| *** thorst has joined #openstack-watcher | 00:34 | |
| *** thorst_ has joined #openstack-watcher | 00:35 | |
| *** thorst has quit IRC | 00:40 | |
| *** thorst_ has quit IRC | 01:27 | |
| *** thorst has joined #openstack-watcher | 01:27 | |
| *** thorst has quit IRC | 01:36 | |
| *** jwcroppe has joined #openstack-watcher | 01:42 | |
| openstackgerrit | junjie.huang proposed openstack/watcher: Workload balance migration strategy implementation https://review.openstack.org/292188 | 01:44 |
|---|---|---|
| *** vtech has quit IRC | 02:27 | |
| *** vtech has joined #openstack-watcher | 02:27 | |
| *** thorst has joined #openstack-watcher | 02:34 | |
| *** thorst has quit IRC | 02:41 | |
| *** jwcroppe has quit IRC | 03:33 | |
| *** jwcroppe has joined #openstack-watcher | 03:34 | |
| *** thorst has joined #openstack-watcher | 03:39 | |
| *** jwcroppe has quit IRC | 03:39 | |
| *** thorst has quit IRC | 03:46 | |
| *** jwcroppe has joined #openstack-watcher | 04:35 | |
| *** thorst has joined #openstack-watcher | 04:44 | |
| *** jwcroppe has quit IRC | 04:45 | |
| *** thorst has quit IRC | 04:52 | |
| *** jwcroppe has joined #openstack-watcher | 05:41 | |
| *** jwcroppe has quit IRC | 05:49 | |
| *** thorst has joined #openstack-watcher | 05:49 | |
| *** thorst has quit IRC | 05:55 | |
| openstackgerrit | Taylor Peoples proposed openstack/watcher-specs: WIP: Cluster model objects wrapper https://review.openstack.org/287019 | 05:58 |
| *** apoorv has joined #openstack-watcher | 06:13 | |
| *** apoorv has quit IRC | 06:35 | |
| *** jwcroppe has joined #openstack-watcher | 06:46 | |
| *** apoorv has joined #openstack-watcher | 06:49 | |
| *** jwcroppe has quit IRC | 06:53 | |
| *** thorst has joined #openstack-watcher | 06:53 | |
| *** apoorv has quit IRC | 06:59 | |
| *** thorst has quit IRC | 07:01 | |
| *** wootehfoot has quit IRC | 07:11 | |
| *** vtech has quit IRC | 07:42 | |
| *** jwcroppe has joined #openstack-watcher | 07:50 | |
| *** thorst has joined #openstack-watcher | 07:58 | |
| *** jwcroppe has quit IRC | 08:00 | |
| *** dtardivel has joined #openstack-watcher | 08:00 | |
| *** vtech has joined #openstack-watcher | 08:02 | |
| *** thorst has quit IRC | 08:06 | |
| *** apoorv has joined #openstack-watcher | 08:21 | |
| apoorv | vincent: thanks a lot you saved much effort | 08:21 |
| *** vincentfrancoise has joined #openstack-watcher | 08:29 | |
| acabot | agenda for our weekly meeting open : https://wiki.openstack.org/wiki/Watcher_Meeting_Agenda#03.2F23.2F2016 | 08:31 |
| openstackgerrit | Merged openstack/watcher-specs: Select destination filter for live migration https://review.openstack.org/291214 | 08:32 |
| openstackgerrit | Merged openstack/watcher-specs: Added doc8 for specs validation https://review.openstack.org/294051 | 08:33 |
| *** jed56 has joined #openstack-watcher | 08:44 | |
| *** jwcroppe has joined #openstack-watcher | 08:56 | |
| *** thorst has joined #openstack-watcher | 09:03 | |
| *** openstackgerrit has quit IRC | 09:03 | |
| *** openstackgerrit has joined #openstack-watcher | 09:03 | |
| *** vtech_ has joined #openstack-watcher | 09:06 | |
| *** vtech has quit IRC | 09:06 | |
| openstackgerrit | junjie.huang proposed openstack/watcher: Workload balance migration strategy implementation https://review.openstack.org/292188 | 09:11 |
| *** thorst has quit IRC | 09:11 | |
| *** thorst has joined #openstack-watcher | 10:08 | |
| *** thorst has quit IRC | 10:16 | |
| *** wootehfoot has joined #openstack-watcher | 10:22 | |
| openstackgerrit | Tomasz Kaczynski proposed openstack/watcher-specs: Add specification for Watcher Scoring Module https://review.openstack.org/289880 | 10:44 |
| *** wootehfoot has quit IRC | 10:50 | |
| *** sudipto has joined #openstack-watcher | 10:53 | |
| sudipto | Hi there. I have installed watcher and wanted to get some idea on a few operations that i could perform using it. I couldn't find documentation that would explain what user commands i could execute to let's say trigger an audit etc...can someone help pls? | 10:55 |
| sudipto | I could find the installation guides explain how to configure/install watcher...but didn't find a doc on how to use watcher as yet... | 10:59 |
| vincentfrancoise | Hi sudipto | 11:00 |
| sudipto | Hi vincentfrancoise | 11:00 |
| vincentfrancoise | The Watcher documentation is currently available at https://factory.b-com.com/www/watcher/doc/watcher/ | 11:00 |
| vincentfrancoise | If you want to get started with it | 11:01 |
| sudipto | vincentfrancoise, thanks. i am following through some usage guidelines there. | 11:01 |
| vincentfrancoise | you can have a look at https://factory.b-com.com/www/watcher/doc/watcher/deploy/user-guide.html where of Watcher CLI commands are explained | 11:01 |
| sudipto | yeah just checked. thx :) | 11:02 |
| vincentfrancoise | feel free to ask if you need help ;) | 11:03 |
| *** sudipto has quit IRC | 11:13 | |
| *** thorst has joined #openstack-watcher | 11:35 | |
| *** jinquan has joined #openstack-watcher | 11:46 | |
| *** apoorv has quit IRC | 12:35 | |
| sballe_ | Morning | 12:52 |
| jed56 | hello sballe_ | 13:00 |
| *** alexchadin has joined #openstack-watcher | 13:03 | |
| acabot | hi alexchadin | 13:04 |
| alexchadin | good day, acabot! | 13:04 |
| acabot | there are currently 24 open comments on your spec, when do you plan to add a new PS ? | 13:05 |
| alexchadin | updated spec? | 13:05 |
| acabot | a new patch set | 13:06 |
| alexchadin | I'm working on it, hope it will be done tomorrow or by friday. | 13:07 |
| acabot | great thx | 13:07 |
| alexchadin | trying to provide it in more clear and better form | 13:07 |
| alexchadin | I'll be on meeting in an hour anyway. | 13:09 |
| *** alexchadin has quit IRC | 13:10 | |
| *** jwcroppe has quit IRC | 13:12 | |
| vtech_ | Hey guys, just a quick question, we noticed that in the decision engine model you represent VM state using string - https://github.com/openstack/watcher/blob/master/watcher/decision_engine/model/vm.py#L23 | 13:28 |
| vtech_ | using enum.value | 13:28 |
| vtech_ | but in hypervisor you use enum object directly | 13:29 |
| vtech_ | https://github.com/openstack/watcher/blob/master/watcher/decision_engine/model/hypervisor.py#L25 | 13:29 |
| vtech_ | is there a reason for that? | 13:29 |
| vincentfrancoise | hi vtech_ | 13:30 |
| jed56 | hello vtech_ | 13:30 |
| jed56 | there is not reasons | 13:31 |
| vincentfrancoise | +1 ^ | 13:31 |
| jed56 | we will fix that we the blueprint object-wrapper | 13:31 |
| jed56 | sorry cluster-model-objects-wrapper | 13:31 |
| vincentfrancoise | #link https://blueprints.launchpad.net/watcher/+spec/cluster-model-objects-wrapper | 13:32 |
| vtech_ | ok, then my question is: is it 100% sure the strategy recieves a model with VMs which states are str type? | 13:32 |
| *** seanmurphy has joined #openstack-watcher | 13:33 | |
| vtech_ | I only found https://github.com/openstack/watcher/blob/master/watcher/metrics_engine/cluster_model_collector/nova.py#L71 | 13:35 |
| vtech_ | which should be string | 13:35 |
| vtech_ | well, in any case I will have to recheck the strategy for that. | 13:38 |
| jed56 | we should use string | 13:41 |
| jed56 | to be consistent | 13:41 |
| jed56 | in my opinion | 13:41 |
| jed56 | :) | 13:41 |
| jed56 | this is a error ( my in this case) to handle sometime with Enum sometime with string :) | 13:42 |
| *** lrensing has joined #openstack-watcher | 13:46 | |
| vtech_ | ok then I will try to use resource.state == resource.STATE.value for both vms and hypervisors | 13:46 |
| vtech_ | e.g. vm.state != vm_state.VMState.ACTIVE.value rather then vm.state != vm_state.VMState.ACTIVE | 13:47 |
| vtech_ | it can be a bit misleading - if the types doesn't match it will produce False using the == operator rather then raising an exception | 13:50 |
| jed56 | I agree | 13:51 |
| jed56 | do you think we should check the state of the vm in the strategies ? | 13:51 |
| jed56 | In my opinion, we can check that in the action or let nova manage that | 13:52 |
| vtech_ | a possibility of checking VM state in strategy allows strategies to adjust e.g. correct migration type, so generally I think it's a good idea to have this information there | 13:54 |
| vtech_ | e.g. as watcher so far only supports live migration (and block live migration) it only can migrate VMs in the active state. | 13:55 |
| vtech_ | and with regards to hypervisor state type, I am pretty sure it's a string in the strategy, but an enum object here https://github.com/openstack/watcher/blob/master/watcher/decision_engine/model/hypervisor.py#L25 | 13:57 |
| jed56 | In my opinion we don't want that the strategy have to manage this use cases. We should focus on the strategy itself. if the strategy says migrate this virtual machine nova have to manage handle case if possible ( enough place, constraints,ect) | 13:58 |
| acabot | time for weekly meeting | 13:59 |
| jed56 | let's go :) | 14:00 |
| jinquan | ok | 14:00 |
| *** hvprash has joined #openstack-watcher | 14:01 | |
| *** alexchadin has joined #openstack-watcher | 14:02 | |
| *** vtech_ has quit IRC | 14:03 | |
| *** vtech has joined #openstack-watcher | 14:04 | |
| *** openstack has joined #openstack-watcher | 14:21 | |
| *** aspiers has joined #openstack-watcher | 14:23 | |
| *** jed561 has quit IRC | 14:25 | |
| *** alexchadin has quit IRC | 15:04 | |
| *** jinquan has left #openstack-watcher | 15:04 | |
| *** sudipto has joined #openstack-watcher | 15:08 | |
| *** seanmurphy has quit IRC | 15:12 | |
| vtech | jed56, to continue the conversation, I think the main point of strategy is not to only say migrate VM, but you should also specify where do yo want to migrate from many different reasons. More information in cluster model within strategy allows better control (and possibly better optimization) within the strategy itself. | 15:20 |
| jed56 | I agree for the where but not for the "how" | 15:23 |
| jed56 | :) | 15:23 |
| jed56 | can you give a example ? | 15:24 |
| jed56 | n | 15:24 |
| vtech | e.g. I saw the current implementation of live migrate action use try-fail-retry approach - try to live migrate a machine which fails if there is no shared storage between src and dst hypervisors and then try again with block-migration flag. The first failure could be avoided if there is an information whether the shared storage is available. but in this case i still agree it should be handled by nova since we are still talking about live migration, | 15:35 |
| vtech | then there is a cold migration in nova which is not the same anymore. | 15:35 |
| vtech | but yes if nova can choose the best type of migration on its own I agree that would be the best solution. | 15:36 |
| vincentfrancoise | vtech: the 'auto' option is now available in Nova | 15:37 |
| vtech | doesn't mean that when you attempt to live migrate a machine which is not "live migratable" it will default to cold migration? | 15:38 |
| vtech | *does it | 15:39 |
| vincentfrancoise | yes | 15:39 |
| vincentfrancoise | the auto is only between block and non-block | 15:40 |
| vtech | exactly | 15:40 |
| vtech | that's the problem | 15:40 |
| vtech | it is still a live migration | 15:40 |
| vtech | if you have a shutdown VM you cannot move it with this operation. | 15:41 |
| vincentfrancoise | Oh ok that's why you want to know about the state | 15:41 |
| vtech | auto between block and nonblock is understandable | 15:41 |
| vincentfrancoise | I guess that by default the migration action should deal with live/non-live by itself though we should still be able to force the former or the latter from the strategy | 15:43 |
| vtech | my point is, watcher should also support cold migration to cover these cases. At least until nova starts to be smart enough to choose migration type on it's own. which would essentially meant to marge live- and cold- migration terms to one (e.g. migration). | 15:44 |
| vincentfrancoise | agree | 15:45 |
| vincentfrancoise | if you want to do it, I would be more than happy to let you do so ;) | 15:45 |
| vincentfrancoise | But at least a whishlist bug or a BP should be created to mention this | 15:46 |
| openstackgerrit | Vojtech Cima proposed openstack/watcher: Integrated consolidation strategy with watcher https://review.openstack.org/289259 | 15:46 |
| vtech | vincentfrancoise: I quite busy with some other stuff now, but I will take a look | 15:52 |
| openstackgerrit | Jeffrey Augustine proposed openstack/watcher: Added oslo.context to requirements.txt https://review.openstack.org/296559 | 15:52 |
| *** jaugustine has joined #openstack-watcher | 15:53 | |
| *** hvprash_ has joined #openstack-watcher | 16:06 | |
| *** hvprash has quit IRC | 16:06 | |
| *** hvprash has joined #openstack-watcher | 16:08 | |
| *** hvprash_ has quit IRC | 16:11 | |
| vincentfrancoise | vtech: ok no problem | 16:19 |
| *** brunograz has joined #openstack-watcher | 16:28 | |
| brunograz | Hi everyone | 16:29 |
| vincentfrancoise | Hi brunograz | 16:29 |
| brunograz | I'm getting the following error when trying to start an action-plan | 16:29 |
| brunograz | Client-side error: Couldn't apply patch '{'path': '/state', 'value': u'TRIGGERED', 'op': 'replace'}'. Reason: Invalid state: TRIGGERED format_exception /root/watcher/watcher_env/lib/python2.7/site-packages/wsme/api.py:221 | 16:29 |
| brunograz | but the action plan is in RECOMMENDED state | 16:30 |
| brunograz | could someone help me out with this? | 16:30 |
| vincentfrancoise | There was a renaming recently | 16:30 |
| vincentfrancoise | the TRIGGERED state was renamed as PENDING | 16:30 |
| vincentfrancoise | so pull the latest version of python-watcherclient and you should be good to go | 16:31 |
| *** jwcroppe has joined #openstack-watcher | 16:34 | |
| *** vincentfrancoise has left #openstack-watcher | 16:37 | |
| *** vincentfrancoise has joined #openstack-watcher | 16:37 | |
| brunograz | that worked | 16:37 |
| brunograz | thanks | 16:37 |
| *** brunograz has quit IRC | 16:58 | |
| *** vtech has quit IRC | 17:03 | |
| *** lrensing has quit IRC | 17:16 | |
| *** vtech has joined #openstack-watcher | 17:31 | |
| *** Kevin_Zheng has quit IRC | 17:32 | |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added Goal and Strategy into Watcher DB https://review.openstack.org/296633 | 17:45 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added Goal object https://review.openstack.org/296634 | 17:45 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added Goal in Strategy + Goal API reads from DB https://review.openstack.org/296635 | 17:45 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Add strategy & goal fields in audit template https://review.openstack.org/296636 | 17:45 |
| *** xenogear has joined #openstack-watcher | 17:47 | |
| *** xenogear has quit IRC | 17:47 | |
| *** vincentfrancoise has quit IRC | 17:49 | |
| *** mestery_ has joined #openstack-watcher | 18:28 | |
| *** mestery has quit IRC | 18:33 | |
| *** mestery_ is now known as mestery | 18:33 | |
| *** mestery_ has joined #openstack-watcher | 18:34 | |
| *** mestery_ has quit IRC | 18:46 | |
| *** openstack has joined #openstack-watcher | 19:07 | |
| *** openstack has joined #openstack-watcher | 19:20 | |
| *** hvprash_ has quit IRC | 19:21 | |
| sballe_ | howdy | 20:08 |
| sballe_ | please review: https://docs.google.com/presentation/d/1xzuIgmvRXiwTxv4R0al8DZoFd7u6OA6hx6aUrid7JPw/edit#slide=id.p4 | 20:08 |
| *** hvprash has joined #openstack-watcher | 20:11 | |
| *** hvprash__ has quit IRC | 20:12 | |
| *** hvprash has quit IRC | 20:18 | |
| *** openstack has joined #openstack-watcher | 20:31 | |
| *** dtardivel has quit IRC | 21:08 | |
| *** thorst has quit IRC | 21:11 | |
| *** thorst has joined #openstack-watcher | 21:12 | |
| *** thorst_ has joined #openstack-watcher | 21:14 | |
| *** thorst__ has joined #openstack-watcher | 21:16 | |
| *** thorst has quit IRC | 21:17 | |
| *** thorst_ has quit IRC | 21:18 | |
| *** thorst__ has quit IRC | 21:21 | |
| *** thorst has joined #openstack-watcher | 21:34 | |
| *** thorst has quit IRC | 21:38 | |
| *** jaugustine has quit IRC | 23:04 | |
| *** openstack has joined #openstack-watcher | 23:22 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!