October 2020 - openstack-discuss - lists.openstack.org

[nova] Weekly meeting canceled
by Balázs Gibizer 29 Oct '20

29 Oct '20

Hi, Today's meeting is canceled in favor of the PTG session. See you there. Cheers, gibi

1 0

[Openstack][Queens] iscsi unity driver
by Ignazio Cassano 28 Oct '20

28 Oct '20

Hello, I installed queens on centos7 with cinder unity iscsi driver. When volume are detached for example for a live migration, faulty devices are left on compute nodes. Must I configure iscsi initiator on both compute and controllers nodes? At this time I configured it only on compute nodes. Faulty could be an os brick module issue ? Thanks Ignazio

1 0

[ironic] Reviewing changes post-gerrit admin account compromise
by Julia Kreger 28 Oct '20

28 Oct '20

Greetings folks, Per the opendev notice that was published on the gerrit outage and compromise[0], I believe it is only prudent to sift through the changes published and verify if their content looks okay and that they were published as we expect. As we have a number of repositories, branches, and ultimately change sets to check, I've created a spreadsheet to help us track and coordinate the work[1]. In the interest of transparency, it is publicly viewable, but I've only shared editing ability with the members of ironic-core. Arne, unfortunately I don't see a google account for yourself, if you have an account you want me to add, let me know! Ironic-cores: If someone wishes to help, I personally think that is okay, but if you feel the need to update editing privileges on the document, please just reply to the mailing list so we have it on the record that this is the case. If I missed a branch that needs to be checked, just add it to the list. Same for any project repositories. And same goes for tabs. Copy/paste is likely easiest. Thanks everyone! -Julia [0]: http://lists.opendev.org/pipermail/service-announce/2020-October/000011.html [1]: https://docs.google.com/spreadsheets/d/1-_on0DOpS4gRriAnpdoLf2B7K98RZN7hNAP…

1 1

[policy] Wallaby Virtual Forum & PTG Summary
by Ghanshyam Mann 28 Oct '20

28 Oct '20

Hello Everyone, 'policy' popup team hosted 2 sessions in Wallaby Virtual PTG to discuss the progress and next steps. Forum: ===== https://etherpad.opendev.org/p/consistent-and-secure-default-policies-walla… Progress: * Three projects completed till now - Keystone, Nova and, Cyborg * One in progress - Barbican * Planning in Wallaby - Neutron, Manila We discussed the current and possible future challenges while migrating to the new policy. policy file in JSON format is one know and we talked about a current workaround and long term plan. Deprecation warnings are still an issue and a lot of warnings are logged. This is logged as a bug in nova[1]. Also, the HTML version of the policy doc does not have the deprecation rule and reason. Example: https://docs.openstack.org/nova/latest/configuration/policy.html We need to add it in these docs too. A clear step by step document about how to use system scope in cloud.yaml as well as in general with all migration steps are much needed. We also asked if any deployment is migrated to new policy but there is none yet. PTG: ==== https://etherpad.opendev.org/p/policy-popup-wallaby-ptg We carried the Forum sessions discussion in PTG with few extra topics. Migrate Default Policy Format from JSON to YAML (All projects) -------------------------------------------------------------------------------- We talked about it and decided that it will be great to do this in advance before projects start moving towards the new policy. and having this as a community goal in wallaby will help this effort to move faster. I have proposed this as a goal in TC PTG and it was agreed to select it for wallaby. - Goal proposal: https://review.opendev.org/#/c/759881/2 We also need to update devstack for the neutron policy file which is policy.json. Deprecation warning on-demand based on config option? ----------------------------------------------------------------------- There is no best solution for deprecation warnings when it is a lot in case of new policy work. We cannot stop logging warnings completely. We discussed an option to provide a config option to disable the warnings (enable by default) and only for default value change, not the policy name change. The policy name change is a little critical even in policy overridden case also that is why we should not make it switchable. This way operators can disable warnings after seeing it for the first time and if it is too noisy. New policy in Horizon ---------------------------- It is challenging to adopt new policy in Horizon when some projects have new policies and some not. We left this for now and will continue brainstorming once we do some investigation on the usage of system token for new policy and old one. For now, amotoki proposed a workaround to keep the policy file with the deprecated rule as the overridden rules. - https://review.opendev.org/#/c/750134/ What are the expectations from project teams? ----------------------------------------------------------- In the end, we discussed what all things to be targeted as part of new policy work and what all as a separate effort. Below is the list and I will be documenting it on wiki. Things to do as part of new policy effort: - Migrating JSON to YAML (in advance though) - Policy Tests coverage - System scoped administrator policy support and "reader" role support - upgrade-check support when changing the default policy As separate effort: - Remove hard-coded admin checks in the API/DB/"Views" - Standardizing policy names (depends on projects) [1] https://bugs.launchpad.net/nova/+bug/1900451 -gmann

1 0

[qa] Wallaby Virtual PTG Summary
by Ghanshyam Mann 28 Oct '20

28 Oct '20

Hello Everyone, I am summarizing the discussion QA team had in PTG and action items for quick ref. Victoria Retrospective: ----------------------------- We talked about the retrospective of the last cycle, one main challenge we discussed is the slow review in QA due to fewer contributors. Below are the action items that came up from improving things: * Action: - microversion need more work also when you try more combination or latest (gmann). - Triage bugs in the office hour as 30 min -> update agenda - Speed up the review. Office hour schedule: --------------------------- With daylight ending in Nov, we decided to shift the QA office hour by an hour late which is at 14 UTC. Move back the single horizon test in tempest tree from tempest-horizon: ------------------------------------------------------------------------------------------- Tempest and Horizon team decided to bring back the single horizon test to Tempest tree and retire the tempest-horizon plugins which was too much maintenance for single tests. It is unlikely to happen but if Tempest team in future plan to remove this test then it needs to be consulted with Horizon team first. * Action: - Move test back to Tempest - gmann - Deprecate and retire tempest-horizon - amotoki - Check with the release team if any other release needed. (amotoki + gmann) - modify the 'horizon-dsvm-tempest-plugin' job to direct to tests' new location Add run_validation as a centralized point for Tempest scenario tests: ------------------------------------------------------------------------------------- For now, validation resources automation and skip ssh part via run_validation is done in API test only and this proposal is to extend it to scenario tests also. There is no reason not to do that which will help to run tempest scenario tests on images where ssh is not allowed. There is a most possible situation where complete tests need to be skipped but we will see case by case and at least skipping automatically will help tester to avoid explicitly excluding the scenario tests to the skip list. *Action: - Implement it in scenario tests - paras333 - Refactor scenario manager's create_server method to consume common/compute.py - paras333 Patrole stable release: ---------------------------- We are lacking the maintainer in Patrole project which is a big challenge to release a stable version of it. Sonia and doug (already helping) will try to spend more bandwidth in Patrole. Another thing to reduce the Patrole tests execution time. We discussed a few options like policy engine to return the API based on the flag or use fake driver etc but we did not conclude any option yet as they need more investigation. * Action: - Keep eyes on gate and how tests are running and based on that we can make a stable release for Wallaby. - gmann - Keep exploring the options for making patrole lightweight. - gmann, doug Grenade: extending the testing for zero downtime upgrade testing: ------------------------------------------------------------------------------------ This is just a thought. There is TC tag 'assert:supports-zero-downtime-upgrade' which is not used by any of the projects and also we do not have any testing framework which can verify this tag. We talked about it if we can do such testing in grenade or not. Before moving forward on discussion and investigation we checked if anyone can volunteer for this work. As of now, no volunteer. Use different guest image for gate jobs to run tempest tests: --------------------------------------------------------------------------- We only test cirros guest image to create the test VM and all proposal from Paras is to try more images to enhance the different configuration scenarios in upstream testing. This will help to able to catch more failure/scenario at the upstream gate itself compare to the current situation where most of them are reported from downstream testing. * Action: - Add a job as experimental - paras333 Tempest Cred Manager proposed changes: ----------------------------------------------------- 'primary' and 'alt_promary' credentials in Tempest are hardcoded to non-admin and they are assigned configured 'tempest_role'. There is no way we can assign a different role to both of these creds. The idea here is to make 'primary' and 'alt_promary' credentials configurable so that different deployment can configure these with different roles in various combinations. We will be adding two config option similar to 'tempest_role' and default to an empty list so that we continue the default behavior with what we have currently. So that there is no backward-incompatible change instead it is an additional new thing provided. * Action: - Doug needs to propose the patch. Tempest Scenario manager stable version brainstorming: ----------------------------------------------------------------------- We talked about a few method changes to make the scenario manager stable. If any method is used among plugins only and not in Tempest then we do not actually need to move it to Tempest and they can stay on the side of the plugin itself. We ran out of time here and will continue the brainstorming in office hours or so. (this etherpad - https://etherpad.opendev.org/p/tempest-scenario-manager) * Action: - Sonia and Martin will continue working on this. Wallaby Priority & Planning: ------------------------------------ We did not prioritize the things as such but listed all working items in below etherpad along with Victoria cycle backlogs. - https://etherpad.opendev.org/p/qa-wallaby-ptg Detailed discussions are in this Etherpad: https://etherpad.opendev.org/p/qa-wallaby-ptg -gmann

1 0

[openstack-ansible] Gerrit breach and auditing all commits since Oct 01
by Dmitriy Rabotyagov 28 Oct '20

28 Oct '20

Hi! Most of you already know that Gerrit has been compromised a while ago [1]. Infra team has dealed with it very efficiently and I can assure you that this issue hasn't touched openstack-ansible project. I have looked through all commits that were merged in period from Oct 1 - Oct 21 (and we've got more than 200 of them during this timeframe, wow!), and all of them look valid and were properly approved. I have also double checked our releases and didn't found any extra or malicious ones. So we are pretty safe and can sleep calm [1] http://lists.opendev.org/pipermail/service-announce/2020-October/000011.html -- Kind Regards, Dmitriy Rabotyagov

1 0

[gate] many jobs failing for nova-compute libvirt driver error
by melanie witt 28 Oct '20

28 Oct '20

Hey all, This is just a FYI that we've got another gate failure where nova-compute is failing with the following error raised from the libvirt driver [1]: "TypeError: Parameterized generics cannot be used with class" which was raised from the eventlet.tpool code. (I don't yet understand what this error means exactly.) Because of this, we suspect this is related to the eventlet package version bump from 0.26.1 to 0.28.0 that happened in the upper-constraints change to openstack/requirements that merged earlier today: https://review.opendev.org/#/c/750084/47/upper-constraints.txt@150 But we're not certain yet exactly what is happening and whether this ^ is indeed the cause. I am currently investigating to find the root cause of failure and how to fix it/confirm which package we should pin for now. I am trying out a DNM nova patch to pin eventlet to 0.26.1 to see what happens, while I look into more depth at what the error means and whether it's something we should fix in the nova code. If anyone can help, it would be appreciated. Best, -melanie [1] https://zuul.opendev.org/t/openstack/build/20f1d309663347c28112b711b82b9c03…

4 7

[qa] Wallaby PTG Wrap-up and office hour was changed to Tue 1400 UTC
by Masayuki Igawa 28 Oct '20

28 Oct '20

Hi, Thank you for attending our sessions! We had discussed many things during the virtual PTG. Topics are the following. And as we discussed our office hour office time 1 hour earlier than the current time. So, the office hour starts at 1400 UTC on every Tuesday. ========= Topics: Mon: @ Juno Greetings and ordering the topics Victoria Retrospective Move back the single horizon test in tempest tree from tempest-horizon Add run_validation as a centralised point for Tempest scenario tests Tue: @Juno Patrole stable release Grenade: extending the testing for zero downtime upgrade testing Use different guest image for gate jobs to run tempest tests Tempest Cred Manager proposed changes Tempest Scenario manager stable version brainstorming Bug triaging -> moved to the office hour Wallaby priority items https://etherpad.opendev.org/p/qa-wallaby-priority Etherpad: https://etherpad.opendev.org/p/qa-wallaby-ptg You can find the recording video files on the etherpad. ========= -- Masayuki Igawa

2 1

[Neutron] no L3 meeting this week (2020-June-01)
by LIU Yulong 28 Oct '20

28 Oct '20

Hi there, Due to the PTG meetings this week, we will cancel the L3 meeting this week. See you guys online next week. Regards, LIU Yulong

1 2

[masakari][ptg] I cannot attend the 2nd Masakari PTG session
by Radosław Piliszek 28 Oct '20

28 Oct '20

Dears, I regret to inform that, due to personal matters (augmented by the pandemic and the current situation in my country), I will not be able to attend the 2nd Masakari PTG session: Thursday October 29, 2020 13:00 - 15:00 (UTC) The good news is we don't have that many topics to go through and I am generally able to get more time from later slots if we need to have it. -yoctozepto

1 0