| Re: [cdt-dev] Frequent cdt-verify job aborts |
I keep an eye on the stuck builds on the deadlock right now. Some do pass after > 2h.
I've set the timeout to 3h and also opened this bug:
https://bugs.eclipse.org/bugs/
show_bug.cgi?id=511989
I'm also investigating the deadlock. I'm monitoring the builds to abort them when I'm fairly sure they are stuck there, but it's not the majority.
I'm also wondering if a recent code change could be making the tests go slower, in cdt.core in particular.
I'll keep you posted.
Marc-André
From: cdt-dev-bounces@xxxxxxxxxxx <cdt-dev-bounces@xxxxxxxxxxx> on behalf of Marc Dumais <marc.dumais@xxxxxxxxxxxx>
Sent: Tuesday, February 7, 2017 8:57:07 AM
To: CDT Mailing List
Subject: Re: [cdt-dev] Frequent cdt-verify job abortsBTW, we do have the option of setting a longer timeout for our builds (ATM: 150 mins). It's of course not ideal to have builds run for very long. OTOH, when we invest 2h30 in a build that ends-up timing-out, it's wasted resources.
From: cdt-dev-bounces@xxxxxxxxxxx <cdt-dev-bounces@xxxxxxxxxxx> on behalf of Marc Dumais <marc.dumais@xxxxxxxxxxxx>
Sent: Tuesday, February 7, 2017 8:36 AM
To: CDT Mailing List
Subject: Re: [cdt-dev] Frequent cdt-verify job abortsHi Nathan,
There was a discussion recently on the Eclipse cross-project-issues-dev mailing list about other projects seeing their builds/tests take multiple times longer than they used-to. Look-for title "hipp9 overloaded".
So it may be that this is a general problem, or at least affecting multiple instances of "hipp". It might help us to run only 2 jobs in parallel, but then again it depends how the resources are divided between projects; other projects sharing the same resources (VM pool?) might need to do the same before we see benefits.
Regards,
Marc
From: cdt-dev-bounces@xxxxxxxxxxx <cdt-dev-bounces@xxxxxxxxxxx> on behalf of Nathan Ridge <zeratul976@xxxxxxxxxxx>
Sent: Friday, February 3, 2017 3:37 PM
To: CDT Mailing List
Subject: [cdt-dev] Frequent cdt-verify job abortsLately I've noticed our cdt-verify jobs are being aborted very frequently, due to a timeout. In such cases, the following appears at the end of the build output:
Build timed out (after 150 minutes). Marking the build as aborted.
Does anyone know the cause of these timeouts? Are we just expecting the tests to run more quickly than they actually do?
I also noticed that the aborts happen more frequently when there are more jobs being done in parallel. Are we perhaps allowing more jobs to run in parallel than the machine can handle?
Thanks,
Nate
_______________________________________________
cdt-dev mailing list
cdt-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cdt-dev
_______________________________________________
cdt-dev mailing list
cdt-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cdt-dev