| opendevreview | melanie witt proposed openstack/tempest master: Add tests for vTPM live migration https://review.opendev.org/c/openstack/tempest/+/957475 | 01:37 |
|---|---|---|
| opendevreview | melanie witt proposed openstack/tempest master: Add more tests for vTPM live migration https://review.opendev.org/c/openstack/tempest/+/977903 | 01:37 |
| opendevreview | melanie witt proposed openstack/whitebox-tempest-plugin master: Test vTPM live migration https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/955969 | 01:53 |
| opendevreview | melanie witt proposed openstack/whitebox-tempest-plugin master: Test vTPM live migration 'deployment' mode https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/977904 | 01:53 |
| opendevreview | melanie witt proposed openstack/whitebox-tempest-plugin master: Test vTPM live migration https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/955969 | 02:01 |
| opendevreview | melanie witt proposed openstack/whitebox-tempest-plugin master: Test vTPM live migration 'deployment' mode https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/977904 | 02:01 |
| opendevreview | Merged openstack/tempest master: Test create VM with unaddressed port https://review.opendev.org/c/openstack/tempest/+/959187 | 03:24 |
| opendevreview | Merged openstack/tempest master: Replace skip with skipTest https://review.opendev.org/c/openstack/tempest/+/977616 | 06:09 |
| opendevreview | Balazs Gibizer proposed openstack/tempest master: Revert "Test create VM with unaddressed port" https://review.opendev.org/c/openstack/tempest/+/977922 | 08:49 |
| gibi | folks ^^ needed to unblock nova-next. the original test landed during the night | 08:53 |
| opendevreview | Merged openstack/tempest master: Revert "Test create VM with unaddressed port" https://review.opendev.org/c/openstack/tempest/+/977922 | 08:57 |
| opendevreview | Balazs Gibizer proposed openstack/tempest master: DNM: Reapply "Test create VM with unaddressed port" https://review.opendev.org/c/openstack/tempest/+/977930 | 09:19 |
| gibi | folks we reverted the new tempest test blocking nova-next and now I proposed the reapply patch ^^ with an addition of nova-next job trigger to show the issue ^^ | 09:22 |
| opendevreview | Pranali Deore proposed openstack/tempest master: Add tests for new add location api of glance https://review.opendev.org/c/openstack/tempest/+/891026 | 12:06 |
| opendevreview | Johannes Kulik proposed openstack/tempest master: Test Nova API 2.101 async volume attachments https://review.opendev.org/c/openstack/tempest/+/977952 | 12:07 |
| opendevreview | Pranali Deore proposed openstack/tempest master: Add tests for new add location api of glance https://review.opendev.org/c/openstack/tempest/+/891026 | 12:11 |
| stephenfin | gmaan: Care to review https://review.opendev.org/c/openstack/releases/+/977953 ? It just gets a release with the testtools skip fix out the door | 12:29 |
| opendevreview | Douglas Viroel proposed openstack/devstack-plugin-prometheus master: Fix exporter URL to support multinode deployments https://review.opendev.org/c/openstack/devstack-plugin-prometheus/+/977956 | 13:14 |
| frickler | gmaan: sean-k-mooney: do you have some context on https://review.opendev.org/c/openstack/tempest/+/977930 ? like could it be that this is only possible with certain neutron backends? I also totally don't understand the use case for it, it seems to be very much contrary to how I expect neutron to work | 13:16 |
| sean-k-mooney | frickler: it requires specific neutron extentions to be enabled | 13:22 |
| sean-k-mooney | however its possibel to have neutron report the extnsion to be avialable but not have ti funcitonal | 13:22 |
| sean-k-mooney | the curernt patch is checkign for ip_allocation | 13:23 |
| sean-k-mooney | and skiping based on that | 13:23 |
| sean-k-mooney | https://github.com/openstack/neutron-lib/blob/master/neutron_lib/api/definitions/ip_allocation.py | 13:24 |
| sean-k-mooney | i suspect that this si a bug in the ml2/ovn extion reporting | 13:24 |
| sean-k-mooney | the simplest fix for now is to just add a feature flag for this to the tempest config | 13:25 |
| sean-k-mooney | until folks have time to figure out why this is not working | 13:25 |
| opendevreview | Johannes Kulik proposed openstack/tempest master: Test Nova API 2.101 async volume attachments https://review.opendev.org/c/openstack/tempest/+/977952 | 13:26 |
| sean-k-mooney | frickler: i dont see the ip_allocation exteion listed here https://github.com/openstack/neutron/blob/master/neutron/common/ovn/extensions.py#L146 | 13:26 |
| sean-k-mooney | so it should not be reported in ovn based jobs and it should have skiped | 13:27 |
| sean-k-mooney | so i dont know why this wedged the nova-next job | 13:27 |
| sean-k-mooney | is it possibel that temepst is not using the api respoce in the skip check https://review.opendev.org/c/openstack/tempest/+/977930/1/tempest/api/compute/servers/test_create_server.py#164 | 13:30 |
| sean-k-mooney | and relyingon its config or something? | 13:30 |
| sean-k-mooney | ralonsoh: ^ any idea why that is not enough or who would be able to help debug that | 13:33 |
| sean-k-mooney | oh wait... nova-next is using ml2/ovs | 13:35 |
| ralonsoh | sean-k-mooney, let me read first | 13:36 |
| sean-k-mooney | really i tought we chagne that years ago and only tested that in the nova-alt-config (formally nova-hybrid-plug) job | 13:36 |
| ralonsoh | sean-k-mooney, in any case, that's a bug for OVN if we are missing this extension there | 13:37 |
| ralonsoh | I'll create a bug and a patch | 13:37 |
| sean-k-mooney | so nova-next is still using ml2/ovs but it is testign the ovs firewall driver | 13:38 |
| sean-k-mooney | i forgot about that we wanted to test all 3 ml2/ovn ml2/ovs with ovs firewall and ml2/ovs with iptables | 13:39 |
| ralonsoh | just to confirm, this is failing with ovs+native FW | 13:39 |
| sean-k-mooney | if i recally corectly when you added the extention filtering fature i think that was only done for ml2/ovn or started with it | 13:39 |
| sean-k-mooney | ralonsoh: its failling in a job that is cofnigure for that yes | 13:40 |
| sean-k-mooney | but i don tknow if we have enabled that fucnctionality correctly | 13:40 |
| sean-k-mooney | i think the problem might be htat ml2/ovs is reporting supprot fro it without use properly configurign it? | 13:40 |
| ralonsoh | let me check the neutron API logs | 13:41 |
| sean-k-mooney | i need ot check how this is ment to be enabled | 13:41 |
| sean-k-mooney | ack | 13:41 |
| ralonsoh | Feb 25 09:36:20.335534 np9ec2ce0d381a4 devstack@neutron-api.service[76980]: DEBUG neutron.api.extensions [None req-ef5af19a-c63e-4033-bd14-1c11688de118 None None] Ext name="IP Allocation" alias="ip_allocation" description="IP allocation extension." updated="2016-06-10T23:00:00-00:00" {{(pid=76980) _check_extension /opt/stack/neutron/neutron/api/extensions.py:415}} | 13:41 |
| ralonsoh | Feb 25 09:36:20.335804 np9ec2ce0d381a4 devstack@neutron-api.service[76980]: INFO neutron.api.extensions [None req-ef5af19a-c63e-4033-bd14-1c11688de118 None None] Extension ip_allocation not supported by any of loaded plugins | 13:41 |
| ralonsoh | so no, it is discarded | 13:41 |
| sean-k-mooney | right so ^ | 13:41 |
| sean-k-mooney | is the issue i was refering too | 13:41 |
| sean-k-mooney | the tempest check is checkign if the extenion is reported and skiping if its not | 13:42 |
| sean-k-mooney | but for ml2/ovs | 13:42 |
| sean-k-mooney | it will report the extions as aviable | 13:42 |
| sean-k-mooney | even if its not properly configured | 13:42 |
| sean-k-mooney | so we either need to work around that in tepemsest, fix that in neutorn with the saem filtering work you did for ovn | 13:42 |
| sean-k-mooney | or fix the nova-next job to cofnigure the extion properly | 13:43 |
| sean-k-mooney | or any combintion of those 3 | 13:43 |
| ralonsoh | for ovn is a matter of adding it to the static list | 13:44 |
| ralonsoh | and for ovs, should be configured as extension | 13:44 |
| sean-k-mooney | does ovn supprot adressles ports properly? not l3 routed network but l2 networks with only a mac and no ip | 13:44 |
| sean-k-mooney | it should just asking | 13:45 |
| ralonsoh | I think it does but must be tested | 13:45 |
| sean-k-mooney | like openfow can match only on the mac/ingress prot no problem | 13:45 |
| sean-k-mooney | well that what the tempest test is trying to do :) | 13:45 |
| sean-k-mooney | we can just use depens on to verify it | 13:45 |
| ralonsoh | one sec | 13:46 |
| ralonsoh | sean-k-mooney, https://review.opendev.org/c/openstack/neutron/+/977961 | 13:51 |
| sean-k-mooney | there is a minor ux issue by the way | 13:52 |
| sean-k-mooney | to enable the ipallcoation exteion to work in neutron | 13:52 |
| sean-k-mooney | you actully have to enabel the entirty of the segments plugin | 13:53 |
| sean-k-mooney | https://github.com/openstack/neutron/blob/master/neutron/services/segments/plugin.py#L67 | 13:53 |
| sean-k-mooney | even if you dont plan to suprpot l3 routed networks at all just unadressed ports | 13:53 |
| sean-k-mooney | its unfortunet that that is requried becuase the 2 feature are mutally exclisive | 13:53 |
| sean-k-mooney | you cant use unadressed prot on l3 routed network so having to enable both via one plugin is not ideal | 13:54 |
| ralonsoh | right, this is what is missing | 13:54 |
| sean-k-mooney | anyway i think that is what we need to do in the nova next job. but the other thing that is missign is for ml2/ovs neutron shoudl add the same extension filterign feature that was implmented for ml2/ovn | 13:55 |
| sean-k-mooney | so that neutron will not report it in the extenion list if no plugin that is enabeld reports supprot | 13:55 |
| ralonsoh | but for ovs is just a matter of adding it to the extension list in the config | 13:55 |
| ralonsoh | and enabling segments | 13:56 |
| sean-k-mooney | right but nova and tempest but cvheck teh exention list api to knwo if the cloud supprots it | 13:56 |
| sean-k-mooney | so its an api bug that a cloud that cant supprot this becuase the plugin is not enbaled reports the extions as aviabel to use | 13:56 |
| sean-k-mooney | the test was passignon the other tempest jobs because ovn didnt reprot it and it was filtered out of the list | 13:56 |
| sean-k-mooney | so the test woudl skip | 13:57 |
| ralonsoh | can you point me to this check? | 13:58 |
| sean-k-mooney | https://review.opendev.org/c/openstack/tempest/+/977930/1/tempest/api/compute/servers/test_create_server.py#164 | 13:58 |
| ralonsoh | are we retrieving the extension list from the API or reading the config | 13:58 |
| sean-k-mooney | the api | 13:59 |
| sean-k-mooney | we also check the api in nova | 13:59 |
| sean-k-mooney | ill get that link now | 13:59 |
| sean-k-mooney | so the fact that you saw that log in the neutron api server means both checks passed | 13:59 |
| ralonsoh | let me check how are we retrieving the extension list | 14:01 |
| sean-k-mooney | ok actully in nova we are just chking if the port value is populated rahter then for the exteion | 14:02 |
| sean-k-mooney | so we are just passign thout the tempest requst | 14:02 |
| sean-k-mooney | we coudl add a check | 14:02 |
| sean-k-mooney | but that woudl nto fix this | 14:02 |
| sean-k-mooney | we do check and reject request for many exteions https://github.com/openstack/nova/blob/master/nova/network/neutron.py#L1417-L1511 | 14:02 |
| sean-k-mooney | but today we are just checkign https://github.com/openstack/nova/blob/master/nova/network/neutron.py#L2563-L2564 | 14:03 |
| sean-k-mooney | it will be set to "imediat" for all normal ports defered is for l3 routned neworks and the string "none" rather then python none is adress less ports | 14:04 |
| ralonsoh | So I can confirm that the "extension list --network" command will return only the loaded extensions | 14:07 |
| sean-k-mooney | did you check that with ml2 ovs | 14:07 |
| sean-k-mooney | because the behvior is diffent unless ye fix that recently | 14:07 |
| ralonsoh | if the extension has been discarded (for example, if you don't use service_plugin=segments), the extension won't be in the list | 14:07 |
| ralonsoh | no, I didn't test that for ovs | 14:08 |
| sean-k-mooney | ok becuase when ye added the filtering it was intially only done for ml2/ovn if i recall | 14:08 |
| ralonsoh | but the behaviour is (should be) the same: if the extension is discarded, is not reported in the extension list | 14:08 |
| ralonsoh | I'll try to spawn an env with ml2/ovs | 14:08 |
| sean-k-mooney | and before you added the filtering feature the api reported all exteions it new about | 14:08 |
| sean-k-mooney | rather then were enabled | 14:08 |
| ralonsoh | slawek added a filtering for the extensions | 14:09 |
| ralonsoh | so if the extensions are not required by any plugin (actually all plugins), they are discarded | 14:09 |
| sean-k-mooney | sorry you i ment neutron | 14:09 |
| ralonsoh | let me create a VM with devstack and ml2/ovs | 14:09 |
| sean-k-mooney | rather then you sepcificly | 14:09 |
| sean-k-mooney | i shoudl have type ye :) or neutron ot make that clear | 14:09 |
| ralonsoh | that could be me, probably! | 14:10 |
| sean-k-mooney | Feb 25 09:36:20.334332 np9ec2ce0d381a4 devstack@neutron-api.service[76980]: DEBUG neutron.api.extensions [None req-ef5af19a-c63e-4033-bd14-1c11688de118 None None] Loading extension file: ip_allocation.py {{(pid=76980) _load_all_extensions_from_path /opt/stack/neutron/neutron/api/extensions.py:449}} | 14:12 |
| sean-k-mooney | Feb 25 09:36:20.335534 np9ec2ce0d381a4 devstack@neutron-api.service[76980]: DEBUG neutron.api.extensions [None req-ef5af19a-c63e-4033-bd14-1c11688de118 None None] Ext name="IP Allocation" alias="ip_allocation" description="IP allocation extension." updated="2016-06-10T23:00:00-00:00" {{(pid=76980) _check_extension /opt/stack/neutron/neutron/api/extensions.py:415}} | 14:12 |
| sean-k-mooney | Feb 25 09:36:20.335804 np9ec2ce0d381a4 devstack@neutron-api.service[76980]: INFO neutron.api.extensions [None req-ef5af19a-c63e-4033-bd14-1c11688de118 None None] Extension ip_allocation not supported by any of loaded plugins | 14:12 |
| sean-k-mooney | so i see that | 14:12 |
| sean-k-mooney | but i dont see anywhere in the logs wehere we do a neutron extions list | 14:12 |
| sean-k-mooney | that might be nice ot add to devstack ro world dump or similar | 14:12 |
| sean-k-mooney | so that we have that for debugging | 14:12 |
| ralonsoh | that would be helpful, yes, a debug log in tempest | 14:13 |
| sean-k-mooney | ralonsoh: https://paste.opendev.org/show/bUzT6XpEgq1nmzKN07tL/ | 14:36 |
| sean-k-mooney | ralonsoh: so as i mentioned the fitlering was only implented for ovn explcityly the other ml2 drier just rely on the base implemetion | 14:40 |
| sean-k-mooney | https://github.com/openstack/neutron/blob/master/neutron/plugins/ml2/plugin.py#L217 | 14:40 |
| sean-k-mooney | with that said the ip_alloctions one is not in the defualt list | 14:40 |
| opendevreview | Takashi Kajinami proposed openstack/devstack master: Drop installation of tox https://review.opendev.org/c/openstack/devstack/+/977970 | 14:40 |
| sean-k-mooney | so it should only get reported if the segments plugin is enabled | 14:41 |
| sean-k-mooney | so im not really sure why the test is failing | 14:42 |
| ralonsoh | sean-k-mooney, but this method `supported_extensions` is the only different thing between ML2/OVN and OVS | 14:52 |
| ralonsoh | the `_filter_extensions_by_mech_driver` is executed too and the extension discarding | 14:52 |
| sean-k-mooney | yep so it shoudl work on master | 14:53 |
| ralonsoh | so if service_plugin=segment is not loaded in any mech driver, the ip_allocation extension won't never be present | 14:53 |
| sean-k-mooney | right | 14:53 |
| sean-k-mooney | there is only 1 mechahim driver enabeld in this job ovs | 14:54 |
| sean-k-mooney | we have ervice_plugins = router,neutron.services.metering.metering_plugin.MeteringPlugin,placement,trunk,qos | 14:56 |
| sean-k-mooney | https://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_c82/openstack/c82065804d7e4297a6882fde834ce558/controller/logs/etc/neutron/neutron_conf.txt | 14:56 |
| sean-k-mooney | so segments is not enabeld and tempest should have skipped | 14:57 |
| opendevreview | Tanvi Nautiyal proposed openstack/tempest master: Add TempURL prefix-based access test https://review.opendev.org/c/openstack/tempest/+/977976 | 15:39 |
| frickler | if someone has a moment for this bionic cleanup that'd be nice https://review.opendev.org/c/openstack/devstack/+/977191 + 977192 | 19:02 |
| gmaan | done | 19:35 |
| gmaan | gibi: frickler thanks for reverting and merging the broken test. it was passing neutron plugin but need to check why it failed in nova-next | 19:35 |
| frickler | gmaan: see the backlog, seems neutron isn't handling the extension reporting correctly | 20:20 |
| gmaan | k, will read it | 20:20 |
| sean-k-mooney | we can make it pass by enableing the segments service plugin in nova-next but it shoudl have been skiped without that and it was not | 21:42 |
| sean-k-mooney | so ya we need to get to the bottom of that | 21:42 |
| opendevreview | Merged openstack/devstack master: Drop installation of tox https://review.opendev.org/c/openstack/devstack/+/977970 | 22:53 |
Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!