From: | Dave Cramer <pg(at)fastcrypt(dot)com> |
---|---|
To: | Kevin Dorne <kevin(at)catalyst(dot)net(dot)nz> |
Cc: | pgsql-jdbc(at)postgresql(dot)org |
Subject: | Re: Printing query durations |
Date: | 2006-03-29 01:50:26 |
Message-ID: | C5733194-C792-421C-B887-EAFF4649EEA1@fastcrypt.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-jdbc |
Kevin,
Well, the server doesn't really know the difference between a query
from jdbc vs a query from psql. The only difference is how they
connect. psql usually connects to a unix domain socket, whereas jdbc
connects via a tcpip socket. Even this behaviour can be changed in
psql by specifying -h <hostname>
What is min_duration set to. Is it possible that the queries are
faster through JDBC. Possibly because the server already has done the
count?
Try setting min_duration to something ridiculously small.
Dave
On 28-Mar-06, at 8:41 PM, Kevin Dorne wrote:
> Simon Riggs wrote:
> [...]
>>> Example output from an interactive query:
>>> LOG: duration: 109.524 ms statement: SELECT count(*) FROM
>>> transaction;
>>
>> This is produced by log_min_duration_statement > -1
>> These lines always have duration prefixes.
>
> Yes, that's what I would expect. My problem is that this setting only
> logs queries via psql; queries via JDBC don't get logged at all.
>
>>> Example output from a JDBC query:
>>> LOG: statement: SELECT count(*) FROM transaction;
>>
>> This is produced by log_statement = 'all'
>> These lines never have durations.
>> If you want the matching durations, use log_duration = on and read
>> the
>> manual to see how to match them up.
>
> Yes, I've done that. Again, I can get those durations to appear for
> queries via psql, but not via JDBC.
>
> [...]
> By the way, the JDBC driver I'm using (thanks Markus Schaber for the
> tip) is "PostgreSQL 8.0 JDBC3 with SSL (build 311)".
>
> -k
>
From | Date | Subject | |
---|---|---|---|
Next Message | Oliver Jowett | 2006-03-29 01:57:03 | Re: Printing query durations |
Previous Message | Kevin Dorne | 2006-03-29 01:41:16 | Re: Printing query durations |