|
From: Lionel B. <lio...@bo...> - 2005-06-24 11:43:23
|
David Rees wrote the following on 23.06.2005 23:52 :
>I just upgraded to 1.6.0 from 1.4.8 yesterday and have had a couple
>problems with it.
>
>I noticed this morning that sqlgrey appeared to have stopped
>responding, then found these errors in /var/log/maillog:
>
>Jun 22 16:45:21 forty sqlgrey: dbaccess: error: couldn't get reconnect
>delay: ERROR: prepared statement "dbdpg_17" does not exist
>
>
That is really puzzling. Which version of:
- perl,
- DBI,
- postgresql DBD driver,
are you using, are they all coming with your distribution ?
This is the first time ever this error gets reported. This kind of error
happens either if the prepare_cached() fails or the execute() fails. In
both cases this means that either:
- there was an error in the SQL syntax,
- the connection with the database is borked.
Either way, the code is designed to get around this (let the message
pass if we can't get a straight answer from the database) and reconnect
to the database hoping it is a temporary DB problem.
>Jun 22 16:45:21 forty sqlgrey: warning: Use of uninitialized value in
>concatenation (.) or string at /usr/sbin/sqlgrey line 1985.
>Jun 22 16:45:21 forty sqlgrey: dbaccess: warning: couldn't do query:
>INSERT INTO from_awl (sender_name, sender_domain, src, first_seen,
>last_seen) VALUES('tcg
>rant','alumni.princeton.edu','64.97.160.33','sql error',NOW()): ERROR:
> invalid input syntax for type timestamp: "sql error" , reconnecting
>to DB
>Jun 22 16:45:21 forty sqlgrey: warning: Use of uninitialized value in
>concatenation (.) or string at /usr/sbin/sqlgrey line 2008.
>
>
>
Given the first error, this is expected. I'll have to clean up a little
to avoid some of these errors though.
>Then it appears to have tried to restart itself later and I found
>these messages:
>
>Jun 23 02:58:53 forty sqlgrey: mail: failed to send: No child processes
>Jun 23 02:58:53 forty sqlgrey: dbaccess: error: couldn't get reconnect
>delay: ERROR: prepared statement "dbdpg_6" does not exist
>
>
Surprising again. AFAIK SQLgrey doesn't play with prepared statements in
a way that could explain this. If a prepared statement doesn't exist,
the call to prepare_cached() should automatically create it. So this
error is really unexpected.
>Jun 23 02:58:53 forty sqlgrey: warning: Use of uninitialized value in
>concatenation (.) or string at /usr/sbin/sqlgrey line 1985.
>
>I tried killing the sqlgrey process, but would only respond to a kill -9.
>
>
Even more surprising! I'm betting on a distro specific problem. A wild
guess: is SELinux active? What happens if you deactivate it?
>Any ideas? I'm running on Fedora Core 4.
>
>
Hum, are there any other FC4 users around?
Lionel.
|