Re: LVM snapshots

From: Murthy Kambhampaty <murthy(dot)kambhampaty(at)goeci(dot)com>
To: 'Tom Lane' <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Matt <matt(at)ymogen(dot)net>
Cc: 'Peter Eisentraut' <peter_e(at)gmx(dot)net>, pgsql-admin(at)postgresql(dot)org
Subject: Re: LVM snapshots
Date: 2003-04-04 22:32:08
Message-ID: 2D92FEBFD3BE1346A6C397223A8DD3FC0921D1@THOR.goeci.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-admin

Tom Lane [mailto:tgl(at)sss(dot)pgh(dot)pa(dot)us] writes:
>> NOTICE: Rel tbl2: Uninitialized page 'nnnnn' - fixing
>Hmm. This amounts to testing what happens after a Postgres crash (the
>copied PGDATA will look exactly like the system had gone down at the
>instant of the freeze). In that scenario zeroed pages aren't too
>improbable, although I wouldn't expect to see a large number of them.
It turns out there aren't many of them, after all, and they seem to be
truncated by vacuum.
Here's the vacuum report from the log, for the two files "afftected":
2003-04-02 18:33:46 [4657] DEBUG: --Relation tbl2--
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54300 -
fixing
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54301 -
fixing
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54302 -
fixing
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54303 -
fixing
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54304 -
fixing
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54305 -
fixing
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54306 -
fixing
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54307 -
fixing
2003-04-02 18:34:33 [4657] NOTICE: Rel tbl2: Uninitialized page 54308 -
fixing
2003-04-02 18:34:34 [4657] DEBUG: Removed 49173 tuples in 3946 pages.
CPU 0.27s/1.07u sec elapsed 1.34 sec.
2003-04-02 18:34:34 [4657] DEBUG: Pages 54309: Changed 54300, Empty 0;
Tup 620366: Vac 49173, Keep 0, UnUsed 0.
Total CPU 9.78s/2.35u sec elapsed 47.74 sec.
2003-04-02 18:34:35 [4657] DEBUG: Truncated 54309 --> 50355 pages.
CPU 0.38s/0.07u sec elapsed 0.44 sec.

2003-04-02 18:34:55 [4657] DEBUG: --Relation tbl3--
2003-04-02 18:35:31 [4657] NOTICE: Rel tbl3: Uninitialized page 33172 -
fixing
2003-04-02 18:35:31 [4657] NOTICE: Rel tbl3: Uninitialized page 33173 -
fixing
2003-04-02 18:35:31 [4657] NOTICE: Rel tbl3: Uninitialized page 33174 -
fixing
2003-04-02 18:35:31 [4657] NOTICE: Rel tbl3: Uninitialized page 33175 -
fixing
2003-04-02 18:35:32 [4657] DEBUG: Removed 38874 tuples in 3230 pages.
CPU 0.20s/0.88u sec elapsed 1.08 sec.
2003-04-02 18:35:32 [4657] DEBUG: Pages 33176: Changed 33172, Empty 0;
Tup 354316: Vac 38874, Keep 0, UnUsed 0.
Total CPU 5.66s/1.71u sec elapsed 36.95 sec.
2003-04-02 18:35:32 [4657] DEBUG: Truncated 33176 --> 29943 pages.
CPU 0.41s/0.02u sec elapsed 0.42 sec.

>Are you running with a nondefault WAL_BUFFERS setting?
WAL_BUFFERS=64

> [ Snip ]
Thanks for the explanation, and for taking a second look ;-)

Responses

Browse pgsql-admin by date

  From Date Subject
Next Message Tom Lane 2003-04-04 22:34:14 Re: LVM snapshots
Previous Message Tom Lane 2003-04-04 18:29:00 Re: LVM snapshots