Earlier  
Later
New Search
Posted Nick Remark
#openstack-nova - 2017-12-11
17:19:05 dtantsur I was thinking about ^^^ as well
17:20:55 mriedem driver.rebalances_nodes?
17:21:08 mriedem that's specific to the problem here
17:21:12 dtantsur sounds good
17:21:14 mriedem doesn't talk about migration
17:21:24 mriedem and is obviously false for everything else
17:21:47 dansmith let's hope no other drivers try to use this to implement HA
17:22:17 dtantsur people never take a random feature and try to misuse it, right?
17:22:35 dansmith never
17:28:30 openstackgerrit Stephen Finucane proposed openstack/nova master: Handle images with no data https://review.openstack.org/526329
17:28:30 openstackgerrit Stephen Finucane proposed openstack/nova master: tests: Use correct response type in tests https://review.openstack.org/527182
17:28:40 sean-k-mooney dtantsur: i mean if i can observer a specific semantic behavior or side effect of a pulic api/config option that is not documented anyware it was totally part of the public api contract that i can build my feature on correct?
17:29:20 dtantsur good plan, go for it
17:29:58 openstackgerrit Chris Dent proposed openstack/nova master: placement: skip authentication on root URI https://review.openstack.org/522002
17:33:03 stephenfin mriedem: I'd to rework that patch. Turns out the glance API docs are wrong. Would appreciate a second look at it https://review.openstack.org/526329
17:33:41 mriedem cdent: https://review.openstack.org/#/c/521640/
17:34:11 cdent ack
17:38:46 sean-k-mooney stephenfin: you have a seperate bug open for the itorator to list conversion on line 179 of glance.py correct?
17:43:11 openstackgerrit Merged openstack/nova stable/pike: placement: avoid returning duplicated alloc_reqs when no sharing rp https://review.openstack.org/508885
17:55:58 cdent mriedem: is your opinion we should go ahead with the want_version thing inside the serialization loop for the last-modified stuff. I read your first comment as maybe we should consider, and then as you progressed is became more certain. So I’ll take it as “doooo iiiiit” unless you feel like saying something to the contrary
18:28:44 mriedem cdent: i think we should probably do that for at least the requests that can generate a lot of results in a big cloud
18:28:53 mriedem so GET /traits and GET /resource_providers
18:29:09 mriedem GET /traits will always be at least reasonable size because of the library
18:29:26 cdent mriedem: i’ve done it for allocations, traits, resource classes, resource providers, but not inventories
18:29:32 dtantsur mriedem: re that patch: should I keep the functional tests changes, given that the behavior is no longer the default?
18:30:30 mriedem cdent: allocations are per provider or consumer right? so i didn't think that one was as big a deal
18:30:38 mriedem but if you're going to establish a pattern...
18:30:52 mriedem i just tend to think of allocations like inventories - they are mapped to a single provider
18:31:00 cdent mriedem: I was imagining something other than libvirt, where maybe the compute-mode is ginormous
18:31:05 mriedem dtantsur: i probably wouldn't
18:31:08 cdent or a shared disk
18:31:26 dtantsur k
18:31:36 mriedem cdent: or vmware :)
18:31:42 mriedem *vcenter
18:32:06 cdent i was mostly thinking of my.petapbytediskfarm.rp-5
18:32:32 mriedem but then the inventory is per resource class per provider isn't it?
18:32:46 mriedem so you'd only have a lot of inventory entries for a provider if you have a lot of resource classes on that provider
18:34:28 _ix Hello friends. I'm having some trouble getting a compute node to show up in the hosts list.
18:34:40 cdent mriedem: right, which isn’t predicted, so that one is left off
18:35:07 cdent but when it turns into a problem, is fixable
18:35:31 mriedem stephenfin: question in https://review.openstack.org/#/c/526329/
18:35:41 mriedem _ix: you have to run the discover_hosts CLI
18:35:47 mriedem nova-manage cell_v2 discover_hosts --verbose
18:35:52 _ix I have. It says there aren't any unmapped hosts.
18:36:05 mriedem what does nova service-list show?
18:36:11 mriedem does the entry show up there?
18:36:44 _ix For nova compute? Yes.
18:37:08 _ix This is just 1/5 of the hosts not showing up.
18:37:20 _ix But, I might be misleading a bit. I see there's an rpc problem on the host.
18:37:31 mriedem what does nova hypervisor-list show?
18:37:40 mriedem nova host-list is deprecated since pike btw
18:37:53 smatzek The Trove gates started failing on Friday or Saturday, right about the time the new attach flows landed. In particular they are failing when resizing or rebuilding instances with attached volumes. Does Nova have gate tests that do this?
18:38:11 openstackgerrit Dmitry Tantsur proposed openstack/nova master: Re-use existing ComputeNode on ironic rebalance https://review.openstack.org/508555
18:38:14 dtantsur mriedem: I hope I did not mess up anything ^^^
18:38:17 mriedem smatzek: yes for resize with volume attached - pretty sure anyway
18:38:24 mriedem smatzek: not sure about rebuild with a volume attached (in tempest anyway)
18:38:51 _ix mriedem: I generally use the openstack client. Is `openstack host list` also deprecated?
18:39:49 mriedem smatzek: test_resize_volume_backed_server_confirm
18:39:51 mriedem in tempest
18:40:07 smatzek mriedem: resize is failing like this, which in looking at the code means that connection_info is missing the driver type, and in looking at the previous log entries I can see that connection_info is indeed missing it: http://logs.openstack.org/39/524539/1/check/legacy-trove-functional-dsvm-mysql/3ef681d/logs/screen-n-cpu.txt.gz#_Dec_11_03_03_28_755007
18:40:43 mriedem _ix: i deprecated nova host-list and didn't update openstack client so it's probably not deprecated
18:41:25 _ix Cool. I see that the host is disabled in the `nova hypervisor-list` Thanks for the help. Odd.
18:41:44 mriedem _ix: starting in pike,
18:41:46 smatzek rebuild, which I haven't dug as deep into today, is failing a bit more nuanced way. The mount point it is getting from either cinder or nova is giving /dev/na, as in "not applicable" which is a default for some fields in Cinder.
18:41:55 mriedem nova-compute will auto-disable the host if it's had 10 failed build attempts
18:41:58 mriedem in a row
18:42:57 _ix Oh, hey. That's a fine tip. Thanks again! I'm sure that's what's going on here.
18:42:57 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: CLI for allocations https://review.openstack.org/457534
18:42:58 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: CLI for usages https://review.openstack.org/457535
18:42:58 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: CLI for aggregates (v1.1) https://review.openstack.org/505643
18:42:59 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: CLI for resource classes (v1.2) https://review.openstack.org/511182
18:42:59 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: RP list: member_of and resources parameters (v1.3, v1.4) https://review.openstack.org/511183
18:43:00 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: RP delete inventories (v1.5) https://review.openstack.org/514642
18:43:00 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: CLI for traits (v1.6) https://review.openstack.org/514643
18:43:01 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: Resource class set (v1.7) https://review.openstack.org/514644
18:43:01 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: Usages per project and user (v1.8, v1.9) https://review.openstack.org/514646
18:43:02 openstackgerrit Andrey Volkov proposed openstack/osc-placement master: CLI allocation candidates (v1.10) https://review.openstack.org/514647
18:43:15 _ix Also, a side note. That's a super handy feature from my previous deployments.
18:43:20 mriedem _ix: there is a config option if you want to change that
18:43:29 mriedem but yeah, ^ is why it's on by default
18:43:30 _ix I don't. I think that's great.
18:43:35 mriedem dansmith: ^ high praise
18:44:11 smatzek mriedem: the tempest test test_resize_volume_backed_server_confirm is doing resize with a volume backed / BFV instance. The trove test is different, the instance has ephemeral disk and a volume attached.
18:44:42 mriedem smatzek: hmm http://logs.openstack.org/39/524539/1/check/legacy-trove-functional-dsvm-mysql/3ef681d/logs/screen-n-cpu.txt.gz#_Dec_11_03_03_28_499795
18:45:02 mriedem {"attachment": {"status": "reserved", "detached_at": "", "connection_info": {}, "attached_at": "", "attach_mode": null, "instance": "85cdb482-63a5-487a-b103-95b9383ffcc7", "volume_id": "b3c82217-38f4-4c1f-b0f0-486a4fda6dda", "id": "848654a4-9647-46b8-9485-b3647fe87cfb"}}
18:45:03 _ix mriedem: Can you offer any conjecture about kernel versions, etc? We're on Centos 7.4, and upgraded to kernel 4.14. It looks like the rpcbind.socket is failing us.
18:45:13 mriedem _ix: nope
18:45:39 mriedem smatzek: that empty connection_info is presumably the problem
18:45:49 mriedem because nova pulls the driver_volume_type out of the connection_info on the attachment
18:46:00 mriedem smatzek: although,
18:46:11 mriedem i think that's normal for a 'reserved' attachment maybe? we'd have to ask jgriffith
18:46:26 mriedem until nova updates the attachment with the host connector, the attachment doesn't have a connection_info
18:46:39 _ix Will the openstack client one day be the only client?
18:46:48 smatzek yep, I came to the same conclusion, that the empty connection_info coming back from Cinder is part of the issue, but was drawing a blank on if it was suppose to be empty there or not.
18:47:06 ildikov mriedem: if it's only reserved connection_info is empty
18:47:18 _ix (That is, the openstack cli client)
18:47:26 mriedem _ix: at this rate i doubt it
18:47:36 _ix That's a sad story.
18:47:38 mriedem _ix: been talking about that for years now
18:47:43 smatzek To my knowledge Trove hasn't had changes here and it just started about when the Nova code landed so I thought I'd check. Also, it's not "mainline". Regular attach works and only resize and rebuild are having issues afaik.

Earlier   Later