Logger misbehaving

[kaffemand]

Hi,

I've noticed a few times over the past week that my server becomes very slow. The slowdown is centered around /opt/zimbra/libexec/logswatch and the logger mysqld.

Logging into the mysqld, and doing a "show processlist;" reveals that it's busy inserting old log records in the raw_logs database..

mysql> show processlist;
+-------+--------+-----------+---------------+---------+------+--------+------------------------------------------------------------------------------------------------------+
| Id    | User   | Host      | db            | Command | Time | State  | Info                                                                                                 |
+-------+--------+-----------+---------------+---------+------+--------+------------------------------------------------------------------------------------------------------+
|     3 | zimbra | localhost | zimbra_logger | Query   |    0 | update | insert into raw_logs(log_date, loghost, app, pid, msg, postfix_qid) values ('2008-03-31 16:04:27','z | 
| 38588 | root   | localhost | NULL          | Query   |    1 | NULL   | show processlist                                                                                     | 
+-------+--------+-----------+---------------+---------+------+--------+------------------------------------------------------------------------------------------------------+

Further searching the raw_logs table reveals that for just this second, every log message is shown twice!

What's going on?

5.0.4 Open source on CENTOS 5.

[mmorse]

Reference: Bug 27336 - Bug in Perl File::Tail affects Zimbra