We have spent considerable time investigating the cause of the problems with Mbox3 and narrowed it down a single, problematic mailbox which was causing sockets to be left in a hung state. Removing this mailbox and restarting the Zimbra service fixed the problem, however we have not been able to conclusively prove whether this was the cause of the problem. We have been able to introduce additional monitoring to proactively alert us in the event of similar problems and before mail is delayed.

There is a related bug open on Zimbra's Bugzilla system, but it remains unconfirmed at present because it's not possible to consistently reproduce. We have tried to reproduce the problem on our test cluster and we also tried moving the problem mailbox there, but we have not been able to glean any further useful information.

Further background is that this problem seems to stem from IMAP connections and it is interesting to note that Zimbra 7 introduced a number of backend IMAP changes, including a whole new system for handling IMAP client connections: http://bugzilla.zimbra.com/show_bug.cgi?id=9470.

Again, please accept our apologies for the inconvenience caused.