Bug: Please add a timeout to omero admin operations

[rleigh]

"bin/omero admin start|stop|restart" take much longer than 5 minutes to complete on some systems. This results in the annoying situation where the startup will be cancelled part way through starting up, making it impossible to start/restart the server.

It would be nice if the timeout could be configurable.

It would also be good to know what on earth omero is doing that takes such huge amounts of time to bring up to readiness. Could much of this be deferred or started lazily after initial start?

[jmoore]

NB: Roger, try not to create tickets without owners or CC if possible.

• Why is it impossible to start/restart the server? Have you tried "startasync"? Have you Ctrl-C'd before the shutdown progresses?
• Can you attach a Blitz-0.log of such a long startup so we can see what's taking so long?

[rleigh]

startasync certainly does work partially. However, the diagnostics shows Blitz-0/Indexer-0/PixelData-0/Processor-0/Tables-0 as "activation timed out (pid=nnnn, enabled). Appears to work, though I am seeing some exceptions in the logs.

Does the upgrade check get run at start or only for diagnostics? The VM does not have internet access; not something we should be relying on if that's a potential cause of delays. I'll investigate this and extract the logs tomorrow.

[jmoore]

upgradechecked definitely gets run during startup. The upgradecheck does have a timeout set (10 seconds: ​https://github.com/openmicroscopy/openmicroscopy/blob/develop/components/common/src/ome/system/UpgradeCheck.java#L38) but it may be misbehaving.

[jmoore]

PR opened ​https://github.com/openmicroscopy/openmicroscopy/pull/826

[jmoore <josh@…>]

(In [9e79baa120774090e40bbdd26e2b3dac474f5cea/ome.git] on branch develop) Add timeout to waiting bin/omero admin commands (Fix #10361)

All commands which use waitup and waitdown internally now
take a --wait parameter (in seconds).