From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
---|---|
To: | Greg Smith <gsmith(at)gregsmith(dot)com> |
Cc: | Greg Stark <greg(dot)stark(at)enterprisedb(dot)com>, Simon Riggs <simon(at)2ndQuadrant(dot)com>, pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Re: Assert Levels |
Date: | 2008-09-21 22:42:36 |
Message-ID: | 13370.1222036956@sss.pgh.pa.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Greg Smith <gsmith(at)gregsmith(dot)com> writes:
> The next time I'm doing some performance testing I'll try to quantify how
> much damage the expensive ones do by playing with pg_config_manual.h.
> Normally I'm testing with 1GB+ of shared_buffers which makes the current
> assert scheme unusable.
There is a commit-time scan of the buffers for the sole purpose of
asserting a few things about their state. That's one of the things
we'd need to turn off for a "cheap asserts only" mode I think.
If you want to try to quantify what "cheap asserts" might do, you
should pull out the #ifdef USE_ASSERT_CHECKING code here:
check_list_invariants in list.c
the loops in AtEOXact_Buffers and AtEOXact_LocalBuffers
the loop in AtEOXact_CatCache
the test that makes AtEOXact_RelationCache scan the relcache
even when !need_eoxact_work
in addition to the memory-related stuff.
regards, tom lane
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2008-09-22 00:09:02 | Re: Toasted table not deleted when no out of line columns left |
Previous Message | Andrew Dunstan | 2008-09-21 22:32:45 | Re: parallel pg_restore |