Best practice for zimbra: restart required?

[mfehr]

I analyzed the content of /opt/zimbra and detected some old data and a large redo.log file of 50 MB. After restarting zimbra, the redo log was reduced to 512 KB. - Until now, I was under the assumption to have zimbra running 7x24, only stopping for upgrade or server restart purpose. This means that zimbra might be running for 2 months. Does this make sense?

What is best practice for a recurring restart and periodic cleanup (e.g. remove old tmp files in /opt/amavisd/tmp, etc.) of a zimbra server?

[p24t]

Generally I only restart my ZCS when it needs it. (occasionally java goes wonky for me) Or, of course for upgrades.

The redolog is an internal function, more like a double-check on what the server has done. Mine seems to peak at about 100-120 megs and cycle onto a new file, and it keeps a few in logrotation.

I don't know that they're used in the OSS version, but I could be wrong. Here's a good explanation:

Question about redo logs and disaster recovery