Travis CI Python 3.3 builds stalled

18 views
Skip to first unread message

Oliver Beckstein

unread,
Jul 11, 2016, 12:52:02 PM7/11/16
to MDAnalysis developers list
Hi,

I had a quick look at some of the travis runs such as develop https://travis-ci.org/MDAnalysis/mdanalysis/builds/143875230 and all the PYTHON=3.3 builds stalled and died after 12 min or so.

Does anyone have any idea what’s going on? Should we just disable the builds?

In other notes: Any idea what to do about the open files problem? It seems to lead to a fair number of test runs that fail randomly. Typically, I just tell travis to re-run the test but that’s not ideal… and we’d like to fix the problem anyway.

Oliver

--
Oliver Beckstein * orbe...@gmx.net
skype: orbeckst * orbe...@gmail.com

Jonathan Barnoud

unread,
Jul 11, 2016, 12:58:28 PM7/11/16
to mdnalys...@googlegroups.com
Hi,

I do not know why the python3 build do not produce any output in that
time scale. I tried some time ago to make them verbose, but it did not
solve the issue.

On the open file issue, my tests seam to indicate that the issue is
solved in the issue363 branch.

Jonathan

Hai Nguyen

unread,
Jul 11, 2016, 1:42:32 PM7/11/16
to mdnalys...@googlegroups.com
minor point: should test with python 3.4 or 3.5.

I don't think peopole ever use 3.3

Hai

--
You received this message because you are subscribed to the Google Groups "MDnalysis-devel" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mdnalysis-dev...@googlegroups.com.
To post to this group, send email to mdnalys...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mdnalysis-devel/5783D02E.4020107%40barnoud.net.

For more options, visit https://groups.google.com/d/optout.

Jonathan Barnoud

unread,
Jul 11, 2016, 1:45:41 PM7/11/16
to mdnalys...@googlegroups.com
3.3 was the only version of python 3 officially supported by travis at the time we added the builds. Indeed, 3.4 would be better nowadays.

Oliver Beckstein

unread,
Jul 11, 2016, 1:48:11 PM7/11/16
to mdnalys...@googlegroups.com
Fair enough. I don't think that there's a specific reason that the tests use 3.3 apart from the fact that py3 is taking a backseat to #363.

--
Oliver Beckstein
email: orbe...@gmail.com

Oliver Beckstein

unread,
Jul 11, 2016, 1:49:21 PM7/11/16
to mdnalys...@googlegroups.com
I stand corrected.

Can we change to >3.3?

--
Oliver Beckstein
email: orbe...@gmail.com


Oliver Beckstein

unread,
Jul 11, 2016, 1:51:13 PM7/11/16
to mdnalys...@googlegroups.com
One more reason to get #363 done!

--
Oliver Beckstein
email: orbe...@gmail.com


Oliver Beckstein

unread,
Jul 20, 2016, 2:58:19 PM7/20/16
to MDnalysis-devel
Could someone familiar with Python 3 and Travis have a look at this?

If we can upgrade to Python 3.4 or 3.5, try it.

If it still stalls then I suggest to disable that part of the build matrix until #363 has arrived. It's wasteful to burn travis' cpu-h on tests that we know are not doing anything useful (and the processes count against our free running processes and can slow down the required tests.)

Oliver


On Monday, July 11, 2016 at 10:49:21 AM UTC-7, Oliver Beckstein wrote:
I stand corrected.

Can we change to >3.3?

--
Oliver Beckstein


> Am Jul 11, 2016 um 10:45 schrieb Jonathan Barnoud
>
Reply all
Reply to author
Forward
0 new messages