bg writer went away after reload

Started by Matthew Chambersabout 12 years ago4 messagesgeneral
Jump to latest
#1Matthew Chambers
mchambers@wetafx.co.nz

Hi, just wondering if this is normal, DB is operating just fine.

I upped bgwriter_lru_maxpages to 200 and issued a reload. Normally, I'd
see the bgwriter constantly churning as one of my main I/O using
processes, but now I have:

postgres: wal writer process
postgres: checkpointer process

The wal writer seems to have taken over. Does this make sense?

-Matt

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

#2Venkata Balaji Nagothi
vbnpgc@gmail.com
In reply to: Matthew Chambers (#1)
Re: bg writer went away after reload

On Tue, Mar 11, 2014 at 8:30 AM, Matthew Chambers <mchambers@wetafx.co.nz>wrote:

Hi, just wondering if this is normal, DB is operating just fine.

I upped bgwriter_lru_maxpages to 200 and issued a reload. Normally, I'd
see the bgwriter constantly churning as one of my main I/O using processes,
but now I have:

postgres: wal writer process
postgres: checkpointer process

The wal writer seems to have taken over. Does this make sense?

What WAL writer does is completely different from the way bgwriter
functions. These two critical background processes of PostgreSQL performing
independent I/O operations independently.One cannot take over another.

which version of Postgres is this ?

Do you see anything in the Postgres logs ? Do you see any message which
indicates that reloading of the new configuration in postgresql.conf file
was successful ?

Venkata Balaji N

Sr. Database Administrator
Fujitsu Australia

#3Matthew Chambers
mchambers@wetafx.co.nz
In reply to: Venkata Balaji Nagothi (#2)
Re: bg writer went away after reload

This is postgres 9.3.2.

This is what the log shows.

Mar 11 08:16:29 jupiter521 postgres[2026]: [8-1] 2014-03-11 08:16:29
NZDTLOG: received SIGHUP, reloading configuration files
Mar 11 08:16:29 jupiter521 postgres[2026]: [9-1] 2014-03-11 08:16:29
NZDTLOG: parameter "bgwriter_lru_maxpages" changed to "200"

Here are the processes I have running besides the connections.

postgres 2028 0.0 8.3 17245532 8279356 ? Ss Mar09 2:42
postgres: checkpointer process
postgres 2029 0.0 0.1 17245272 107900 ? Ss Mar09 0:08
postgres: writer process
postgres 2030 0.2 0.0 17245272 34248 ? Ss Mar09 6:44
postgres: wal writer process
postgres 2031 0.0 0.0 17246164 2596 ? Ss Mar09 0:09
postgres: autovacuum launcher process
postgres 2032 0.0 0.0 18152 1244 ? Ss Mar09 0:06
postgres: archiver process last was 0000000100000202000000F8
postgres 2033 0.0 0.0 18568 1636 ? Ss Mar09 1:47
postgres: stats collector process
postgres 3914 0.4 0.0 17246520 2844 ? Ss Mar09 14:04
postgres: wal sender process postgres 192.168.122.54(48686) streaming
202/F996C000

Is it the "writer process"? I was sure it was called the background
writer before.

-Matt

Show quoted text

On 11/03/14 12:03, Venkata Balaji Nagothi wrote:

On Tue, Mar 11, 2014 at 8:30 AM, Matthew Chambers
<mchambers@wetafx.co.nz <mailto:mchambers@wetafx.co.nz>> wrote:

Hi, just wondering if this is normal, DB is operating just fine.

I upped bgwriter_lru_maxpages to 200 and issued a reload.
Normally, I'd see the bgwriter constantly churning as one of my
main I/O using processes, but now I have:

postgres: wal writer process
postgres: checkpointer process

The wal writer seems to have taken over. Does this make sense?

What WAL writer does is completely different from the way bgwriter
functions. These two critical background processes of PostgreSQL
performing independent I/O operations independently.One cannot take
over another.

which version of Postgres is this ?

Do you see anything in the Postgres logs ? Do you see any message
which indicates that reloading of the new configuration in
postgresql.conf file was successful ?

Venkata Balaji N

Sr. Database Administrator
Fujitsu Australia

#4Venkata Balaji Nagothi
vbnpgc@gmail.com
In reply to: Matthew Chambers (#3)
Re: bg writer went away after reload

Yes. It is the "writer process". It is still called as background writer
process. It displays as "writer process" since PostgreSQL-8.0.

Venkata Balaji N

Sr. Database Administrator
Fujitsu Australia

On Tue, Mar 11, 2014 at 10:32 AM, Matthew Chambers
<mchambers@wetafx.co.nz>wrote:

Show quoted text

This is postgres 9.3.2.

This is what the log shows.

Mar 11 08:16:29 jupiter521 postgres[2026]: [8-1] 2014-03-11 08:16:29
NZDTLOG: received SIGHUP, reloading configuration files
Mar 11 08:16:29 jupiter521 postgres[2026]: [9-1] 2014-03-11 08:16:29
NZDTLOG: parameter "bgwriter_lru_maxpages" changed to "200"

Here are the processes I have running besides the connections.

postgres 2028 0.0 8.3 17245532 8279356 ? Ss Mar09 2:42 postgres:
checkpointer process
postgres 2029 0.0 0.1 17245272 107900 ? Ss Mar09 0:08 postgres:
writer process
postgres 2030 0.2 0.0 17245272 34248 ? Ss Mar09 6:44 postgres:
wal writer process
postgres 2031 0.0 0.0 17246164 2596 ? Ss Mar09 0:09 postgres:
autovacuum launcher process
postgres 2032 0.0 0.0 18152 1244 ? Ss Mar09 0:06 postgres:
archiver process last was 0000000100000202000000F8
postgres 2033 0.0 0.0 18568 1636 ? Ss Mar09 1:47 postgres:
stats collector process
postgres 3914 0.4 0.0 17246520 2844 ? Ss Mar09 14:04 postgres:
wal sender process postgres 192.168.122.54(48686) streaming 202/F996C000

Is it the "writer process"? I was sure it was called the background
writer before.

-Matt

On 11/03/14 12:03, Venkata Balaji Nagothi wrote:

On Tue, Mar 11, 2014 at 8:30 AM, Matthew Chambers <mchambers@wetafx.co.nz

wrote:

Hi, just wondering if this is normal, DB is operating just fine.

I upped bgwriter_lru_maxpages to 200 and issued a reload. Normally, I'd
see the bgwriter constantly churning as one of my main I/O using processes,
but now I have:

postgres: wal writer process
postgres: checkpointer process

The wal writer seems to have taken over. Does this make sense?

What WAL writer does is completely different from the way bgwriter
functions. These two critical background processes of PostgreSQL performing
independent I/O operations independently.One cannot take over another.

which version of Postgres is this ?

Do you see anything in the Postgres logs ? Do you see any message which
indicates that reloading of the new configuration in postgresql.conf file
was successful ?

Venkata Balaji N

Sr. Database Administrator
Fujitsu Australia