Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-05-23
21:44:36 dansmith you're welcome
21:45:36 melwitt I totes forgot I used target_cell in my patches. go me
21:46:00 melwitt I was like, why are my counts totally whack?
21:47:07 dansmith because you forgot to use target_cel and were in 1998?
21:47:22 mriedem i'm assuming pants were backward
21:47:25 mriedem *weren't
21:47:49 melwitt dansmith: does target_cell magically make objects queried from the db under it have self._context set to the cell db?
21:48:45 dansmith melwitt: if you use the yielded context to initialize them then yes
21:48:54 melwitt like with target_cell(ctxt, mapping): instance = Instance.get()
21:49:28 dansmith target_cell(ctxt, mapping) as cctxt
21:49:36 dansmith instance = Instance.get_by_uuid(cctx)
21:49:38 dansmith will work
21:50:11 melwitt dansmith: like this, the instances are queried inside of group.count_members_by_user() https://review.openstack.org/#/c/457859/7/nova/quota.py@1890
21:50:40 melwitt and I'm not sure if the instances will have the right context in that case?
21:51:07 dansmith melwitt: no, see that's the problem.. group already has a context, right?
21:51:15 dansmith the target won't do anything it that context anymore
21:52:15 melwitt okay, I see
21:52:53 dansmith melwitt: not sure that code would actually work before actually
21:52:55 dansmith with multiple cells
21:53:09 melwitt yeah, I don't think so, now that I realize what's going on here
21:53:15 dansmith because it'll start several greenthreads, but by the time they run they'll all be back to untargeted I think
21:53:31 dansmith unless you copied it first, but that's what you get with the yield now anyway
21:54:20 penick mriedem did you just make a kris kross reference?
21:54:36 mriedem penick: you need to be around more,
21:54:40 mriedem those get made nearly weekly
21:54:46 melwitt yep, that
21:55:07 penick I felt like I had to say something, so you'd know at least someone got it
21:55:28 cdent don't encourage him
21:55:29 melwitt we all know kris kross
21:56:52 penick Ah ok. I wont jump (jump) next time I see it then
21:56:52 melwitt dansmith: I think things worked by chance bc the already targeted (via set_target or something) got the instance_group from the right cell earlier in the request
21:57:07 dansmith melwitt: sounds right
22:01:24 mriedem efried is gone?
22:01:40 melwitt instance create does a get for the requested server group to check the number of members quota. it currently assumes the group is in the same cell as the instance but I think that might not be always true. ugh
22:09:36 melwitt I guess that's out of scope for my patches though
23:19:59 openstackgerrit melanie witt proposed openstack/nova master: Add FloatingIP.get_count_by_project() https://review.openstack.org/446247
23:19:59 openstackgerrit melanie witt proposed openstack/nova master: Add FixedIP.get_count_by_project() https://review.openstack.org/446246
23:20:00 openstackgerrit melanie witt proposed openstack/nova master: Add get_count_by_vm_state() to Instance object https://review.openstack.org/446244
23:20:01 openstackgerrit melanie witt proposed openstack/nova master: Remove useless quota_usage_refresh from nova-manage https://review.openstack.org/446243
23:20:01 openstackgerrit melanie witt proposed openstack/nova master: Remove 'reserved' count from used limits https://review.openstack.org/446242
23:20:01 openstackgerrit melanie witt proposed openstack/nova master: Add SecurityGroup.get_counts() https://review.openstack.org/446245
23:20:02 openstackgerrit melanie witt proposed openstack/nova master: Count networks to check quota https://review.openstack.org/446241
23:20:02 openstackgerrit melanie witt proposed openstack/nova master: Count server groups to check quota https://review.openstack.org/446240
23:20:05 openstackgerrit melanie witt proposed openstack/nova master: Add InstanceGroup.get_counts() https://review.openstack.org/457857
23:20:05 openstackgerrit melanie witt proposed openstack/nova master: Count instances to check quota https://review.openstack.org/416521
23:20:06 openstackgerrit melanie witt proposed openstack/nova master: Add check_deltas() and limit_check_project_and_user() to Quotas https://review.openstack.org/446239
23:20:07 openstackgerrit melanie witt proposed openstack/nova master: Count server group members to check quota https://review.openstack.org/457859
23:20:07 openstackgerrit melanie witt proposed openstack/nova master: Add InstanceGroup._remove_members_in_db https://review.openstack.org/457858
23:20:08 openstackgerrit melanie witt proposed openstack/nova master: Count security groups to check quota https://review.openstack.org/457860
23:20:11 openstackgerrit melanie witt proposed openstack/nova master: Count fixed ips to check quota https://review.openstack.org/457861
23:20:11 openstackgerrit melanie witt proposed openstack/nova master: Make Quotas object favor the API database https://review.openstack.org/410945
23:20:13 openstackgerrit melanie witt proposed openstack/nova master: Add online migration to move quotas to API database https://review.openstack.org/410946
23:20:15 openstackgerrit melanie witt proposed openstack/nova master: Count floating ips to check quota https://review.openstack.org/457862
23:29:30 mriedem melwitt: are the members of the group not created at the same time in the same cell?
23:29:47 mriedem could be on different hosts if an anti-affinity group, that would be the one to wonder about
23:41:54 openstackgerrit Takashi NATSUME proposed openstack/nova-specs master: Amend the "list-show-all-server-migration-types" spec https://review.openstack.org/460847
23:47:20 mriedem dansmith: is your fleetify devstack change expected to pass the live migration job?
23:47:25 mriedem i don't think that runs on devstack does it?
23:47:39 mriedem looks like it doesnt
23:47:44 dansmith mriedem: a nova change that depends on it should pass yeah,
23:47:48 dansmith it was passing all last week
23:48:12 mriedem ok i hit this http://logs.openstack.org/83/467383/2/check/gate-tempest-dsvm-multinode-live-migration-ubuntu-xenial/05a0806/console.html#_2017-05-23_21_53_09_472165
23:48:13 dansmith mriedem: there is a patch on the top of my set that youneed tho
23:48:21 mriedem oh i see
23:48:37 dansmith mriedem: https://review.openstack.org/#/c/458537/
23:49:21 mriedem ok, everything else pass on https://review.openstack.org/#/c/467382/ and down so i'll consider that a success for now
23:49:25 mriedem *passed
23:49:44 dansmith cool
23:54:22 openstack Launchpad bug 1691545 in OpenStack Compute (nova) ocata "Significant increase in DB connections with cells" [High,Fix committed] - Assigned to melanie witt (melwitt)
23:54:22 cfriesen under what circumstances would we hit https://bugs.launchpad.net/nova/+bug/1691545 ? Do you need to have multiple cells?
#openstack-nova - 2017-05-24
00:50:29 openstackgerrit OpenStack Proposal Bot proposed openstack/nova master: Updated from global requirements https://review.openstack.org/467344
00:52:33 mriedem cfriesen: you don't need to have multiple cells,
00:52:36 mriedem it wasn't about that
00:52:53 mriedem it was an issue where we thought db connections were being cached by oslo.db or sqla but they weren't
00:53:01 mriedem so every time one was created it was new and they weren't getting gc'ed
01:31:43 openstackgerrit Huan Xie proposed openstack/nova master: XenAPI use os-xenapi v2 in nova https://review.openstack.org/453493
02:35:06 melwitt mriedem: guh, instance groups and members are in the API db. I forgot
02:36:58 melwitt but the way it counts members is to find instances owned by the user that are also members of the group, and returns len(instances)
02:38:23 melwitt the problem is there's no user association with instance group members, so you can't just query the API db. have to lookup the group in the API db, then count how many instances belong to the user that are also in that group
02:39:32 melwitt so I would think if you boot an instance with some group, boot another instance with the same group that lands in a different cell, you're at the point where you have to query all cells. I dunno if it's possible for that to happen based on the available policies, which is what you were thinking about
02:40:45 openstackgerrit melanie witt proposed openstack/nova master: Add FloatingIP.get_count_by_project() https://review.openstack.org/446247
02:40:45 openstackgerrit melanie witt proposed openstack/nova master: Add FixedIP.get_count_by_project() https://review.openstack.org/446246
02:40:46 openstackgerrit melanie witt proposed openstack/nova master: Remove 'reserved' count from used limits https://review.openstack.org/446242
02:40:46 openstackgerrit melanie witt proposed openstack/nova master: Add SecurityGroup.get_counts() https://review.openstack.org/446245
02:40:46 openstackgerrit melanie witt proposed openstack/nova master: Add get_count_by_vm_state() to Instance object https://review.openstack.org/446244
02:40:47 openstackgerrit melanie witt proposed openstack/nova master: Remove useless quota_usage_refresh from nova-manage https://review.openstack.org/446243
02:40:48 openstackgerrit melanie witt proposed openstack/nova master: Count networks to check quota https://review.openstack.org/446241
02:40:50 openstackgerrit melanie witt proposed openstack/nova master: Count instances to check quota https://review.openstack.org/416521
02:40:51 openstackgerrit melanie witt proposed openstack/nova master: Count server group members to check quota https://review.openstack.org/457859
02:40:52 openstackgerrit melanie witt proposed openstack/nova master: Count security groups to check quota https://review.openstack.org/457860
02:40:54 openstackgerrit melanie witt proposed openstack/nova master: Count fixed ips to check quota https://review.openstack.org/457861
02:40:54 openstackgerrit melanie witt proposed openstack/nova master: Make Quotas object favor the API database https://review.openstack.org/410945
02:40:57 openstackgerrit melanie witt proposed openstack/nova master: Add online migration to move quotas to API database https://review.openstack.org/410946
02:40:58 openstackgerrit melanie witt proposed openstack/nova master: Count floating ips to check quota https://review.openstack.org/457862
02:51:23 mriedem melwitt: huh, not sure. we could write a tempest test or something for this which might tell us
02:51:32 mriedem and run that against dan's multi-cell devstack change
02:52:02 mriedem if you can come up with the test scenario i could help work on that in tempest
02:53:18 openstackgerrit jichenjc proposed openstack/nova master: Use plain routes list for os-migrations endpoint instead of stevedore https://review.openstack.org/461628
03:41:51 openstackgerrit OpenStack Proposal Bot proposed openstack/nova master: Updated from global requirements https://review.openstack.org/467344
04:22:07 openstackgerrit Tristan Cacqueray proposed openstack/nova master: Mask the token used to allow access to consoles https://review.openstack.org/220622

Earlier   Later