Re: Reduce SyncRepLock contention on the commit path

From: Vadim Ponomarev <vbponomarev(at)gmail(dot)com>
To: Yura Sokolov <y(dot)sokolov(at)postgrespro(dot)ru>
Cc: pgsql-hackers(at)postgresql(dot)org
Subject: Re: Reduce SyncRepLock contention on the commit path
Date: 2026-10-06 15:13:35
Message-ID: CANeUpr_1LF1rEjPq3fEfRUSCdFkeivQv+GhcXSLnze4Ktq4BDA@mail.gmail.com
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

v5 attached. cfbot failed 058_syncrep_release on both Windows tasks:

# die: SIGCONT walsender: Invalid argument at
.../058_syncrep_release.pl line 92.

Both scenarios in that test hold a process with SIGSTOP/SIGCONT, which
Windows doesn't have. 0003 now skips the test there, the same way
019_replslot_limit.pl and 038_walsnd_shutdown_timeout.pl skip their
kill-based parts. Nothing else changed.

Regards,
Vadim Ponomarev

Attachment Content-Type Size
v5-0001-Wake-the-released-sync-rep-waiters-after-the-queu.patch text/x-patch 6.5 KB
v5-0002-Compute-the-synced-positions-before-taking-the-sy.patch text/x-patch 3.1 KB
v5-0003-Release-the-sync-rep-waiters-once-per-drained-bat.patch text/x-patch 16.2 KB
v5-0004-Let-a-committer-whose-acknowledgement-already-arr.patch text/x-patch 7.9 KB
v5-0005-Add-stream-regress-test-for-synchronous-replicati.patch text/x-patch 12.1 KB

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Heikki Linnakangas 2026-10-06 15:29:00 Re: enhancing pg_basebackup speeds up to ~23Gbps (small fixes + io_uring/Direct I/O)
Previous Message Manu 2026-10-06 15:07:54 Re: Logical replication: lost updates/deletes and invalid log messages caused by SnapshotDirty + concurrent updates