Note on scalix 11.4.4 to 11.4.5 Upgrade

[propagandhi]

Just in case others experience this, one of the 13 nodes I upgraded last night (the central node in fact) experienced a long delay at around the section where it installs/upgrades the management services.

I noted that queue.manager process was utilising 100% cpu and the system was totally bogged down.

Initially i assumed that the queue manager process was hung when i stopped the services to do the upgrade, so I performed a kill -9 on it (silly me i know).

After the installation then succeeding, every restart of the scalix services resulted in the queue.manager process still utilising 100% cpu, and no services starting.

I then ran the installer again and performed the upgrade of all components. This time I just waited for the queue manager to complete, and everything was fine.

So, just a word of warning to anyone else who sees this, the install may take a while and the queue manager might go nuts, but just leave it alone until it completes.

[smpoole7]

Nice to know. Thanks.

For curiosity's sake, what kind of system are you running Scalix on?

[propagandhi]

Our servers are all virtualized on IBM hardware.

[dirk]

I think you should have a deeper look into your queue status. Probably best with omqdump.
If there is a queue with lots of messages (including the error queues) the behaviour is expected and nothing new in 11.4.5. The queue manager is just upgrading all messages.
The content of ˜/xport.hold (to be resubmitted Scalix-Scalix messages) might be worth a look too.