From: | Jaime Casanova <jaime(at)2ndquadrant(dot)com> |
---|---|
To: | PostgreSQL-development <pgsql-hackers(at)postgresql(dot)org> |
Subject: | pg_stats_recovery view |
Date: | 2012-01-15 07:50:00 |
Message-ID: | CAJKUy5jkCT_xUOVF0Tqd_ofYSuiWjC7dFMxVB+D_wbw5HxKYJw@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hi,
Attached is a patch thats implements a pg_stat_recovery view that
keeps counters about processed wal records. I just notice that it
still lacks documentation but i will add it during the week.
Because it tracks redo time this introduces to GetCurrentTimestamp()
calls to the redo main loop, so i add a track_recovery GUC so only the
people that wants the view has to spent time in those calls.
Probably the most controversial part of the patch will be the addition
of a new column in RmgrData that is a pointer to a new function for
*_short_desc() these functions are similar to the *_desc functions
that already exists and that is called via rm_desc but instead of
giving full details about the record being processed it just inform of
the type of the record, for example the 2 first columns will look
something like:
rmgr: XLOG
wal_record_type: xlog switch
--
Jaime Casanova www.2ndQuadrant.com
Professional PostgreSQL: Soporte 24x7 y capacitación
Attachment | Content-Type | Size |
---|---|---|
stats_recovery.v1.patch | text/x-patch | 58.0 KB |
From | Date | Subject | |
---|---|---|---|
Next Message | Magnus Hagander | 2012-01-15 08:17:54 | Re: automating CF submissions (was xlog location arithmetic) |
Previous Message | Peter Geoghegan | 2012-01-15 07:26:31 | Our poll() based WaitLatch implementation is broken |