I've never seen this one before.
It appears that when amavis fails an ldap search, it retries, first disconnecting and reconnecting to the ldap server (from the logs you posted, this reconnect is succeeding).
Some questions:
Are all the spams going to the same address? Are all the messages the same?
What does df -h show? Both before and after the crash.
Here's what I would suggest:
edit /opt/zimbra/amavisd.conf.in and change the log_level line to "5".
zmamavisdctl stop
zmamavisdctl start
Then recreate the problem, and post the (extremely verbose) logs from the session. After the crash:
zmcontrol status - that should reveal what all is down.
If that doesn't reveal the cause, we'll work on cranking up the ldap log level. |