Re: duration logging setting in 7.4 - Mailing list pgsql-performance

Ryszard Lach wrote:
> If I turn on log_min_duration_statement (i.e. set to 0), log_statement and
> log_duration, then I receive something like that
>
> Nov 17 22:33:27 postgres[22945]: [29231-1] LOG:  statement:
> Nov 17 22:33:27 postgres[22945]: [29232-1] LOG:  duration: 0.198 ms
> Nov 17 22:33:27 postgres[22945]: [29233-1] LOG:  duration: 0.198 ms  statement:
> Nov 17 22:33:27 postgres[22946]: [29231-1] LOG:  statement:
> Nov 17 22:33:27 postgres[22946]: [29232-1] LOG:  duration: 0.191 ms
> Nov 17 22:33:27 postgres[22946]: [29233-1] LOG:  duration: 0.191 ms  statement:
> Nov 17 22:33:27 postgres[22678]: [147134-1] LOG:  statement: select * from cms where id=1465
> Nov 17 22:33:27 postgres[22679]: [154907-1] LOG:  statement:
> Nov 17 22:33:27 postgres[22679]: [154908-1] LOG:  duration: 0.867 ms
> Nov 17 22:33:27 postgres[22679]: [154909-1] LOG:  duration: 0.867 ms  statement:
> Nov 17 22:33:27 postgres[22678]: [147135-1] LOG:  duration: 1.458 ms
> Nov 17 22:33:27 postgres[22678]: [147136-1] LOG:  duration: 1.458 ms  statement: select * from cms where id=1465
> Nov 17 22:33:27 postgres[22680]: [158366-1] LOG:  statement:
> Nov 17 22:33:27 postgres[22680]: [158367-1] LOG:  duration: 0.620 ms
> Nov 17 22:33:27 postgres[22680]: [158368-1] LOG:  duration: 0.620 ms  statement:
> Nov 17 22:33:27 postgres[22681]: [161294-1] LOG:  statement:
> Nov 17 22:33:27 postgres[22681]: [161295-1] LOG:  duration: 0.650 ms
>
> It seems, that log_duration is responsible only for "duration:" lines,
> log_statement - for "statement:" ones, and "log_min_duration_statement" - for
> "duration: .* statement:". I think, that the above output should exclude losing
> of data by syslog from further delibarations. Do you thing that could be
> a bug?

Yes, the problem is not related to syslog.  Are you using prepared
queries, perhaps?  I don't think those show the query, but it seems we
should display something better than blanks.

> There is another one thing: logs from the same database running on 7.3 and the same
> application contained lines like 'select getdatabaseencoding()', 'select
> datestyle()' and similar (not used by application explicite, probably
> added by JDBC driver), now they are missed - maybe this is the
> problem?

No, those are missing because the new 7.4 wire protocol doesn't require
those queries anymore --- the data is send automatically.

--
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 359-1001
  +  If your life is a hard drive,     |  13 Roberts Road
  +  Christ can be your backup.        |  Newtown Square, Pennsylvania 19073