Because!
Would need a fair bit of testing, I suppose. We are already running a fairly 'mixed' environment (precise and trusty exec hosts, with a precise master / shadow)
Description
Description
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | Andrew | T143349 Deprecate precise instances in Labs by 2017-03-31 | |||
| Resolved | yuvipanda | T94790 Phase out precise instances from Tool Labs | |||
| Resolved | coren | T94791 Move tools-master and tools-shadow to trusty | |||
| Resolved | Andrew | T103390 Labs: Move tools-shadow off the same host as tool-master |
Event Timeline
Comment Actions
Thankfully, that shouldn't be /too/ hard thanks to the master/shadow setup. Reinstalling shadow as Trusty, then forcibly switching over to test seems like the simplest approach.
Comment Actions
Doing simultaneously with the parent ticket (that is, the new tools-shadow-01 is constructed with Trusty from the start)
Comment Actions
tools-grid-shadow is now on Trusty and happily chugging along as a shadow master. Once it's tested as being able to take over as master, I'll create a new master node in Trusty too.
Comment Actions
tools-grid-shadow is now the active master, and seems to be running without issues. I'm going to give it a little while then create the new (trusty) master.
Comment Actions
tools-grid-master is now the active master and tools-grid-shadow has resumed its role as shadow master.
There is an apparent lingering issue (almost certainly Trusty-based) that the service start invocation for gridengine-shadow hangs (even though the daemon itself works fine). It looks like the systemd unit expects a fork() that never comes.
Comment Actions
The original masters (tools-master and tools-shadow) have been turned off but not yet deleted to monitor how things go over the weekend.