| opendevreview | Merged openstack/devstack stable/ussuri: nova: Ensure each compute uses a unique iSCSI initiator https://review.opendev.org/c/openstack/devstack/+/812929 | 00:21 |
|---|---|---|
| opendevreview | Merged openstack/devstack stable/victoria: nova: Ensure each compute uses a unique iSCSI initiator https://review.opendev.org/c/openstack/devstack/+/812928 | 00:48 |
| opendevreview | Merged openstack/devstack stable/wallaby: nova: Ensure each compute uses a unique iSCSI initiator https://review.opendev.org/c/openstack/devstack/+/812926 | 00:51 |
| opendevreview | Merged openstack/devstack master: Rehome functions to enable Neutron's Trunk service plugin https://review.opendev.org/c/openstack/devstack/+/812657 | 00:51 |
| opendevreview | Merged openstack/devstack master: Fix displaying usage for make_cert.sh https://review.opendev.org/c/openstack/devstack/+/803731 | 00:53 |
| *** flaviof_ is now known as flaviof | 01:26 | |
| opendevreview | Merged openstack/devstack stable/train: nova: Ensure each compute uses a unique iSCSI initiator https://review.opendev.org/c/openstack/devstack/+/812930 | 02:10 |
| *** bhagyashris is now known as bhagyashris|rover | 04:43 | |
| *** abhishekk is now known as akekane|home | 04:50 | |
| *** akekane|home is now known as abhishekk | 04:50 | |
| *** pojadhav- is now known as pojadhav | 05:44 | |
| *** jpena|off is now known as jpena | 07:32 | |
| opendevreview | Soniya Murlidhar Vyas proposed openstack/patrole master: Add tests for log resource https://review.opendev.org/c/openstack/patrole/+/795548 | 07:48 |
| soniya29 | slaweq, do we require show_log policy in the logging extension of neutron - https://opendev.org/openstack/neutron/src/branch/master/neutron/conf/policies/logging.py ? | 08:00 |
| *** sboyron_ is now known as sboyron | 08:00 | |
| *** sboyron is now known as Guest2498 | 08:01 | |
| soniya29 | slaweq, I mean, Is it missing there or isn't required? | 08:02 |
| opendevreview | Merged openstack/devstack master: tools: Fix use of continue https://review.opendev.org/c/openstack/devstack/+/812507 | 08:10 |
| slaweq | soniya29: I think we are missing it | 08:13 |
| soniya29 | slaweq, okay, so shall i add it? | 08:14 |
| slaweq | soniya29: would be great, thx :) | 08:14 |
| soniya29 | slaweq, thanks for the info :) | 08:15 |
| opendevreview | Merged openstack/devstack master: Remove cinder from service names https://review.opendev.org/c/openstack/devstack/+/811310 | 08:27 |
| isabek | Hi team! How we can run tempest tests not in parallel? During running tempest api tests creates resources in the same time... | 08:35 |
| opendevreview | Merged openstack/devstack master: Further fixup for Ubuntu cloud images https://review.opendev.org/c/openstack/devstack/+/813148 | 09:26 |
| *** sboyron_ is now known as sboyron | 09:26 | |
| *** sboyron is now known as Guest2502 | 09:27 | |
| *** Guest2502 is now known as sboyron_ | 09:29 | |
| opendevreview | Soniya Murlidhar Vyas proposed openstack/patrole master: Add tests for log resource https://review.opendev.org/c/openstack/patrole/+/795548 | 09:43 |
| opendevreview | Soniya Murlidhar Vyas proposed openstack/patrole master: Add tests for loggable resource https://review.opendev.org/c/openstack/patrole/+/807131 | 10:08 |
| soniya29 | slaweq, By observing the other policies of neutron, there is no implementations of show_network/show_flavor policy. Hence IMHO, we won't require show_log policy as well, what do you think? | 10:14 |
| slaweq | soniya29: sorry, it was to early for me I guess | 10:30 |
| slaweq | :) | 10:30 |
| slaweq | It's of course defined in https://opendev.org/openstack/neutron/src/branch/master/neutron/conf/policies/logging.py#L74 | 10:30 |
| slaweq | there is GET for collection (list) and for resource (show) | 10:30 |
| *** dviroel|afk is now known as dviroel | 11:18 | |
| soniya29 | slaweq, yes...got it!! | 11:19 |
| soniya29 | slaweq, thanks :) | 11:19 |
| *** jpena is now known as jpena|lunch | 11:22 | |
| *** jpena|lunch is now known as jpena | 12:16 | |
| sean-k-mooney | o/ have we started to see any time out with devstack installing packages form pip | 12:23 |
| sean-k-mooney | locally on a newlew providioned ubuntu 20.04 vm im seeing | 12:23 |
| sean-k-mooney | Collecting oslo.policy>=3.8.1 | 12:23 |
| sean-k-mooney | Using cached oslo.policy-3.8.2-py3-none-any.whl (86 kB) | 12:23 |
| sean-k-mooney | INFO: This is taking longer than usual. You might need to provide the dependency resolver with stricter constraints to reduce runtime. If you want to abort this run, you can press Ctrl + C to do so. To improve how pip performs, tell us what happened here: https://pip.pypa.io/surveys/backtracking | 12:24 |
| sean-k-mooney | for other packages too not just oslo.policy | 12:24 |
| sean-k-mooney | i kicked of an install on friday and it timed out installing packages | 12:25 |
| sean-k-mooney | right now this is installing cinder where it is failng | 12:25 |
| sean-k-mooney | sudo -H LC_ALL=en_US.UTF-8 SETUPTOOLS_USE_DISTUTILS=stdlib http_proxy= https_proxy= no_proxy= PIP_FIND_LINKS= SETUPTOOLS_SYS_PATH_TECHNIQUE=rewrite python3.8 -m pip install -c /opt/stack/requirements/upper-constraints.txt -e /opt/stack/cinder | 12:25 |
| sean-k-mooney | perhaps i need to update my local copy of the requirements repo although i have RECLONE=True set so devstack should do that | 12:26 |
| sean-k-mooney | yoctozepto: fungi do you know if we are seeing anything like that in the gate jobs or if they have recently started taking longer where this might be happening but not to a degree that would make the job fail just make it slower | 12:28 |
| fungi | sean-k-mooney: the examples i looked at were barbican and pylint getting installed unconstrained in very large dependency sets | 12:32 |
| fungi | pylint manifests as pip struggling to find an appropriate version of its astroid dependency i think | 12:33 |
| yoctozepto | sean-k-mooney: it was reported on launchpad already | 12:33 |
| yoctozepto | I have not hit it but, alas, I am not doing enough openstack these days | 12:33 |
| fungi | probably the most effective ways to reduce that are to not install unconstrainted dependencies in large sets where there may be overlap with transitive deps | 12:34 |
| sean-k-mooney | well we nolonger install test-requirements | 12:34 |
| sean-k-mooney | so pylint should not be installed by devstack | 12:34 |
| fungi | that ought to help | 12:34 |
| fungi | ahh, you're seeing it in a devstack install? | 12:34 |
| sean-k-mooney | yes | 12:35 |
| sean-k-mooney | fresh ubuntu 20.04 vm i created on friday | 12:35 |
| fungi | i was running it down on tox-based jobs on friday | 12:35 |
| sean-k-mooney | i had to remove python3-simplejson and i removed python3-yaml just incase | 12:35 |
| sean-k-mooney | then i stacked and hit resolver timesouts | 12:35 |
| fungi | yeah, make sure you have as current an upper-constraints.txt for the branch as possible | 12:35 |
| sean-k-mooney | also issues with suds-jurko>=0.6 | 12:36 |
| sean-k-mooney | where it tried to use use_2to3 | 12:36 |
| sean-k-mooney | ya i just deleted the requirement repo and did pull on devstack to latest | 12:36 |
| sean-k-mooney | looks like im going to hit the smae issue agin with cinder | 12:36 |
| sean-k-mooney | INFO: pip is looking at multiple versions of oslo-policy to determine which version is compatible with other requirements. This could take a while. | 12:37 |
| sean-k-mooney | its just a guess but i would assume https://paste.opendev.org/show/809899 might be the issue | 12:37 |
| sean-k-mooney | it might be forceing use to try and find some weird set of deps | 12:38 |
| sean-k-mooney | i wonder if we can just remove suds-jurko | 12:41 |
| sean-k-mooney | https://codesearch.opendev.org/?q=suds-jurko&i=nope&literal=nope&files=&excludeFiles=&repos= | 12:42 |
| sean-k-mooney | it looks like its there for vmware or powervm | 12:42 |
| sean-k-mooney | im not sure why it woudl be installed by as part of installing cidner | 12:42 |
| sean-k-mooney | i suspect this is being caused by cinder dep on oslo.vmware | 12:47 |
| sean-k-mooney | https://opendev.org/openstack/cinder/src/branch/master/requirements.txt#L58 | 12:47 |
| sean-k-mooney | i wonder if we can make that optional and not installed by default | 12:47 |
| sean-k-mooney | e.g. so you woudl have do do pip install cinder[vmware] or something using extras | 12:48 |
| fungi | is oslo.vmware installable on its own with the same constraints set? | 13:04 |
| fungi | once you get things installed, maybe check to see if there are any packages present which are not in upper-constraints.txt | 13:04 |
| fungi | i have a feeling something is being installed which isn't in the list of projects the constraints generator is smashing together | 13:05 |
| *** sboyron_ is now known as sboyron | 13:05 | |
| *** sboyron is now known as sboyron_ | 13:06 | |
| *** sboyron_ is now known as sboyron | 13:06 | |
| fungi | sean-k-mooney: pip shouldn't be doing that if there's not something else unconstrained also depending on that package | 13:32 |
| fungi | is there a build in the opendev zuul which exhibits this behavior? i'm happy to take a closer look at the logs for it | 13:32 |
| fungi | problem is, just looking at the cinder install output is insufficient since pip is installing into a shared environment and these days it takes the requirements of everything that's already installed into account, so you really need to know what the initial state of the environment was before (via pip list or pip freeze) | 13:35 |
| sean-k-mooney | fungi: not that im aware of this is more or less a defautl devstack install on ubuntu 20.04 https://paste.opendev.org/show/809902/ | 13:38 |
| fungi | we do perform default ubuntu installs on ubuntu 20.04 in ci as well | 13:39 |
| fungi | er, default devstack installs i mean | 13:39 |
| fungi | however we go out of our way to have as little preinstalled on the servers as possible, because `sudo pip install` fights with any distro-supplied python packages on the system too | 13:39 |
| fungi | sean-k-mooney: any chance you could try with one of our ci images? | 13:40 |
| fungi | just curious if that's what's making it not break in our jobs | 13:40 |
| fungi | that's one of the reasons we don't use cloud-init, it has far too many python package dependencies | 13:42 |
| sean-k-mooney | am maybe. im using one of our internal cloud i can upload image but it takes a while since it has to be coplied over the vpn and trans atlantic | 13:46 |
| fungi | oh, yeah you probably don't want to do that then, our images are gigantic because of including a cache of all the git repos in opendev | 13:46 |
| fungi | but regardless, look at `dpkg -l|grep '^ii python3-'` and see what all you can remove. anything which is listed in upper-constraints.txt is a possible candidate | 13:48 |
| sean-k-mooney | some of those i tought i removed already | 13:50 |
| sean-k-mooney | i have 77 packages | 13:50 |
| sean-k-mooney | i really dont need twisted installed | 13:50 |
| fungi | there's already some machinery in devstack for forcibly uninstalling conflicting distro packages, so if you can identify which ones are causing the problem (assuming that's even the problem) then we can probably extend devstack to make it install more automatically on a generic ubuntu image | 13:53 |
| sean-k-mooney | ya i have seen that code in the past. we have two ways one is unsitalling htem and the other were its not removable we manually mask/delete stuff | 13:54 |
| sean-k-mooney | if it continues to be an issue i can just move to another base image or distro | 13:57 |
| fungi | right, obviously prefer the first, the second is a very hacky last resort | 13:57 |
| sean-k-mooney | i removed an hand full of packates that i dont think need to be there so ill see if that helps | 13:57 |
| fungi | well, it's intended that people be able to install devstack on generic cloud images, we just don't have an easy way to test that | 13:57 |
| fungi | so if we can fix it to uninstall things which are causing dep solver resolution issues for pip now, that would be good | 13:58 |
| sean-k-mooney | it might be helping a littl but im in the same situraiton with suds-jurko then everything fails in cinder after that fials to isntall form pip | 13:59 |
| sean-k-mooney | it was resolveing a little faster but i think we really do need to get rid of that dep form oslo.vmware | 14:00 |
| sean-k-mooney | or make oslo.vmware an optional dep of cidner and remove it out of its requirements.txt | 14:00 |
| sean-k-mooney | i might just do that locally to check | 14:00 |
| kopecmartin | isabek: hi, just pass --serial arg to run the tests not in parallel - https://docs.openstack.org/tempest/latest/run.html#test-execution | 14:03 |
| sean-k-mooney | fungi: so ya removing oslo.vmware form cinder fixes the issue | 14:05 |
| sean-k-mooney | it really does like like suds-jurko is the problem | 14:06 |
| sean-k-mooney | suds-jurko is still using use_2to3 and we require >=0.6 | 14:06 |
| sean-k-mooney | there is no version that is >=0.6 that does not use use2to3 since there has been no release of that since 2014 | 14:07 |
| fungi | sean-k-mooney: ahh, i wonder if jobs are avoiding that by pinning to older setuptools | 14:07 |
| sean-k-mooney | ya perhaps | 14:14 |
| sean-k-mooney | the install is still running but just removing oslo.vmware which is what is pulling it in seams to do the trick | 14:15 |
| isabek | kopecmartin: Thanks for link. I have tried with this arg, still failing. I have used: https://docs.openstack.org/tempest/latest/HACKING.html#parallel-test-execution LockFixture failing too :( | 14:15 |
| frickler | sean-k-mooney: fungi: see https://bugs.launchpad.net/cinder/+bug/1946340 | 14:20 |
| kopecmartin | isabek: hm, what is it failing with? maybe it's not an issue of concurrency | 14:20 |
| frickler | it is not seen in CI because we install from old wheels | 14:20 |
| fungi | frickler: thanks! | 14:20 |
| fungi | makes sense, we do prebuild wheels, and likely built those ages ago with older setuptools | 14:20 |
| frickler | see also https://review.opendev.org/c/openstack/requirements/+/813292 for a job that tries to run without those | 14:21 |
| frickler | on top of that is the list of things in u-c that currently would fail | 14:21 |
| frickler | designate and cinder are the projects that I know to be affected directly | 14:21 |
| sean-k-mooney | frickler: yep same issue on ubuntu 20.04 | 14:22 |
| frickler | hotfix is to remove suds-jurko or oslo.vmware from their req.txt | 14:22 |
| sean-k-mooney | yes | 14:22 |
| sean-k-mooney | i just removed oslo.vmware and it just finished stacking | 14:23 |
| sean-k-mooney | form cinder | 14:23 |
| sean-k-mooney | not deploying with designate currently | 14:23 |
| isabek | kopecmartin: it is concurrency, during list resource when getting required resource body is passing. but sometimes is failing | 14:24 |
| isabek | my tests creating own resources | 14:25 |
| sean-k-mooney | oslo.vmware will either need to move to suds https://github.com/cackharot/suds-py3 or otherwise drop the dempency | 14:25 |
| sean-k-mooney | or better yet https://github.com/suds-community/suds | 14:25 |
| frickler | yes, someone proposed a patch to designate with the latter I think | 14:30 |
| frickler | https://review.opendev.org/c/openstack/designate/+/813380 | 14:31 |
| sean-k-mooney | i guess we will have to do the same for oslo.vmware | 14:40 |
| opendevreview | Carlos Eduardo proposed openstack/devstack-plugin-ceph master: WIP Fix Ganesha repo urls https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/813446 | 14:53 |
| opendevreview | Archit Modi proposed openstack/whitebox-tempest-plugin master: Wait after image copy https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/813455 | 16:18 |
| opendevreview | Balazs Gibizer proposed openstack/tempest master: Simplify compute schema handling https://review.opendev.org/c/openstack/tempest/+/813459 | 16:30 |
| opendevreview | Balazs Gibizer proposed openstack/tempest master: DNM: Example of a compute schema change https://review.opendev.org/c/openstack/tempest/+/813460 | 16:30 |
| *** jpena is now known as jpena|off | 16:45 | |
| opendevreview | Luigi Toscano proposed openstack/devstack-plugin-ceph master: WIP Allow the usage of Clone v2 API https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/782624 | 17:39 |
| opendevreview | Luigi Toscano proposed openstack/tempest master: DNM test snapshot deletion with deps on ceph https://review.opendev.org/c/openstack/tempest/+/782392 | 17:39 |
| *** dviroel is now known as dviroel|out | 21:16 | |
| opendevreview | Merged openstack/devstack-plugin-ceph stable/wallaby: Use Ceph Pacific for devstack-plugin-ceph-tempest-py3 jobs https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/810059 | 23:20 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!