IMAP Timeout Errors From Messaging Server Clients To Backend Imap Servers (Doc ID 1275280.1)

Symptoms

I am seeing "General Error: Error reading from IMAP server: Timeout" in the http log on a front-end messaging express server. I have had at least one user so far report that they get a lot of errors usually when sending or forwarding mail, the error they report was "Timed out while waiting for server response" or "copy was not saved, unable to send message".

I also recall getting a call last week from a Thunderbird mail client user reporting a timeout, and occasional reports of similar timeouts in the past few weeks but the users usually tried again and it worked for them so I didn't look into it. It doesn't appear that the IMAP proxy is configured to log anything about IMAP transactions.

When the timeouts are happening, I check the "iostat -x 5" output on the backend stores and it seems OK (largest times I saw was 250ms once but usually maxing out around 30 and normally under 10.) Storage is 2x1TB ZFS LUNs (1 per store) on a Sun StorageTek 2540 and all appears in good health.

imap.maxsessions is set to 6000 on both backend stores and the load balancer in front of these machines only reports about 1600-1700 current open sessions.

I've enabled IMAP telemetry for the user who reported problems to see if anything shows up there, and right now am thinking I will restart the stores tonight to see if that helps at all.

The stores are running a current version of the Messaging Server, the frontend is running an older 6.3 version but it doesn't appear that the issues we are seeing are related to the frontend.

I got an IMAP alarm to the mail_administrator address today as well, that IMAP response time was too long. The text is below:

Cause