frickler | I had this bug in early jammy testing where pip was missing in a venv (duplicate of https://bugs.launchpad.net/ubuntu/+source/python3.10/+bug/1962791) | 05:46 |
---|---|---|
frickler | I wonder whether the fix for that may be related to this | 05:46 |
*** lajoskatona_ is now known as lajoskatona | 08:29 | |
fungi | clarkb: dansmith: something similar-sounding came up in the python discourse today, regarding apple's patched sysconfig in the python shipped in xcode, i wonder if the distro package could be doing something similar? https://github.com/pypa/pip/blob/22.1/src/pip/_internal/locations/__init__.py#L486-L498 | 11:48 |
fungi | https://discuss.python.org/t/understanding-site-packages-directories/12959/17 | 11:52 |
noonedeadpunk | Hey there! Sorry, I didn't dig too much, decieded to ask stupid question at once. Were there some changes of merging heat plugin to core of tempest or smth like that? | 15:03 |
noonedeadpunk | As when I'm installing heat-tempest-plugin tempest cli just falls apart | 15:03 |
noonedeadpunk | ie `tempest workspace list` silently exist with rc1 | 15:03 |
noonedeadpunk | the moment I drop heat plugin things are good | 15:04 |
noonedeadpunk | Seing https://opendev.org/openstack/tempest/src/branch/master/tempest/services/orchestration/json/orchestration_client.py sounds like it it | 15:06 |
noonedeadpunk | but don't see tests at same time | 15:06 |
noonedeadpunk | Most frustrating part is that even no stack trace is shown | 15:15 |
gmann | noonedeadpunk: no, there is no plan to merge heat plugin into tempest. orchestration_client.py in tempest is supposed to be removed. I remember i had kept it as todo because it has some dependency on murano tempest plugin. I will start fixing the deps and remove it from tempest tree,. | 15:18 |
gmann | noonedeadpunk: orchestration_client.py is pending work of migrating heat tests into heat-tempest-plugin | 15:18 |
noonedeadpunk | Then I wonder if I'm the only one who can't init workspace when heat plugin is installed | 15:18 |
noonedeadpunk | I don't know tempest code good enough. So if you have some guess where to start looking - that would be helpful:) | 15:19 |
noonedeadpunk | just in case I tried tempest==30.1.0 and 29.2.0, heat plugin - 1.5.0, 1.4.0 | 15:20 |
noonedeadpunk | I can reproduce it with just `pip install tempest heat-tempest-plugin` | 15:23 |
noonedeadpunk | https://paste.openstack.org/show/bIVgD948OUQ2e1XpScoT/ | 15:24 |
clarkb | fungi: as far as I could tell the behavior I described and patched for us all upstream code related | 15:24 |
clarkb | fungi: the problem arises on my suse machine as well if I do editable installs in a virtualenv which should be very isolated from the distro | 15:25 |
noonedeadpunk | And I bet that happens just after some dependancy update, as this started failing several weeks ago without changing a thing in CI | 15:25 |
clarkb | fungi: reading that linked message this may explain why only 3.10 beraks hard though | 15:26 |
gmann | noonedeadpunk: it seems heat-tempest-plugin installation issue. Tempest is installing all fine. | 15:27 |
gmann | noonedeadpunk: " ERROR: Failed building wheel for Yappi" may be clarkb or fungi can help here | 15:28 |
noonedeadpunk | It fallsback actually | 15:28 |
noonedeadpunk | `Running setup.py install for Yappi ... done` | 15:28 |
noonedeadpunk | so wheel fails, but installation of it is done at the end | 15:29 |
noonedeadpunk | But should actually tempest cli jsut silently fail because of some plugin being weird at all? | 15:29 |
gmann | that's the latest run of heat-tempest-plugin but 10 days ago, not sure if that is happening in our jobs too https://review.opendev.org/c/openstack/heat-tempest-plugin/+/841136 | 15:29 |
clarkb | noonedeadpunk: gmann: that isn't a failure | 15:30 |
clarkb | its a weird pip behavior where it tries to build a wheel for the package first to install the wheel. But if that fails it falls back to making an sdist | 15:30 |
noonedeadpunk | yup^ | 15:30 |
clarkb | in this case the wheel build failed because you don't have the `wheel` packge installed which gives it the ability to run | 15:30 |
clarkb | but it isn't a failure | 15:30 |
gmann | ok | 15:31 |
gmann | noonedeadpunk: you can check if plugin is installed or not via 'tempest list-plugins'. but tempest workspace cli does not fail if any explicit installation of plugin is failed. | 15:31 |
gmann | noonedeadpunk: what is exact error you are getting in tempest cli | 15:33 |
gmann | noonedeadpunk: and you are not creating any workspace just running 'tempest workspace list' ? | 15:34 |
clarkb | on a newly held jammy node I have a dist-info dir for dbcounter but no dbcounter.py under dist-packages which is what I expect based on a local install | 16:39 |
clarkb | so it seems there is another problem on jammy | 16:39 |
clarkb | dbcounter.py is not ending up in the SOURCES.txt or RECORD files | 16:41 |
opendevreview | Clark Boylan proposed openstack/devstack master: DNM debugging jammy pip install stuff https://review.opendev.org/c/openstack/devstack/+/842620 | 16:52 |
clarkb | I think ^ may fix it. If that manages to pass I'll clean up the change and propose it as something that is mergable | 16:53 |
noonedeadpunk | gmann: it;s installed - `| heat | heat_tempest_plugin.plugin:HeatTempestPlugin |` | 17:42 |
noonedeadpunk | gmann: in tempest workspace list I don't get literally anything | 17:42 |
noonedeadpunk | *for `tempest workspace list` Same goes if I want to init new workspace | 17:43 |
gmann | noonedeadpunk: you need to use this 'tempest list-plugins' to know/list all installed plugins. 'tempest workspace' is separate things. let me know if you created/init the workspace and list does not show that workspace | 17:44 |
gmann | this is workflow of 'tempest workspace' https://docs.openstack.org/tempest/16.1.0/workspace.html | 17:45 |
gmann | https://docs.openstack.org/tempest/latest/workspace.html | 17:45 |
noonedeadpunk | gmann: https://paste.openstack.org/show/b0u2RAEsc8bjYDJVzVMr/ | 17:46 |
gmann | cool, so heat plugin is there | 17:46 |
noonedeadpunk | yup. And that was the only thing I installed for testing :) | 17:47 |
clarkb | what is the actual failure? I don't think we've seen that yet? | 17:56 |
bkopilov_ | gmann, Hi , please review , let me know what do you think ? https://review.opendev.org/c/openstack/tempest/+/840410 | 17:59 |
gmann | noonedeadpunk: so you are not creating/registering the workspace so what is the expectation and did not work | 17:59 |
gmann | bkopilov_: ack, I was waiting for unit test which you added. will check today | 17:59 |
bkopilov_ | gmann, perfect thanks a lot ! | 18:00 |
gmann | np! | 18:00 |
noonedeadpunk | gmann: sorry didn't really got question. Expectation is that `tempest init --name workspace $path` or `tempest workspace list` works :) | 18:08 |
noonedeadpunk | gmann: basically we have script https://opendev.org/openstack/openstack-ansible-os_tempest/src/commit/c3a2a979744a617b04e384846a6a2b016784e19d/tasks/tempest_post_install.yml#L89-L123 | 18:08 |
noonedeadpunk | and when we have heat installed, all CI fails | 18:09 |
noonedeadpunk | Which basically also blocks magnum, sahara and several more CI | 18:09 |
gmann | ok, let me check. Its been while I have tried workspace cli | 18:10 |
noonedeadpunk | shouldn't it be used? what's the recommended way? | 18:15 |
noonedeadpunk | as I guess if you want to run tempest not in CI, workspaces make sense, right? | 18:16 |
gmann | noonedeadpunk: yes, workspace make more sense and should work. btw where you are registering the workspace ? | 18:20 |
noonedeadpunk | gmann: Tried `/root/workspace` but in script it's basically $HOME | 18:21 |
noonedeadpunk | or well, `tempest init --name workspace /root/workspace` but it also fails silently with return code 1 | 18:21 |
gmann | noonedeadpunk: 'tempest workspace list' and register working fine. let me try with tempest init | 18:39 |
noonedeadpunk | um.... | 18:40 |
noonedeadpunk | how's that possible.... | 18:41 |
noonedeadpunk | I guess I know what's the issue.... | 18:41 |
noonedeadpunk | It doesn't like `[heat_plugin]/endpoint_type = internalURL` in /etc/tempest/tempest.conf. I didn't know that workspace init would read and rely on that | 18:43 |
noonedeadpunk | because it should be ``[heat_plugin]/endpoint_type = internal` | 18:44 |
noonedeadpunk | uh | 18:44 |
opendevreview | Clark Boylan proposed openstack/devstack master: Fix dbcounter installation on Jammy https://review.opendev.org/c/openstack/devstack/+/842620 | 18:44 |
noonedeadpunk | I wish failure output was more verbose :) | 18:44 |
clarkb | dansmith: frickler ^ I expect that change is mergeable now and hopefully I've written enough text about the situation to explain all the weirdness here | 18:44 |
noonedeadpunk | gmann: thanks a lot for help and sorry for wasting your time | 18:45 |
noonedeadpunk | But I wouldn't catch that without you ;) | 18:45 |
gmann | noonedeadpunk: ok, i ran init also and worked fine for me. | 18:52 |
gmann | noonedeadpunk: I see, so tempest config things can cause these error as you mentioned | 18:52 |
gmann | because init do parse the config file and udpate it | 18:53 |
noonedeadpunk | the worst part is it's silent... | 18:53 |
noonedeadpunk | having stack trace would be so much a relief.... | 18:53 |
gmann | yeah, we should put some log or stop there | 18:54 |
* noonedeadpunk really hopes that `internal` is valid this for _all_ plugins | 18:54 | |
gmann | noonedeadpunk: tempest accept both but seems like heat does not https://github.com/openstack/heat-tempest-plugin/blob/master/heat_tempest_plugin/config.py#L79 | 18:56 |
gmann | octavia plugin also https://github.com/openstack/octavia-tempest-plugin/blob/master/octavia_tempest_plugin/config.py#L48-L51 | 18:57 |
gmann | but 'internal' should work for all | 18:57 |
noonedeadpunk | yeah. already pushed https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/842758 | 19:01 |
noonedeadpunk | thanks a lot! | 19:01 |
gmann | +1 | 19:02 |
opendevreview | Ghanshyam proposed openstack/tempest master: Add py310 job in gate https://review.opendev.org/c/openstack/tempest/+/832144 | 19:09 |
opendevreview | Ghanshyam proposed openstack/grenade master: Do not run grenade-skip-level job on older stable branches https://review.opendev.org/c/openstack/grenade/+/842799 | 19:25 |
opendevreview | Merged openstack/eslint-config-openstack master: Switch testing to Zed testing runtime https://review.opendev.org/c/openstack/eslint-config-openstack/+/840494 | 22:28 |
opendevreview | Merged openstack/karma-subunit-reporter master: Switch testing to Zed testing runtime https://review.opendev.org/c/openstack/karma-subunit-reporter/+/840496 | 22:29 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!