regress/", strlen("pg_regress/")) != 0 &&
+ strcmp(appname, "pg_regress") != 0))
{
elog(NOTICE, "Waiting for %d transactions to finish. First PID: %s", num, PQgetvalue(res, 0, 0));
}
^ permalink raw [nested|flat] 6+ messages in thread
* Re: [reorg/pg_repack] PR #480: Fix NOTICE suppression for pg_regress with PGAPPNAME
@ 2026-02-28 15:50 ` "za-arthur (@za-arthur)" <noreply+za-arthur@github.com>
4 siblings, 0 replies; 6+ messages in thread
From: za-arthur (@za-arthur) @ 2026-02-28 15:50 UTC (permalink / raw)
To: reorg/pg_repack <noreply+reorg-pg_repack@github.com>
@bayandin thank you for the fix.
It looks good to me. Although I'm curious do you reproduce this reliably? Locally I cannot reproduce it when I run `make installcheck`.
^ permalink raw [nested|flat] 6+ messages in thread
* Re: [reorg/pg_repack] PR #480: Fix NOTICE suppression for pg_regress with PGAPPNAME
@ 2026-03-02 15:47 ` "bayandin (@bayandin)" <noreply+bayandin@github.com>
4 siblings, 0 replies; 6+ messages in thread
From: bayandin (@bayandin) @ 2026-03-02 15:47 UTC (permalink / raw)
To: reorg/pg_repack <noreply+reorg-pg_repack@github.com>
> Although I'm curious do you reproduce this reliably?
Unfortunately no, I didn't manage to reproduce it reliably either. It happens from time to time in our CI
^ permalink raw [nested|flat] 6+ messages in thread
* Re: [reorg/pg_repack] PR #480: Fix NOTICE suppression for pg_regress with PGAPPNAME
@ 2026-03-02 16:14 ` "za-arthur (@za-arthur)" <noreply+za-arthur@github.com>
4 siblings, 0 replies; 6+ messages in thread
From: za-arthur (@za-arthur) @ 2026-03-02 16:14 UTC (permalink / raw)
To: reorg/pg_repack <noreply+reorg-pg_repack@github.com>
Thanks! The fix looks correct anyway. I'm merging it.
^ permalink raw [nested|flat] 6+ messages in thread
* Re: [reorg/pg_repack] PR #480: Fix NOTICE suppression for pg_regress with PGAPPNAME
@ 2026-03-02 16:14 ` "za-arthur (@za-arthur)" <noreply+za-arthur@github.com>
4 siblings, 0 replies; 6+ messages in thread
From: za-arthur (@za-arthur) @ 2026-03-02 16:14 UTC (permalink / raw)
To: reorg/pg_repack <noreply+reorg-pg_repack@github.com>
@za-arthur merged this (commit f142143e3cdf).
^ permalink raw [nested|flat] 6+ messages in thread
* Re: [reorg/pg_repack] PR #480: Fix NOTICE suppression for pg_regress with PGAPPNAME
@ 2026-03-02 16:14 ` "za-arthur (@za-arthur)" <noreply+za-arthur@github.com>
4 siblings, 0 replies; 6+ messages in thread
From: za-arthur (@za-arthur) @ 2026-03-02 16:14 UTC (permalink / raw)
To: reorg/pg_repack <noreply+reorg-pg_repack@github.com>
@za-arthur closed this.
^ permalink raw [nested|flat] 6+ messages in thread
end of thread, other threads:[~2026-03-02 16:14 UTC | newest]
Thread overview: 6+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2026-02-27 12:10 [reorg/pg_repack] PR #480: Fix NOTICE suppression for pg_regress with PGAPPNAME "bayandin (@bayandin)" <noreply+bayandin@github.com>
2026-02-28 15:50 ` "za-arthur (@za-arthur)" <noreply+za-arthur@github.com>
2026-03-02 15:47 ` "bayandin (@bayandin)" <noreply+bayandin@github.com>
2026-03-02 16:14 ` "za-arthur (@za-arthur)" <noreply+za-arthur@github.com>
2026-03-02 16:14 ` "za-arthur (@za-arthur)" <noreply+za-arthur@github.com>
2026-03-02 16:14 ` "za-arthur (@za-arthur)" <noreply+za-arthur@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