Wednesday, 2020-11-04

*** lamt has quit IRC00:03
*** hamalq has quit IRC01:38
*** DSpider has quit IRC01:39
openstackgerritIan Wienand proposed opendev/system-config master: borg-backup: disambiguate for multiple servers
*** stevebaker has quit IRC02:13
openstackgerritIan Wienand proposed zuul/zuul-jobs master: merge-output-to-logs: convert to untrusted executor safe code
openstackgerritMerged zuul/zuul-jobs master: More E208 fixes
openstackgerritMerged opendev/system-config master: letsencrypt-install-txt-record: pause after adding TXT records
openstackgerritMerged opendev/system-config master: RAX DFW/IAD : add internal mirror DNS to cert
openstackgerritMerged opendev/system-config master: borg-backup: disambiguate for multiple servers
*** ykarel|away has joined #opendev04:44
*** ykarel|away is now known as ykarel04:47
*** lamt has joined #opendev04:54
*** kindwindfall__ has quit IRC05:38
*** ykarel has quit IRC05:49
*** ykarel has joined #opendev05:54
openstackgerritIan Wienand proposed opendev/system-config master: [wip] borg: add fuse
*** ysandeep|away is now known as ysandeep|ruck06:05
*** marios has joined #opendev06:12
*** sboyron has joined #opendev06:55
*** sboyron has quit IRC06:56
*** sboyron has joined #opendev06:57
*** ykarel_ has joined #opendev06:59
*** ykarel has quit IRC07:02
*** fressi has joined #opendev07:08
*** whoami-rajat__ has joined #opendev07:16
*** ykarel_ has quit IRC07:18
*** eolivare has joined #opendev07:41
*** ralonsoh has joined #opendev07:54
openstackgerritzbr proposed zuul/zuul-jobs master: Enable progressive mode with ansible-lint
*** slaweq has joined #opendev08:03
*** sshnaidm|afk is now known as sshnaidm|rover08:08
*** andrewbonney has joined #opendev08:09
openstackgerritzbr proposed zuul/zuul-jobs master: More E208
openstackgerritzbr proposed zuul/zuul-jobs master: More E208 (22)
*** fressi has quit IRC08:36
*** tosky has joined #opendev08:42
*** ykarel has joined #opendev08:48
*** hashar has joined #opendev08:54
openstackgerritzbr proposed zuul/zuul-jobs master: More E208 (final)
openstackgerritIan Wienand proposed opendev/system-config master: [wip] borg: add fuse
*** rpittau|afk is now known as rpittau09:05
*** fressi has joined #opendev09:14
*** ralonsoh has quit IRC09:16
*** ralonsoh has joined #opendev09:19
*** larainema has joined #opendev09:23
*** DSpider has joined #opendev09:37
*** fressi has quit IRC09:59
*** sauloasilva1 has quit IRC10:01
*** hashar has quit IRC10:10
*** fressi has joined #opendev10:21
*** marios has quit IRC10:38
*** lpetrut has joined #opendev10:55
*** sboyron has quit IRC11:14
*** sboyron has joined #opendev11:14
*** sboyron has quit IRC11:25
*** sboyron has joined #opendev11:25
*** marios has joined #opendev11:36
*** sboyron has quit IRC12:15
*** sboyron has joined #opendev12:15
*** sboyron has quit IRC12:30
*** sboyron has joined #opendev12:30
*** hashar has joined #opendev12:34
*** sboyron has quit IRC12:35
*** sboyron has joined #opendev12:35
*** sboyron has quit IRC12:45
*** sboyron has joined #opendev12:47
*** sboyron has quit IRC12:51
*** sboyron has joined #opendev12:51
*** sboyron has quit IRC13:00
*** sboyron has joined #opendev13:00
*** sboyron has quit IRC13:11
*** sboyron has joined #opendev13:11
*** ralonsoh has left #opendev14:07
zbrfungi: clarkb mnaser: hi! can we help pip folks by providing extra testing for pip? (use of zuul as 3rd party ci)14:10
zbrThis is related to thread.14:10
sean-k-mooneyi think that would require the maintainer of pi to install the zuul appliction for the opendev or openstack tenants to the pypa/pip repo so that the existing github connect we have could trigger off the PR agains pip then we would need to add the repo to either the openstack or preferable opendev tenant14:14
sean-k-mooneywhat type of job were you thinking woudl be run14:14
*** ysandeep|ruck is now known as ysandeep|away14:15
sean-k-mooneythe pip tests or deploying openstack with the modifed pip or running tox install/unit/funcitonal tests for some openstack projects?14:15
zbri can work with them but for the moment I consider using two nodes, likely a fedora and a debian one.14:16
zbrstill, i need some kind of informal acceptance before proposing a CR14:17
sean-k-mooneywell i assume that would be 2 different jobs14:18
sean-k-mooneyor the same base job but with 2 differnt inerited jobs for fedroa and debian althouh i proably would have chosen centos/ubuntu but whatever works for them14:19
sean-k-mooneyfor pip i defnietly see the advantage of chasing faster updated distro to test the latest python14:20
zbryeah, centos fine too. as long we have two different os, should be fine for start.14:21
*** xavpaice has quit IRC14:48
*** xavpaice has joined #opendev14:50
*** fressi has quit IRC14:58
fungizbr: i suppose it would be similar to what we're doing to help provide third-party arm64 testing and wheel builds for pyca/cryptography14:58
fungii haven't read through the issue yet though, will take a look shortly14:58
zbri was looking to follow cryptography as a model14:59
zbrthe issue is very long but last comments are important: I offered to maintain few zuul jobs that test extra platforms for them14:59
fungialso saw you mention resolver loops over in #openstack-infra, note that prometheanfire has at least tried installing all of openstack/requirements global-requirements.txt with the new resolver and not run into problems15:00
zbrpypa team explained lack of hw/human resources for better testing pip.15:00
fungibut that's also a fairly broad test, no doubt the issues crop up in the corners that hasn't tested15:01
zbri know, i am already using new resolver on few projects, but those with more complex extra, can get into loop bugs.15:01
zbrespecially if you have a dependency that is listing you+extra, it goes infinite.15:01
zbri disabled it for these special cases.15:01
zbrin general is better than old one, at least you do not endup with conflicts15:01
zbrbefore the 2020-resolver I had a habit of adding a "pip check" as first command in tox.ini15:02
fungiright, or worse, incorrect versions of dependencies15:02
*** xavpaice has quit IRC15:10
fungisome of the comments mention which sounds a bit like zuul's project gating scheduler15:11
fungihad never seen that before15:12
*** xavpaice has joined #opendev15:12
openstackgerritzbr proposed openstack/project-config master: Reconfigure testinfra new location
zbrevery time i see the name of "bors" project I have a big laugh, in easter europe this is used to make a sour soup.15:17
openstackgerritzbr proposed openstack/project-config master: Remove testinfra project
fungiso i guess it's not entirely clear to me what third-party testing we'd be performing for pypa/pip pull requests... what i sort of pieced together is they don't want to be responsible for running their functional testsuite on debian, fedora and centos, but would be amenable to someone else possibly doing it and reporting as a third-party ci system15:33
*** ykarel has quit IRC15:41
*** lpetrut has quit IRC15:42
sean-k-mooneywhat was the zuul based travic alternitive15:46
sean-k-mooneyboni ci or something15:46
fungibonnie-ci, yeah. SpamapS was running that but i think he dropped it a year or more ago15:47
fungiand yeah, if that's the suggestion then i don't see any problem with it (for the same reason the pyca tenant is also within our remit), but i would be cautious about pushing this on them... people generally don't get excited about $tool they've never used before unless it solves actual problems they have which their current tools cannot. and it doesn't sound like the pip maintainers are convinced they need15:47
fungito test on debian/fedora/centos15:47
fungiso they're just as likely to feel they're being pressured/marketed at and will be instinctively skeptical of whatever alternative solution is suggested if it doesn't originate from within their inner circle15:49
fungidoing arm64 builds for pyca/cryptography mostly worked out because they saw a need for arm64 wheels and were unable to make them happen with the solutions already available to them15:50
fungi(at least in a timely enough fashion to satisfy other requirements they had)15:51
fungithey were still skeptical at first, and that was even with the maintainers having been former openstack contributors so already having some familiarity with earlier versions of zuul15:52
clarkbI'm not opposed but considering my last major interaction with the project was "your code is so bad I won't even review it" I'm not really personally inclined to get involved much15:53
sean-k-mooneyjust as a general comment it might be nice to update or add a new doc for how to create a new ci tenant for external projects15:56
sean-k-mooneye.g. what are the consideration to review when evaluating if opendev is the right place to do testing15:57
sean-k-mooneywhat are teh critia for new projects, how to get in contact  and set up the ci ecetra15:57
fungiwe still need to hack apart the infra manual, that project creators guide was initially written by dhellmann as a checklist for adding new oslo libs, so it's somewhat openstackish15:58
sean-k-mooneysome of that is in the openstack project creators guide but not all of it15:58
fungiwe probably need to move more of the openstackisms into some openstack-specific reference like their project teams guide or contributor guide15:59
sean-k-mooneyto be fair its in the faq15:59
sean-k-mooneyor rather called out in the faq
sean-k-mooneythat this is still a do do to decuple16:00
fungialso in this case they woudn't be doing any code hosting in opendev probably, other than a very tiny tenant config repo16:02
*** elod has quit IRC16:26
*** elod has joined #opendev16:26
*** hashar has quit IRC16:28
*** hashar has joined #opendev16:28
zbrtrue, in fact doubt they need a separated tenant16:48
fungiwell, i think when we did pyca/cryptography we opted to use a separate tenant for good reasons, and talked about doing that for any similar situations in the future16:50
*** marios is now known as marios|out16:59
*** marios|out has quit IRC17:06
*** eolivare has quit IRC17:12
*** tosky has quit IRC17:15
*** sboyron has quit IRC17:18
*** sboyron has joined #opendev17:18
*** rpittau is now known as rpittau|afk17:35
*** hamalq has joined #opendev17:46
*** mlavalle has quit IRC18:08
fungi#status log deleted broken 7t0pduk9xwiy spreadsheet from ethercalc via rest api, since it was malformed in such a way that it was crashing the service on page load18:12
openstackstatusfungi: finished logging18:12
*** mugsie has quit IRC18:18
*** mugsie has joined #opendev18:21
*** andrewbonney has quit IRC18:24
*** slittle1 has left #opendev18:38
*** mlavalle has joined #opendev18:50
*** mlavalle has quit IRC18:59
*** mlavalle has joined #opendev19:19
*** _mlavalle_1 has joined #opendev19:22
*** stevebaker has joined #opendev19:23
*** mlavalle has quit IRC19:24
*** _mlavalle_2 has joined #opendev19:25
*** _mlavalle_1 has quit IRC19:27
*** slaweq has quit IRC20:32
*** tosky has joined #opendev20:35
*** rchurch has quit IRC20:46
*** rchurch has joined #opendev20:49
ianwsean-k-mooney: yes, this actually came up in the ptg multiarch discussions where i took an action item to describe better how to hook up things from the pyca experience20:50
ianwi hope to get that written up soon, and we can shuffle the content as appropriate i'd say20:50
ianwi think that given the hours we have all lost to pip updates, it would be useful for us to get involved21:02
sean-k-mooneyya that would be useful for third part cis too. i have deployed zuul for third party ci 4 times now and while i vaguly know where to look in the docs i still find them kind of hard to navigate21:03
sean-k-mooneythe info is there21:03
sean-k-mooneybut the fact there is no config ref like we have for other projects trips me up every time21:04
sean-k-mooneyoh tiny feature request put the links to nodepool/other docs in the zuul docs21:04
sean-k-mooneye.g. in the header some where havign the links to the different docs listed here
sean-k-mooneywoudl be awsome21:05
fungisomething like the docs pull-down on the main pages?21:06
fungiprobably a better suggestion for #zuul since the project is in conrol of its site/docs/sphinx theme21:07
sean-k-mooneykind of but ya i kept havign to sawp betteen and to find the  config options and how to write .zuul.yaml then havign to job to nodepool docs21:08
sean-k-mooneybut ya its a better suggestion for #zuul21:10
sean-k-mooneythe user admin split is just kind of unhelpful when your trin to get things intially deployed becaue you are using both initailly21:12
sean-k-mooneyonce its deployed it make more sense but grouping by zuul.conf/tenant.yaml vs .zuul.yaml woudl be simpler IMO21:12
*** sboyron has quit IRC21:24
openstackgerritIan Wienand proposed openstack/project-config master: Add pypa/project-config
openstackgerritIan Wienand proposed openstack/project-config master: Add pypa tenant
*** hashar has quit IRC22:01
*** whoami-rajat__ has quit IRC22:13
ianw2020-11-04 22:09:35.684296 | ubuntu-bionic |     from ansible.plugins.loader import add_all_plugin_dirs22:51
ianw2020-11-04 22:09:35.684316 | ubuntu-bionic | ImportError: cannot import name 'add_all_plugin_dirs'22:51
ianwi'm guessing ansible-lint is broken?22:51
fungiis this the "you can't upgrade ansible in-place" bug still?22:56
ianwi'm ... not quite sure23:05
ianw ... investigating23:06
ianwohhh, it's "ansible.plugins.loader import add_all_plugin_dirs"23:06
ianwi think it's not compatible with the version of ansible we're installing in the tox environment23:06
ianwhrm, ansible-lint requires ansible >= 2.8 in setup.cfg, but we pin it to <2.8 for tox23:08
fungiyeah, i think that's the problem zbr was talking about. there may already be a job change up to work around that, not positive though23:28
openstackgerritIan Wienand proposed openstack/project-config master: Add pypa/project-config
openstackgerritIan Wienand proposed openstack/project-config master: Add pypa tenant
openstackgerritIan Wienand proposed openstack/project-config master: tox.ini : update Ansible pin
ianwit's not 100% clear we need the pin.  but having a version of ansible in the linter that doesn't run in production is probably bad23:44
clarkbianw: I think zuul is using 2.9 by default23:54
ianwclarkb: yeah23:54
clarkbbut I guess we support 2.8 so linting against that ensures we don't break jobs running with 2.8?23:54
ianwi'm not sure how much the ansible version influences the linting?  perhaps you're right we should keep it the lowest possible version23:55
clarkbI think ansible-lint uses ansible's yaml parser23:55
clarkbso there is some impact23:55

Generated by 2.17.2 by Marius Gedminas - find it at!