- SLES 10, GW 8.0.1 (upgraded from 8.0HP2 a week ago)
- Approx 500 clients using IMAP (Thunderbird), 50 clients using GW8 client, 50 clients using WebAccess

Since the installation of SP1 a week ago, the IMAP service on the POA randomly stops accepting connections. When the problem arises I can look at the status of the threads, and I can see that all threads are in use (for 200+ seconds and counting). Doing a '/etc/rc.d/grpwise status' shows that all services (PO and MTA) are running. When things are working normally, the IMAP threads process requests in under 3 seconds (usually under a second - extremely fast!).

I've noticed that no more than 8 IMAP threads are ever created. The default setting is to use as many as 40 threads.

I haven't been able to figure out two things in this situation:
1. What is happening to cause the IMAP service to take 200+ seconds to process a request? (Verbose logging hasn't been much help.) My theory is that if I wait long enough, the threads will release themselves, but I've witnessed one thread process a request for 900+ seconds. I'm not willing to let my clients wait that long so I can test my theory...:)

2. Why aren't more threads created?

I've also noticed that there was a change in IMAP behavior in SP1: Before SP1, an expunge process had to be run in Thunderbird to make "trashed" messages appear in the Trash folder. With SP1 that behavior has changed - messages now appear almost immediately in the Trash. (Could something else have changed to cause the IMAP problem I'm seeing?)

If you have any suggestions, please throw them out there!