| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-03-30 | |||
| 20:14:27 | xgerman | wonder what the status of soft-anti-affinity is? I am on master nd it doesn’t work | |
| 20:14:37 | xgerman | https://www.irccloud.com/pastebin/zfgU1sJu/ | |
| 20:17:03 | sdague | jaypipes: ok, most of the clarifications are good, I'll look at doing another quick edit to be consistent there. When it comes to why I think there needs to be a library, see the various quota rule sets here that were being explored - https://review.openstack.org/#/c/441203/ | |
| 20:18:22 | mriedem | xgerman: i'd talk to gibi or ildikov | |
| 20:18:35 | xgerman | thanks | |
| 20:18:49 | openstackgerrit | Anusha Unnam proposed openstack/nova master: Add description to policies in servers.py https://review.openstack.org/451966 | |
| 20:19:07 | xgerman | when I grepped my nova dir it was there so odd | |
| 20:19:42 | ildikov | xgerman: is that latest master? | |
| 20:19:52 | xgerman | pretty much | |
| 20:20:07 | xgerman | build my devstack 3-27 | |
| 20:20:44 | xgerman | or is there something I need to enable - I am run ing with default settings | |
| 20:20:47 | ildikov | hmm, I wonder whether there's any config/filter option that you need to set | |
| 20:21:12 | xgerman | yeah, me, too — had trouble finding docs at all for soft-anti-affinity | |
| 20:21:28 | ildikov | I need to look into it as it was quite a while ago when we introduced it and I cannot recall any config related change in it now | |
| 20:23:51 | melwitt | xgerman, ildikov: I have seen some usage of it in a functional test before, maybe it will help. it implies you need to add ServerGroupAntiAffinityFilter to the enabled_filters config option https://github.com/openstack/nova/blob/master/nova/tests/functional/test_server_group.py | |
| 20:25:07 | xgerman | I have it: | |
| 20:25:24 | mriedem | it's not a default filter i don't think | |
| 20:25:36 | xgerman | it is on devstack | |
| 20:25:38 | xgerman | scheduler_default_filters = RetryFilter,AvailabilityZoneFilter,RamFilter,DiskFilter,ComputeFilter,ComputeCapabilitiesFilter,ImagePropertiesFilter,ServerGroupAntiAffinityFilter,ServerGroupAffinityFilter,SameHostFilter,DifferentHostFilter | |
| 20:25:50 | mriedem | ah | |
| 20:26:00 | xgerman | wonder if I need the weights? | |
| 20:26:01 | mriedem | the weighers are all loaded up by default | |
| 20:26:13 | mriedem | weight_classes=all i think | |
| 20:26:36 | mriedem | we recently updated the api-ref for this | |
| 20:26:40 | mriedem | when creating the server group | |
| 20:27:24 | xgerman | mmh, what throws me off is that the REST call response only gives me tow options | |
| 20:29:43 | mriedem | https://developer.openstack.org/api-ref/compute/?expanded=create-server-group-detail#create-server-group | |
| 20:29:54 | mriedem | see the 'policies' attribute | |
| 20:30:05 | mriedem | are you specifying the correct policy when creating the server group? | |
| 20:32:28 | ildikov | it's also a weigher: ServerGroupSoftAntiAffinityWeigher | |
| 20:32:45 | xgerman | mriede: Yes | |
| 20:32:53 | xgerman | https://www.irccloud.com/pastebin/o38ZPMUT/ | |
| 20:32:56 | sdague | jaypipes: ok, I think I cleaned it up | |
| 20:33:07 | mriedem | xgerman: and you specified the hint when creating the server? | |
| 20:33:10 | mriedem | *servers | |
| 20:33:13 | sdague | I left 2 references in there to "quota system" to mean limits + usage | |
| 20:33:28 | xgerman | well, I can’t create a servergroup with that polciy | |
| 20:33:34 | sdague | but it should hopefully be pretty distinct from the rest of it | |
| 20:35:13 | mriedem | xgerman: yeah but you also have to specify the group hint when creating the servers | |
| 20:35:25 | mriedem | https://docs.openstack.org/mitaka/config-reference/compute/scheduler.html#servergroupaffinityfilter | |
| 20:35:28 | mriedem | it's a whole dance | |
| 20:35:32 | mriedem | sdague loves it | |
| 20:35:58 | xgerman | well, my flow is: create server-group and then create severs with hint (servergrouo) | |
| 20:36:16 | xgerman | I fail when creating a server-group | |
| 20:37:08 | openstackgerrit | Dan Smith proposed openstack/nova master: Teach HostAPI about cells https://review.openstack.org/442162 | |
| 20:37:09 | openstackgerrit | Dan Smith proposed openstack/nova master: Make scheduler target cells to get compute node instance info https://review.openstack.org/439891 | |
| 20:37:10 | mriedem | RESP BODY: {"badRequest": {"message": "Invalid input for field/attribute 0. Value: soft-anti-affinity. u'soft-anti-affinity' is not one of ['anti-affinity', 'affinity']", "c | |
| 20:37:10 | openstackgerrit | Dan Smith proposed openstack/nova master: Sort CellMappingList.get_all() for safety https://review.openstack.org/443174 | |
| 20:37:11 | mriedem | ode": 400}} | |
| 20:37:11 | openstackgerrit | Dan Smith proposed openstack/nova master: Add get_by_instance_uuids() to InstanceMappingList https://review.openstack.org/443292 | |
| 20:37:12 | openstackgerrit | Dan Smith proposed openstack/nova master: Clean up ClientRouter debt https://review.openstack.org/444487 | |
| 20:37:13 | openstackgerrit | Dan Smith proposed openstack/nova master: Make server_groups determine deleted-ness from InstanceMappingList https://review.openstack.org/443293 | |
| 20:37:14 | openstackgerrit | Dan Smith proposed openstack/nova master: Remove Mitaka-era service version check https://review.openstack.org/442861 | |
| 20:37:15 | openstackgerrit | Dan Smith proposed openstack/nova master: Add workaround to disable group policy check upcall https://review.openstack.org/442736 | |
| 20:37:29 | mriedem | something wrong with your jsonschema | |
| 20:37:37 | xgerman | ok | |
| 20:37:47 | xgerman | then I will look there | |
| 20:37:50 | oomichi | mriedem: sdague: do we still need to put _LI() etc on LOGs? I guess _LI() could be unnecessary after https://review.openstack.org/#/c/447528 | |
| 20:38:12 | ildikov | xgerman: it looks like you have only the base from here: https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/schemas/server_groups.py | |
| 20:38:28 | mriedem | https://github.com/openstack/nova/blob/master/nova/api/openstack/compute/schemas/server_groups.py#L50 | |
| 20:38:35 | sdague | oomichi: no, we don't | |
| 20:38:36 | mriedem | you need 2.15 | |
| 20:38:59 | mriedem | you're using the wrong microversion | |
| 20:39:03 | sdague | the hacking rule was removed, we just need to do the rest of the deletes | |
| 20:39:11 | xgerman | I was afraid of that | |
| 20:39:28 | mriedem | https://developer.openstack.org/api-ref/compute/?expanded=create-server-group-detail#create-server-group | |
| 20:39:39 | oomichi | sdague: oh, that is a good way. | |
| 20:39:42 | mriedem | soft-anti-affinity - servers in this group should be scheduled to different hosts if possible, but if not possible then they should still be scheduled instead of resulting in a build failure. This policy was added in microversion 2.15. | |
| 20:43:46 | mriedem | oomichi: can we customize a schema validation error to remind users to verify they are using the correct microversion? | |
| 20:44:02 | jaypipes | sdague: cool, +1 from me :) | |
| 20:44:46 | oomichi | mriedem: now users can know which microversion is specifed by users in a response header instead of response body. Isn't it enough? | |
| 20:45:27 | mriedem | wifi on this flight is terrible | |
| 20:45:46 | oomichi | oh, you are flying :) | |
| 20:46:09 | ildikov | mriedem: it usually is in my experience... | |
| 20:47:00 | mriedem | oomichi: the plane is yes; i havent gotten that power yet. | |
| 20:48:17 | xgerman | sorry, I didn’t know we were locking to a specific version in Octavia | |
| 20:48:26 | xgerman | now I know what to look for | |
| 20:48:36 | oomichi | mriedem: hehe, you need to work fast before wasting the power | |
| 20:48:56 | oomichi | mriedem: the specifed microversion is in a response header with https://github.com/openstack/nova/blob/master/nova/api/openstack/wsgi.py#L1050 | |
| 20:50:08 | mriedem | oomichi: i know but i don't think that's helpful here | |
| 20:50:25 | mriedem | in this case, the user is requesting to create a server group with a specific policy that isn't available until 2.15 | |
| 20:50:41 | mriedem | so you get a schema validation error back saying that policy value is invalid | |
| 20:51:13 | mriedem | it would be user-friendlier in my opionon to give a sort of hint or reminder in the error message reminding the user to check which microversion they are using | |
| 20:51:29 | mriedem | https://www.irccloud.com/pastebin/o38ZPMUT/ | |
| 20:51:35 | mriedem | instead of the compute jargon in ^ | |
| 20:52:52 | openstackgerrit | Ildiko Vancsa proposed openstack/nova master: Add Cinder V3 Detach calls https://review.openstack.org/438750 | |
| 20:53:09 | mriedem | "does not compute, bleep bloop, invalid error death to humans" | |
| 20:55:00 | ildikov | xgerman: sorry, I forgot about that too :( | |
| 20:56:23 | oomichi | mriedem: maybe you want to see like "... u'soft-anti-affinity' is not one of ['anti-affinity', 'affinity'] on microversion 2.14" or something | |
| 20:57:10 | melwitt | that's sounds pretty good to me | |
| 20:57:30 | mriedem | oomichi: yeah that might work | |
| 20:57:38 | mriedem | pretty damn good | |
| 20:58:03 | mriedem | ooo rough air | |
| 21:02:39 | oomichi | mriedem: I did dig some code, that could be easy I guess | |
| 21:17:01 | openstackgerrit | Ken'ichi Ohmichi proposed openstack/nova master: DNM: Add the specified microversion into error msg https://review.openstack.org/451986 | |
| 21:17:09 | oomichi | mriedem: ^^^ prototype | |
| 21:18:02 | mriedem | thanks | |
| 21:26:44 | openstackgerrit | Sujitha proposed openstack/nova master: Add BDM to InstancePayload https://review.openstack.org/448779 | |
| 21:32:54 | ildikov | mriedem: and we had the meeting today, but it wasn't the same without you :) | |
| 21:34:15 | mriedem | ildikov: you mean it went the full hour? | |