| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-01-12 | |||
| 17:58:21 | mriedem | devstack starts n-cpu before placement | |
| 17:58:33 | armax | mriedem: I see | |
| 17:58:35 | mriedem | there is placement in this job http://logs.openstack.org/08/418108/4/gate/gate-tempest-dsvm-neutron-src-neutron-lib-ubuntu-xenial/dafcfca/logs/apache/placement-api.txt.gz | |
| 17:58:40 | mriedem | it's on by default in all jobs in master | |
| 17:58:45 | mriedem | it's just the startup order of the services | |
| 17:58:58 | dansmith | oh it's not in a screen, right | |
| 17:59:14 | mriedem | yeah | |
| 17:59:19 | dansmith | didn't see the log, but forgot it was under apache | |
| 17:59:21 | mriedem | armax: what i did notice yesterday was this http://logs.openstack.org/08/418108/4/gate/gate-tempest-dsvm-neutron-src-neutron-lib-ubuntu-xenial/dafcfca/logs/screen-n-cpu.txt.gz?level=TRACE#_2017-01-12_17_05_25_623 | |
| 17:59:36 | dansmith | mriedem: but there are warnings much later in the run about not being able to contact placement | |
| 17:59:36 | mriedem | armax: ^ has started showing up a lot, it's not new either apparently | |
| 17:59:56 | armax | um | |
| 18:00:10 | dansmith | I guess it's only a minute later | |
| 18:00:37 | mriedem | http://logstash.openstack.org/#dashboard/file/logstash.json?query=message%3A%5C%22NeutronAdminCredentialConfigurationInvalid%3A%20Networking%20client%20is%20experiencing%20an%20unauthorized%20exception.%5C%22%20AND%20message%3A%5C%22deallocate_for_instance%5C%22%20AND%20tags%3A%5C%22screen-n-cpu.txt%5C%22%20AND%20voting%3A1&from=7d | |
| 18:01:53 | mriedem | http://logs.openstack.org/08/418108/4/gate/gate-tempest-dsvm-neutron-src-neutron-lib-ubuntu-xenial/dafcfca/logs/screen-n-cpu.txt.gz?level=TRACE#_2017-01-12_17_05_25_061 | |
| 18:01:57 | openstackgerrit | John Garbutt proposed openstack/nova: Pass objects into _update_ports_for_instance https://review.openstack.org/418513 | |
| 18:01:57 | openstackgerrit | John Garbutt proposed openstack/nova: Extract preallocate_for_instance https://review.openstack.org/418876 | |
| 18:01:58 | openstackgerrit | John Garbutt proposed openstack/nova: Stop passing network when you only need dns_domain https://review.openstack.org/419577 | |
| 18:02:00 | mriedem | Neutron client was not able to generate a valid admin token, please verify Neutron admin credential located in nova.conf | |
| 18:02:18 | armax | mriedem: let me look into that | |
| 18:02:24 | openstackgerrit | Davanum Srinivas (dims) proposed openstack/nova: [WIP][py35] Trying to get cinder+rally working https://review.openstack.org/419476 | |
| 18:02:58 | mriedem | armax: ok i'll report a bug for it too just to track it | |
| 18:03:05 | armax | mriedem: cool | |
| 18:06:05 | openstack | Launchpad bug 1656056 in OpenStack Compute (nova) "NeutronAdminCredentialConfigurationInvalid all over n-cpu logs in successful gate runs during network teardown" [High,Confirmed] | |
| 18:06:05 | mriedem | https://bugs.launchpad.net/nova/+bug/1656056 | |
| 18:08:37 | mriedem | armax: https://review.openstack.org/#/q/b8a05333ddc4e248e18080750e3cfa9cbedbca53 ? | |
| 18:10:07 | mriedem | nvm that's not released yet | |
| 18:13:27 | mriedem | could also be this https://review.openstack.org/#/c/407147/ | |
| 18:14:39 | openstackgerrit | Rafael Folco proposed openstack/nova: DELETE all inventory for a resource provider https://review.openstack.org/416669 | |
| 18:18:46 | openstackgerrit | John Garbutt proposed openstack/nova: WIP: introduce preallocate_for_instance method. https://review.openstack.org/418881 | |
| 18:18:46 | openstackgerrit | John Garbutt proposed openstack/nova: WIP: Call preallocate_for_instance before scheduler https://review.openstack.org/413124 | |
| 18:30:53 | openstackgerrit | Dan Smith proposed openstack/nova: WIP: Make api_samples tests use simple cell environment https://review.openstack.org/410473 | |
| 18:30:57 | openstackgerrit | Dan Smith proposed openstack/nova: WIP Move instance creation to conductor https://review.openstack.org/319379 | |
| 18:33:45 | sean-k-mooney | mriedem: regarding the os-vif 1.4 release are the only failing test in https://review.openstack.org/#/c/385061/7 or are there others | |
| 18:34:07 | mriedem | sean-k-mooney: i'm trying to fix it here https://review.openstack.org/#/c/419558/ | |
| 18:34:18 | mriedem | but that's failing with something different now: http://logs.openstack.org/21/418421/5/check/gate-cross-nova-python27-db-ubuntu-xenial/5b2478f/console.html#_2017-01-12_17_50_19_015480 | |
| 18:34:18 | sean-k-mooney | if its just effecting https://review.openstack.org/#/c/385061/7 that is because the upperconstraits is not allowing 1.4 to be used | |
| 18:34:45 | sean-k-mooney | looking at you fix | |
| 18:35:58 | mriedem | sean-k-mooney: there is some earlier scrollback conversation between myself and jaypipes where i don't really understand what's going on in the original failure, i.e. we're getting a 1.1 VIFHostUser object from os-vif but it doesn't have the vif_name field set in the primitive | |
| 18:36:07 | mriedem | which doesn't make any sense to me | |
| 18:36:58 | sean-k-mooney | that is strange. does it only get added if it is used? | |
| 18:37:24 | sean-k-mooney | we are not currently using the vif_name field as we havent gone to 1.4 yet | |
| 18:37:31 | mriedem | sean-k-mooney: ah yeah | |
| 18:37:40 | mriedem | getattr(self, name)) | |
| 18:37:40 | mriedem | primitive[name] = field.to_primitive(self, name, | |
| 18:37:40 | mriedem | if self.obj_attr_is_set(name): | |
| 18:39:49 | sean-k-mooney | mriedem: the issue is even with the version bump in nova in that change the upperconstrains is downgrading to 1.3.0 | |
| 18:39:51 | sean-k-mooney | http://logs.openstack.org/61/385061/7/check/gate-nova-python27-db-ubuntu-xenial/3f6ab1c/console.html.gz#_2016-12-29_09_57_59_936076 | |
| 18:40:11 | sean-k-mooney | so untill the upper-constraints patch merges the tests wont pass | |
| 18:40:39 | sean-k-mooney | the upper-constratins is downgrading below the minium set in nova's requirements.txt | |
| 18:41:22 | sean-k-mooney | if we merge teh upperconstraints change the test should pass then we can raise the minium in the global requirements | |
| 18:41:36 | mriedem | sean-k-mooney: i really thought the u-c and g-r changes would be in the same patch | |
| 18:41:50 | sean-k-mooney | the were in my patch | |
| 18:42:26 | sean-k-mooney | but i abandoned it becase francesco had one open too | |
| 18:42:50 | sean-k-mooney | his just raised the gloabl requirements patch and there was a seperate patch for upper constraints | |
| 18:44:01 | sean-k-mooney | mriedem: global reques https://review.openstack.org/#/c/415660/ upper constraitns https://review.openstack.org/#/c/418421/ | |
| 18:44:23 | mriedem | yeah i know, which now depends on the nova fix | |
| 18:44:32 | mriedem | which fails http://logs.openstack.org/21/418421/5/check/gate-cross-nova-python27-db-ubuntu-xenial/5b2478f/console.html#_2017-01-12_17_50_19_015480 | |
| 18:44:40 | mriedem | oslo_versionedobjects.exception.ObjectActionError: Object action obj_make_compatible failed because: No rule for port_profile | |
| 18:45:46 | mriedem | i'm not sure where ^ is coming from | |
| 18:45:52 | mriedem | i assume it's VIFBase | |
| 18:46:20 | sean-k-mooney | neither am i i dont recall any changes related to the port_profile | |
| 18:46:44 | mriedem | i could just be i need to pass a manifest in for the backport conversion | |
| 18:46:55 | sean-k-mooney | i still think merging https://review.openstack.org/#/c/418421/ would resolve the conflict but ill dig a little deeper | |
| 18:46:58 | dansmith | mriedem: see that test example I pointed you to? | |
| 18:47:06 | mriedem | dansmith: i couldn't really make heads or tails of that | |
| 18:47:16 | dansmith | mriedem: it passes in a manifest of what it wants the system to think the remote looks like | |
| 18:47:24 | mriedem | sean-k-mooney: it's a chicken and egg, we can't merge https://review.openstack.org/#/c/418421/ | |
| 18:47:30 | mriedem | because the nova unit tests are failing in that chnage | |
| 18:47:44 | dansmith | mriedem: in that test it mocks get_versions or whatever, but there are others in there that just pass a manifest with one or two object versions | |
| 18:49:53 | sean-k-mooney | your right hum ok im just trying to figure out what is different between them. | |
| 18:50:24 | sean-k-mooney | ah the 'vif_name': u'nicdc065497-3c' | |
| 18:50:29 | mriedem | sean-k-mooney: yeah :) | |
| 18:50:49 | mriedem | it's in the expected object from the nova test, but not in the actual object from the os-vif side | |
| 18:50:57 | mriedem | which is weird because it's version on the object is 1.1 | |
| 18:51:02 | mriedem | which was the version when vif_name was added to VIFHostUser | |
| 18:51:09 | mriedem | oh but you said, | |
| 18:51:13 | sean-k-mooney | the expect result is incorrect as before os-vif 1.4 that field did not exist | |
| 18:51:15 | mriedem | we don't use that yet? | |
| 18:51:23 | mriedem | sean-k-mooney: yeah that's another thing i noted earlier | |
| 18:51:26 | mriedem | and removed in my fix patch, | |
| 18:51:29 | mriedem | but maybe that's just the fix? | |
| 18:52:05 | mriedem | i'll try that out as i have a recreate locally | |
| 18:52:15 | sean-k-mooney | https://review.openstack.org/#/c/410737/2 is the change that should add the vif_name to the unit test | |
| 18:52:35 | mriedem | farq | |
| 18:52:40 | mriedem | ok, well just removing vif_name makes it work | |
| 18:52:44 | sean-k-mooney | yes | |
| 18:52:44 | mriedem | so i'll cleanup my patch | |
| 18:53:17 | sean-k-mooney | so if you remove it now it can be reintoduce in https://review.openstack.org/#/c/410737/2 when the filed actully exists in the os-vif object | |
| 18:54:51 | sean-k-mooney | this is related to a previous conversation i had with danile berange about the fact that oslo versioned objects allows you to set filds in objecst that are not in the defined set of fields | |
| 18:56:25 | mriedem | yeah...i remember that | |
| 18:56:32 | mriedem | we use it to our advantage at times in nova, | |
| 18:56:51 | mriedem | for fields we want to set in the nova-net layer for example and then those are used in the nova-api layer, but aren't persisted in the nova db | |
| 18:57:14 | mriedem | i think i had mentioned at some point about changing ovo to add a flag to be strict about which attributes can be set | |
| 18:57:43 | openstackgerrit | Matt Riedemann proposed openstack/nova: Make unit tests work with os-vif 1.4.0 https://review.openstack.org/419558 | |
| 18:57:52 | sean-k-mooney | yes you had. by the way i just book my ptg flights/hotel. one thing i would like to add to os-vif is a dsvm gate to increase the level of testing. | |
| 18:58:20 | mriedem | want to throw that in https://etherpad.openstack.org/p/nova-ptg-pike ? | |
| 18:58:29 | sean-k-mooney | im not sure it would have caught this but it would be better to have some intergration testing of os-vif | |
| 18:59:08 | sean-k-mooney | am sure. should i add a generic os-vif section or just os-vif gateing | |
| 18:59:23 | sean-k-mooney | there is also the topic of neutron intergration | |