pantsbuild #announce

we typically would cut releases directly from master at whatever floating sha is at HEAD vs backdating to the release sha - but sounds like you guys actually cut from/respect the release tags

rough-minister-58256

09/08/2017, 11:23 PM

but yeah, hopefully a fix lands early next week

rough-minister-58256

09/08/2017, 11:23 PM

and then if anyone was relying on a dev release etc it’d be included for next fri

limited-country-20626

09/08/2017, 11:23 PM

yep…I basically cut out the practice of forking Pants, I feel like the habit cost us more than it gained.

limited-country-20626

09/08/2017, 11:23 PM

yeah, =you. Sounds good

limited-country-20626

09/08/2017, 11:24 PM

I am gonna go grab a train, will release based on feedback when I get to my house

rough-minister-58256

09/08/2017, 11:25 PM

cya

limited-country-20626

09/09/2017, 1:06 AM

wow…multiple shards timed out. Will click some buttons

rough-minister-58256

09/09/2017, 1:07 AM

integration test shard 1 has been timing out for a couple days - tracked in https://github.com/pantsbuild/pants/issues/4850

rough-minister-58256

09/09/2017, 1:07 AM

we’ve pinged Travis asking for a timeout bump and Dorothy is looking into topical sharding fixes

limited-country-20626

09/09/2017, 1:08 AM

okay, I see the relevant PR now

limited-country-20626

09/09/2017, 1:08 AM

sorry, I am a bit out of touch with the CI issues

limited-country-20626

09/09/2017, 1:10 AM

Okay, so it looks like for the most part we are crossing our fingers and landing code. I already generated the markdown locally, so I know it compiles. I can go forward, I think

rough-minister-58256

09/09/2017, 1:11 AM

for the last one I merged, I just copied the ci.sh invocation from shard 1 and ran it locally to confirm

rough-minister-58256

09/09/2017, 1:12 AM

but yeah, possible that timeout is cutting off e.g. failing tests

rough-minister-58256

09/09/2017, 1:12 AM

all week travis seems to either be super slow or having major network issues on fetches

rough-minister-58256

09/09/2017, 1:12 AM

almost wonder if we’re just being opaquely resource constrained/oversubscribed by travis

rough-minister-58256

09/09/2017, 1:13 AM

c’est la vie

limited-country-20626

09/09/2017, 1:16 AM

We probably would find out about broken tests, and I happened to run all the unit tests yesterday as I worked on a patch for upstream

limited-country-20626

09/09/2017, 1:16 AM

so hopefully generally green. But yeah, a real red flag for TravisCI (and/or zinc 🧌 )

limited-country-20626

09/09/2017, 1:40 AM

Hmm…so the deprecation is blocking the tests from passing during the dry run

limited-country-20626

09/09/2017, 1:41 AM

Copy code

.7/site-packages/urllib3/contrib/pyopenssl.py:46: DeprecationWarning: OpenSSL.rand is deprecated - you should use os.urandom instead
  import OpenSSL.SSL
**** Failed to install cryptography-2.0.3 (caused by: NonZeroExit("received exit code 1 during execution of `['/private/var/folders/yh/7xl9dstx33vc_xwqtbzpxy2h0000gn/T/pants.XXXXX.EDz00HVk/bin/python2.7', '-', 'bdist_egg', '--dist-dir=/var/folders/yh/7xl9dstx33vc_xwqtbzpxy2h0000gn/T/tmplv70zO']` while trying to execute `['/private/var/folders/yh/7xl9dstx33vc_xwqtbzpxy2h0000gn/T/pants.XXXXX.EDz00HVk/bin/python2.7', '-', 'bdist_egg', '--dist-dir=/var/folders/yh/7xl9dstx33vc_xwqtbzpxy2h0000gn/T/tmplv70zO']`",)

limited-country-20626

09/09/2017, 1:49 AM

gonna try and hack through here, possibly I just need to adjust the environment

limited-country-20626

09/09/2017, 3:58 AM

I only tried this for a few moments. But my gut-level interpretation is that the chroot is exiting non-zero trying to translate this library. We could either temporarily remove the requests[security] lib or accept this as a temporary blocker until the library releases. https://github.com/requests/requests/blob/master/setup.py#L72 I am done for the evening, but I can sit down for a couple hours and release from an alternate SHA, or after consuming a patch fix of either stripe.

limited-country-20626

09/09/2017, 8:50 PM

I am going to see about using vanilla requests. FWICT, the

requests[security]

was intended for CI consumption, but since the CI is broken for other reasons, perhaps we can just not gate on that fix here

rough-minister-58256

09/11/2017, 5:22 PM

hmm, interesting that it passes unit test shards without that.. wonder if there’s some caching taking place?