Queues missing after upgrade from RT 4.2.12 to RT 4.4.4


I’m trying to upgrade from RT 4.2.12 + RT::IR 3.2.1 to RT 4.4.4 + RT::IR 4.0.1 as a first step in
migrating our ticket database towards RT 5.0.x. Following the upgrade instructions however, it appears
all the previous queues have disappeared during the upgrade to RT 4.4.4. There are no queues
at all visible in the web gui of RT 4.4.4 when checked using the RT admin account. The dropdown
menu for creating new tickets on the start page in the web gui (“New ticket in” button) presents one queue in the queue selection box called “#4” but even that is a queue that doesn’t exist.

Any idea as to what goes wrong in the upgrade process?

Further inspection of the Queues table in the mysql database indicates the old queues are there, so it
seems the problem is related to how RT+RT::IR tries to fetch the queue info. Still, ideas for fixing this are more than welcome.

And two more observations: I have REST API 2.0 installed to RT 4.4.4 and searching for a ticket
with the URL


returns the expected ticket info including its queue data, whereas


reports the number of active queues correctly but fails to list them.

Did you clear the mason cache and restart the webserver? Maybe some kind of weird caching?

I did and afaik there’s no other caching involved. After having compared the Queues table structure from 4.2.12 with the one in 4.4.4 I noticed the upgrade from the former to the latter is unable to create
the same table structure as a fresh install of 4.4.4 does. Why is still a mystery …

If you go to Admin->Tools->system configuration and find the upgrades section, does it say any upgrade steps failed?

I have an extremely long Makefile that aims to both install and upgrade RT+RTIR. It seems one of the steps in the upgrade procedure was skipped due to variables set incorrectly, so it looks better after fixing them, the queues appear. Sorry for the trouble.

However, now it seems I have multiple queues in my RT 4.4.4 + RT::IR 4.0.1, e.g. there is one Incident queue and then there are queues named as “Incident - <constituency_name_here>”, one for each constituency. Is this the way it should be?