did anyone else experienced the fact that this action just hangs? we have included an action that checks the status with the release-status REST call but had one situation where it just did hang for 15min untill we manually stopped it, although the action has the default timeout of 60s
If any of you had this situation, is there any way to prevent it or react to it during the deployment? Because I was only able to stop the step and so fail the deployment although everything worked fine and succeeded.
p.s. it only happens once so far, so I assume it might have been a network issue, still, would be good to know how to react to this stuff automatically
Sorry replying to your thread very late but wanted to check if you see any recurrence of this again. If so I will request to open a support ticket may be support can help to root cause if this is N/W issue or caused by something else.
no, it didn't happen again yet, so I believe my assumption was correct, that it most likely was a network issue.
I marked this question as "assumed answered"
Thanks for updating the thread, Michael. This is an interesting situation. It may translate more into an idea. I don't know of any way that you could reliably "handle" this without a traffic cop (so to speak). Keeping an eye on the "Long Running" jobs widget might help you identify which jobs are experiencing this condition. I'm not sure what criteria it checks for a job to appear through this widget. Maybe it is our version of a traffic cop. But I would think that you might want something that is looking at jobs whose step hasn't progressed in X (preferably a configurable value) amount of time.
Does anyone know the criteria used for adding jobs to the Long Running widget?
you thing one need also to keep in mind, that it mustn't really be an error that happens and the action is somehow stuck in an infinite state. it also might be, that the action really does take quite some time (e.g. downloads or copy of large files or a huge amount of files)
so yeah...it is a tricky situation, maybe it might need something like a special flag on all actions, where you can define "monitor long running" and the threshold is set somewhere. I know that you guys already have internal timeouts as well, but it seems they don't fire in all situations (which is good)