| *** jwcroppe has joined #openstack-watcher | 00:22 | |
| *** zhurong has joined #openstack-watcher | 00:28 | |
| *** hanrong has left #openstack-watcher | 00:40 | |
| *** thorst_ has joined #openstack-watcher | 00:53 | |
| *** thorst_ has quit IRC | 00:58 | |
| *** thorst_ has joined #openstack-watcher | 01:17 | |
| *** hidekazu has joined #openstack-watcher | 01:30 | |
| openstackgerrit | licanwei proposed openstack/watcher: check the state of action plan https://review.openstack.org/391383 | 01:56 |
|---|---|---|
| *** thorst_ has quit IRC | 02:38 | |
| *** thorst_ has joined #openstack-watcher | 02:39 | |
| *** thorst_ has quit IRC | 02:43 | |
| openstackgerrit | licanwei proposed openstack/python-watcherclient: Add param 'goal' and 'strategy' in list() https://review.openstack.org/417790 | 03:30 |
| *** zhurong has quit IRC | 03:39 | |
| *** sferna1_ has joined #openstack-watcher | 04:00 | |
| *** diga has joined #openstack-watcher | 04:21 | |
| *** adisky_ has joined #openstack-watcher | 05:18 | |
| *** thorst_ has joined #openstack-watcher | 05:40 | |
| *** thorst_ has quit IRC | 05:44 | |
| *** absubram has quit IRC | 06:17 | |
| *** zhurong has joined #openstack-watcher | 06:23 | |
| *** thorst_ has joined #openstack-watcher | 06:40 | |
| *** Yumeng__ has joined #openstack-watcher | 06:44 | |
| *** thorst_ has quit IRC | 06:45 | |
| *** logan- has quit IRC | 07:47 | |
| *** logan- has joined #openstack-watcher | 07:56 | |
| *** vincentfrancoise has joined #openstack-watcher | 08:24 | |
| *** diga has quit IRC | 08:26 | |
| *** alexchadin has joined #openstack-watcher | 08:30 | |
| alexchadin | good morning | 08:34 |
| vincentfrancoise | alexchadin: morning | 08:37 |
| vincentfrancoise | how were your holidays? | 08:38 |
| *** jwcroppe has quit IRC | 08:41 | |
| alexchadin | vincentfrancoise: great:) 8 days in Prague and Wien | 08:41 |
| *** thorst_ has joined #openstack-watcher | 08:41 | |
| *** jwcroppe has joined #openstack-watcher | 08:42 | |
| vincentfrancoise | Wunderbar! I hope you enjoyed the beers they have in both places! | 08:42 |
| *** jwcroppe has quit IRC | 08:46 | |
| *** thorst_ has quit IRC | 08:46 | |
| alexchadin | vincentfrancoise: very tasty beer! | 08:50 |
| *** tmetsch has joined #openstack-watcher | 08:54 | |
| *** hidekazu has quit IRC | 09:11 | |
| openstackgerrit | Alexander Chadin proposed openstack/watcher-specs: Add improvements to Applier to make it parallel https://review.openstack.org/413079 | 09:17 |
| *** Yumeng__ has quit IRC | 09:21 | |
| vincentfrancoise | alexchadin: did you have a look at the code I pushed following our discussion on the planner with validate_parents() ? | 09:37 |
| alexchadin | vincentfrancoise: not yet, I manage with my patches and then will take a look at yours | 09:38 |
| vincentfrancoise | alexchadin: ok, just give me your thoughts about it once you get to it ;) | 09:38 |
| alexchadin | vincentfrancoise: of course, there will be a lot of pings :D | 09:40 |
| vincentfrancoise | alexchadin: feel frree ;) | 09:40 |
| *** dtardivel has joined #openstack-watcher | 09:44 | |
| *** zhurong has quit IRC | 09:44 | |
| openstackgerrit | Alexander Chadin proposed openstack/python-watcherclient: Add functional tests to watcherclient https://review.openstack.org/389181 | 09:51 |
| openstackgerrit | Alexander Chadin proposed openstack/python-watcherclient: Support parents field along with planner changes https://review.openstack.org/409126 | 09:51 |
| *** akuleshov has joined #openstack-watcher | 09:55 | |
| *** alexchadin has quit IRC | 09:56 | |
| *** alexchadin has joined #openstack-watcher | 09:56 | |
| akuleshov | Hi team! I've been digging through code of WorkloadBalance and WorkloadStabilization strategies and it seems that they use their own logic for choosing a host to migrate a VM to without leveraging nova scheduler. Are there any plans to use nova scheduler from watcher to get an initial list of eligible hosts for VM migration (so that availability zones are respected, for example)? | 10:03 |
| vincentfrancoise | hi akuleshov | 10:04 |
| vincentfrancoise | this is a very good question | 10:05 |
| vincentfrancoise | as of now nova does not provide any way to submit a VM candidate to the nova scheduler to see if it can be migrated to a given node | 10:07 |
| vincentfrancoise | the nova guys know about this and that's why they are introducing a placement API | 10:08 |
| vincentfrancoise | there was a spec submitted on our side explaining this problem (see https://review.openstack.org/#/c/329873/) | 10:09 |
| akuleshov | Oh, I haven't heard about that, will look it up. I was thinking about using nova scheduler's RPC API directly or maybe extending nova's REST API. Thanks for the info! Will dig a bit more. | 10:10 |
| vincentfrancoise | ah the RPC API thing :) | 10:11 |
| vincentfrancoise | this is a somewhat sensitive subject. Watcher should not use the RPC API of Nova to do this and we should only pass through their Rest API | 10:13 |
| vincentfrancoise | but you can fork/extend Watcher to do so if you want but we cannot maintain this in the main repo | 10:14 |
| *** zhurong has joined #openstack-watcher | 10:25 | |
| openstackgerrit | Alexander Chadin proposed openstack/watcher-specs: Add improvements to the planner and workflow mechanisms https://review.openstack.org/385871 | 10:30 |
| *** alexchadin has quit IRC | 10:34 | |
| *** alexchadin has joined #openstack-watcher | 10:34 | |
| *** alexchadin has quit IRC | 10:38 | |
| *** alexchadin has joined #openstack-watcher | 10:38 | |
| *** thorst_ has joined #openstack-watcher | 10:42 | |
| *** thorst_ has quit IRC | 10:47 | |
| *** zhurong has quit IRC | 11:15 | |
| *** alexchadin has quit IRC | 11:50 | |
| *** alexchadin has joined #openstack-watcher | 12:06 | |
| *** alexchadin has quit IRC | 12:07 | |
| *** alexchadin has joined #openstack-watcher | 12:07 | |
| *** alexchadin has quit IRC | 12:07 | |
| *** alexchadin has joined #openstack-watcher | 12:08 | |
| *** alexchadin has quit IRC | 12:08 | |
| *** alexchadin has joined #openstack-watcher | 12:09 | |
| *** alexchadin has quit IRC | 12:09 | |
| *** sferna1_ has quit IRC | 12:32 | |
| akuleshov | Another question: are there any plans for implementing DRS in nova scheduler itself? I mean to make nova scheduler responsible not only for the initial placement of the VM, but for dynamic VM rescheduling? The reason I'm asking this is because I'm concerned about race conditions between watcher audits and new VM placements. | 12:35 |
| *** alexchadin has joined #openstack-watcher | 12:41 | |
| *** thorst_ has joined #openstack-watcher | 12:46 | |
| vincentfrancoise | akuleshov: the question you are asking is again a matter of which project should do what | 12:51 |
| vincentfrancoise | overall, "vanilla" Watcher is only supposed to intervene after the initial placement which is handled by Nova | 12:52 |
| vincentfrancoise | however, you can implement a Nova scheduler plugin that would then call Watcher to handle all scheduling tasks if need be, although that's not our primary objective | 12:54 |
| vincentfrancoise | as of now Watcher only focuses on compute resources | 12:54 |
| vincentfrancoise | but the idea would be to extend it to storage and network resources as well | 12:55 |
| vincentfrancoise | akuleshov: I personally wasn't the one handling most of these discussions with the other teams so that's pretty much the extent of my knowledge on the matter | 12:58 |
| openstackgerrit | Matthew J Black proposed openstack/puppet-watcher: Allow python-memcache install from authtoken class https://review.openstack.org/419201 | 12:59 |
| akuleshov | My primary concern is that currently watcher is not aware of new VM placements and action plans built by watcher will most likely become outdated if new VMs are created after the audit, so I'm interested to know what are the plans for handling these situations. | 13:04 |
| akuleshov | The spec that you mentioned (Watcher should enforce the Nova policies) briefly mentions this and suggests validating each action against nova scheduler before execution, but that doesn't solve the problem of candidate host becoming an unsuitable target for VM migration because some new VMs were created on it after watcher has formed an action plan. | 13:05 |
| *** tmetsch has quit IRC | 13:08 | |
| vincentfrancoise | akuleshov: ok so for the new VMs coming in, it's not currently handled but the idea will be to play it safe and mark as superseded all the action plans that are in the RECOMMENDED state | 13:12 |
| vincentfrancoise | akuleshov: we now have the notifications handlers for nova notifications, we "just" need to enforce the rule | 13:12 |
| vincentfrancoise | what is about to be merged though is the fact that if an action plan is triggered, all the other action plans will be subsequently marked as SUPERSEDED | 13:14 |
| alexchadin | vincentfrancoise: ping | 13:15 |
| vincentfrancoise | akuleshov: as for the second question you asked | 13:15 |
| alexchadin | vincentfrancoise: https://review.openstack.org/#/c/415505/2/watcher/decision_engine/planner/weight.py line 132 | 13:16 |
| vincentfrancoise | akuleshov: the problem is kind of shared between what should Watcher do and on the other hand, what should Nova do | 13:16 |
| alexchadin | vincentfrancoise: actions are list here? | 13:17 |
| vincentfrancoise | alexchadin: pong | 13:17 |
| vincentfrancoise | alexchadin: let me have a look | 13:17 |
| vincentfrancoise | alexchadin: yes, it's a list of dict coming from solution.actions | 13:19 |
| vincentfrancoise | alexchadin: oh no sorry | 13:20 |
| alexchadin | vincentfrancoise: but this is not full list of actions, right? | 13:20 |
| vincentfrancoise | alexchadin: it's a list of objects.Action() | 13:20 |
| vincentfrancoise | that are built using solution.actions | 13:20 |
| alexchadin | vincentfrancoise: I mean, actions are separeted by weights | 13:20 |
| alexchadin | separated* | 13:20 |
| vincentfrancoise | alexchadin: at line 209 | 13:20 |
| vincentfrancoise | yes that's a limitation to simplify the implementation | 13:21 |
| vincentfrancoise | alexchadin: does the code make sense to you? | 13:23 |
| alexchadin | IMHO it is more hard to understand code than my solution | 13:24 |
| alexchadin | I will continue to read it | 13:24 |
| vincentfrancoise | alexchadin: yes but we have no coupling between the decision engine and the applier | 13:24 |
| vincentfrancoise | alexchadin: this was something I also discussed with jed56 BTW | 13:24 |
| alexchadin | vincentfrancoise: line 112 and 113 | 13:30 |
| alexchadin | vincentfrancoise: could you please describe it? | 13:31 |
| vincentfrancoise | How about: | 13:33 |
| vincentfrancoise | Split a flat list in a list of chunks of size n.e.g chunkify([0, 1, 2, 3, 4], 2) -> [[0, 1], [2, 3], [4]] | 13:33 |
| edleafe | akuleshov: Hi, I'm currently working on the placement API. The long-term plan is to provide an API that will allow you to specify an instance to be migrated and a list of potential hosts, and then return a list of those hosts that are "acceptable" for that instance | 13:49 |
| edleafe | akuleshov: "acceptable" meaning that the host satisfies not only the raw resource requirements, but also any higher stuff like affinity requirements | 13:50 |
| edleafe | akuleshov: We hope to have much of that done by the end of Pike, but nothing's guaranteed. | 13:51 |
| *** absubram has joined #openstack-watcher | 13:51 | |
| akuleshov | edleafe: Thanks for the info | 13:54 |
| vincentfrancoise | edleafe: sounds good BTW :) | 13:54 |
| edleafe | akuleshov: you're welcome | 13:55 |
| *** lamt has joined #openstack-watcher | 13:57 | |
| *** absubram has quit IRC | 14:00 | |
| *** absubram has joined #openstack-watcher | 14:00 | |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Add first alembic version for db migration https://review.openstack.org/419031 | 14:33 |
| *** alexchadin has quit IRC | 14:52 | |
| *** Yumeng has joined #openstack-watcher | 14:52 | |
| *** alexchadin has joined #openstack-watcher | 14:52 | |
| *** Yumeng has quit IRC | 14:55 | |
| *** alexchadin has quit IRC | 14:57 | |
| *** tmetsch has joined #openstack-watcher | 14:58 | |
| *** lamt has quit IRC | 15:02 | |
| *** lamt has joined #openstack-watcher | 15:03 | |
| *** absubram has quit IRC | 15:08 | |
| *** alexchadin has joined #openstack-watcher | 15:15 | |
| *** jwcroppe has joined #openstack-watcher | 15:28 | |
| *** alexchadin has quit IRC | 15:32 | |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added Monasca Helper https://review.openstack.org/348891 | 15:43 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Multi datasource support for Basic Consolidation https://review.openstack.org/414617 | 15:43 |
| *** absubram has joined #openstack-watcher | 15:59 | |
| *** michaelgugino has joined #openstack-watcher | 16:04 | |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Fix multinode tempest test failure https://review.openstack.org/419544 | 16:10 |
| *** sferna1_ has joined #openstack-watcher | 16:17 | |
| akuleshov | edleafe: Is there a blueprint for the API that you mentioned? (The long-term plan is to provide an API that will allow you to specify an instance to be migrated and a list of potential hosts...) | 16:26 |
| edleafe | akuleshov: not yet - that's still a ways off. We're focusing now on just getting the placement engine/API working and used by Nova | 16:26 |
| akuleshov | edleafe: Oh, I see. Thanks. | 16:29 |
| edleafe | akuleshov: I tried to push a change into Nova in Newton, but it was felt that it would be better to do it right in the placement API than to add a hack into the nova API | 16:30 |
| akuleshov | edleafe: Can I have a look at it anyway? | 16:32 |
| edleafe | akuleshov: sure. Spec: https://review.openstack.org/#/c/276840/ | 16:33 |
| edleafe | akuleshov: Code: https://review.openstack.org/327809 | 16:34 |
| akuleshov | edleafe: Thanks a lot! | 16:36 |
| *** akuleshov has quit IRC | 16:41 | |
| *** akuleshov1 has joined #openstack-watcher | 16:42 | |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Fix multinode tempest test failure https://review.openstack.org/419544 | 16:47 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added Monasca Helper https://review.openstack.org/348891 | 16:53 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Multi datasource support for Basic Consolidation https://review.openstack.org/414617 | 16:53 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added tempest test for workload_balance https://review.openstack.org/393356 | 16:57 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added tempest test for workload_stabilization https://review.openstack.org/393355 | 16:58 |
| *** akuleshov1 has left #openstack-watcher | 16:58 | |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Enable notification for vm task state update https://review.openstack.org/419562 | 17:04 |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Enable, on DevStack, notifications for vm task state update https://review.openstack.org/419562 | 17:05 |
| *** hvprash has quit IRC | 17:08 | |
| *** hvprash has joined #openstack-watcher | 17:10 | |
| *** vincentfrancoise has quit IRC | 17:55 | |
| *** tmetsch has quit IRC | 18:28 | |
| *** openstackgerrit has quit IRC | 18:33 | |
| *** tmetsch has joined #openstack-watcher | 19:06 | |
| *** tmetsch has quit IRC | 20:04 | |
| *** dtardivel has quit IRC | 20:14 | |
| *** wootehfoot has joined #openstack-watcher | 20:18 | |
| *** jwcroppe has quit IRC | 21:01 | |
| *** adisky_ has quit IRC | 21:19 | |
| *** thorst_ has quit IRC | 21:57 | |
| *** ss4 has joined #openstack-watcher | 22:03 | |
| *** wootehfoot has quit IRC | 22:06 | |
| *** absubram has quit IRC | 22:07 | |
| *** gabor_antal has quit IRC | 22:11 | |
| *** gabor_antal_ has joined #openstack-watcher | 22:11 | |
| *** gabor_antal_ has quit IRC | 22:23 | |
| *** gabor_antal has joined #openstack-watcher | 22:23 | |
| *** jwcroppe has joined #openstack-watcher | 22:26 | |
| *** thorst_ has joined #openstack-watcher | 22:32 | |
| *** thorst_ has quit IRC | 22:37 | |
| *** thorst_ has joined #openstack-watcher | 23:33 | |
| *** thorst_ has quit IRC | 23:37 | |
| *** openstackgerrit has joined #openstack-watcher | 23:43 | |
| openstackgerrit | Hidekazu Nakamura proposed openstack/watcher: Add period input parameter to basic strategy https://review.openstack.org/416432 | 23:43 |
| *** lamt has quit IRC | 23:55 | |
| *** ss4 has quit IRC | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!