| *** esberglu has joined #openstack-watcher | 00:09 | |
| *** thorst has joined #openstack-watcher | 01:21 | |
| *** thorst has quit IRC | 01:47 | |
| *** thorst has joined #openstack-watcher | 01:47 | |
| *** thorst has quit IRC | 01:51 | |
| *** thorst has joined #openstack-watcher | 01:55 | |
| *** thorst has quit IRC | 01:56 | |
| *** thorst has joined #openstack-watcher | 01:56 | |
| *** thorst has quit IRC | 02:02 | |
| *** diga_ has quit IRC | 02:23 | |
| *** jinquan has quit IRC | 03:20 | |
| *** jinquan has joined #openstack-watcher | 03:21 | |
| *** diga has joined #openstack-watcher | 03:50 | |
| *** Kevin_Zheng has quit IRC | 05:23 | |
| *** esberglu has quit IRC | 05:41 | |
| openstackgerrit | hongzhezheng proposed openstack/python-watcherclient: Remove the blank space between the function name and the parenthesis. https://review.openstack.org/340690 | 05:54 |
|---|---|---|
| openstackgerrit | licanwei proposed openstack/watcher: Fix unittest in test_api.py https://review.openstack.org/340694 | 06:06 |
| *** vincentfrancoise has joined #openstack-watcher | 06:52 | |
| openstackgerrit | Merged openstack/python-watcherclient: Remove the blank space between the function name and the parenthesis. https://review.openstack.org/340690 | 07:02 |
| tkaczynski | hi guys | 07:14 |
| vincentfrancoise | tkaczynski: hi | 07:15 |
| tkaczynski | I have a quick question about openstack/watcher plugin architecture | 07:15 |
| tkaczynski | so if I want to deliver a new strategy (and in the future a new scoring engine as well) as a plugin, I need to prepare a stevedore plugin which essentially is similar to the watcher plugin itself and configured in local.conf file, correct? | 07:16 |
| *** dtardivel has joined #openstack-watcher | 07:16 | |
| vincentfrancoise | yes but then it depends on whether you want to enable many scoring engine at once or if you just want some flag option to decide which one to enable | 07:19 |
| openstackgerrit | David TARDIVEL proposed openstack/python-watcherclient: Add support continuously-optimization https://review.openstack.org/339026 | 07:20 |
| vincentfrancoise | also, local.conf would be in the case of a deployment, but ultimately, that's the watcher.conf file that needs to be changed | 07:20 |
| openstackgerrit | David TARDIVEL proposed openstack/watcher-dashboard: Add support continuously-optimization https://review.openstack.org/327095 | 07:21 |
| tkaczynski | ideally what I want is to have "dynamic" behavior. that means that I would like to have a single entry point for the plugin, but then have the ability to add/remove scoring engines on the fly | 07:21 |
| tkaczynski | the plugin would take care of internal dispatching, but still somehow I need to sync the db, so the new plugins will be visible in the watcher API and watcher CLI | 07:22 |
| vincentfrancoise | oh yeah I remember | 07:22 |
| tkaczynski | so what I really need is to somehow expose "db.syncScoringEngines()" method from watcher plugin to my plugin | 07:24 |
| tkaczynski | so I kind of have a dependency between plugins as well (scoring engine plugin depends on watcher plugin) | 07:24 |
| tkaczynski | any suggestions for an elegant solution? | 07:25 |
| vincentfrancoise | what is the difference you make when you say scoring engine plugin vs watcher plugin (too much plugins I'm getting confused :p) | 07:25 |
| tkaczynski | I'm talking now about general plugins which you declare in the local.conf file for your openstack / devstack. watcher is simply a plugin there, right? | 07:26 |
| tkaczynski | I plan to implement also an external plugin for Intel's TAP scoring engines, which will allow to integrate TAP (Intel product) with Watcher | 07:27 |
| tkaczynski | so I don't want it to integrate directly with watcher (so the watcher has no dependency on TAP), but deliver as a separate, completely optional plugin | 07:28 |
| tkaczynski | this is why I'm working on scoring engine abstraction in Watcher. the ultimate goal is to enable such scenarios, so everybody can implement their own scoring engines and configure as openstack plugins and watcher will recognize them | 07:29 |
| vincentfrancoise | ok I see | 07:32 |
| tkaczynski | so I believe it should just work for simple scenarios. if my additional plugin will define an entry point /class for a scoring engine, watcher during startup will recognize that and load in the same way as it does with built in strategies, goals etc. correct? | 07:34 |
| openstackgerrit | Merged openstack/watcher: Make default Planner generic to handle new action https://review.openstack.org/338921 | 07:35 |
| tkaczynski | I should be able now to define such plugin for a strategy, configure it in local.conf file (point to some Git repo) and it should work. the thing is I've never tried that, so maybe my understanding is wrong :) | 07:36 |
| vincentfrancoise | yes correct ^^ | 07:36 |
| openstackgerrit | Merged openstack/python-watcherclient: Add support continuously-optimization https://review.openstack.org/339026 | 07:36 |
| tkaczynski | this is essentially talking about that: http://docs.openstack.org/developer/watcher/dev/plugin/base-setup.html | 07:37 |
| vincentfrancoise | so I recap what I understood jsut to make sure | 07:37 |
| tkaczynski | so I want to do the same thing for scoring engines! with one exception: normally watcher loads and syncs with db all the plugins during start-up | 07:38 |
| tkaczynski | I would like to be able to force db-resync | 07:38 |
| tkaczynski | let's say from a command line | 07:38 |
| vincentfrancoise | you will write a stevode scoring engine plugin which is a an client that can call any scoring engine given that the associated (devstack-like) watcher plugin has been enabled behind the hood | 07:38 |
| vincentfrancoise | tkaczynski: ^^ I believe we can make that happen fairly easily | 07:39 |
| tkaczynski | how? | 07:40 |
| vincentfrancoise | And because it's sync, you should be able to run as many time as you want | 07:40 |
| tkaczynski | yes, but this is run as part of the watcher decision engine service | 07:40 |
| tkaczynski | from a command line I need a way to trigger it again | 07:40 |
| vincentfrancoise | yes but it shouldn't break anything if you re-run it afterwards | 07:41 |
| tkaczynski | correct. but how to re-run it from command line? | 07:41 |
| vincentfrancoise | the only issue would be be a race condition | 07:41 |
| vincentfrancoise | you will have to create a new command line then | 07:42 |
| vincentfrancoise | or add a subcommand to an existing one | 07:42 |
| vincentfrancoise | say, watcher-db-manage | 07:42 |
| vincentfrancoise | but the functionality is already there so it won't be complicated | 07:43 |
| tkaczynski | I was thinking about that, but watcher-db-manage seems quite specific to administrative tasks, like create db, purge db, etc. | 07:44 |
| vincentfrancoise | in https://github.com/openstack/watcher/blob/master/watcher/cmd/dbmanage.py#L55-L58 you can add a sync() command based on how the other commands are implemented | 07:44 |
| vincentfrancoise | or if you prefer an other command | 07:44 |
| vincentfrancoise | that's easy as well | 07:44 |
| tkaczynski | sure, but will that be easy if that other command will come in a third party plugin? | 07:45 |
| vincentfrancoise | Create a new module like watcher/cmd/dbsync.py | 07:45 |
| vincentfrancoise | and add the main() function into the setup.cfg like https://github.com/openstack/watcher/blob/master/setup.cfg#L36-L40 | 07:45 |
| tkaczynski | because it should know the watcher DB credentials etc | 07:45 |
| vincentfrancoise | the command cannot comment from your plugin | 07:46 |
| tkaczynski | yes, I know how to do that! but this additional tool should be in the third party plugin | 07:46 |
| vincentfrancoise | watcher should go and load the extra from your 3rd party plugin instead | 07:46 |
| tkaczynski | because it will be specific to TAP | 07:46 |
| tkaczynski | correct! ^^ | 07:47 |
| tkaczynski | but I still need to trigger watcher to load that and this trigger has to be from third party plugin | 07:47 |
| tkaczynski | the use case is: third party plugin will provide a scoring engine plugin (one entry point for watcher-defined entry points) and a command line tool, which will have options like add/update/remove scoring TAP engine. each call should then trigger watcher to reload scoring engines and sync DB | 07:49 |
| tkaczynski | this is what I want :) | 07:49 |
| tkaczynski | question is how to expose this db sync from watcher to third party plugin | 07:50 |
| tkaczynski | and I think it's ok to limit this sync to scoring engines for now, if you don't want that for other plugins (strategies or goals) | 07:51 |
| vincentfrancoise | (sorry daily) | 07:53 |
| tkaczynski | no worries, please take your time :) | 07:54 |
| vincentfrancoise | ok so let's say you have a new watcher command I will call watcher-sync | 07:55 |
| vincentfrancoise | this command has an initial command that does the syncing | 07:56 |
| vincentfrancoise | and you can register subcommands via 3rd party plugins | 07:56 |
| vincentfrancoise | then when the TAP subcommand is called, it can ask the parent to call the sync with the extra | 07:57 |
| vincentfrancoise | would it be enough or am I still missing something? | 08:00 |
| openstackgerrit | Merged openstack/watcher: Add new documentation section for Watcher policies rules https://review.openstack.org/338305 | 08:14 |
| tkaczynski | thinking :) | 08:18 |
| tkaczynski | it seems that this could be enough. my CLI tool (from third party plugin) would then call watcher-sync and that's it. | 08:20 |
| vincentfrancoise | yes because it will have access to the parent | 08:20 |
| tkaczynski | I don't think that the subcommands are needed | 08:21 |
| vincentfrancoise | if not then that's fine by me both ways anyway ") | 08:21 |
| vincentfrancoise | :) | 08:21 |
| tkaczynski | what about authentication? everybody could call watcher-sync? | 08:21 |
| tkaczynski | or this ". openrc admin" would be needed? | 08:21 |
| vincentfrancoise | IMHO no need for protecting the script because running it means you already have access to the machine and to other commands like create_schema | 08:22 |
| tkaczynski | sounds good. so I will prepare this watcher-sync script as part of my next changes with scoring engines. for the moment I will only do the scoring engine sync. is that ok? or better to include everything (strategies and goals)? | 08:24 |
| vincentfrancoise | the smaller for reviews the better | 08:24 |
| tkaczynski | thanks for help! | 08:25 |
| vincentfrancoise | you're welcome :) | 08:25 |
| openstackgerrit | licanwei proposed openstack/watcher: Fix unittest in test_api.py https://review.openstack.org/340694 | 08:54 |
| *** alexchadin has joined #openstack-watcher | 09:07 | |
| hanrong1 | I meet a problem. I upload a patchset for wachter-dashboard policies: https://review.openstack.org/#/c/339082/11/watcher_dashboard/api/watcher.py. In this file I wrote some code to modify horizon's settings' POLICY_FILES attribute which is defined in /horizon/openstack_dashboard/settings.py file. Although the effect is achieved, but I think there may be a better way to achieve it for these horizon's plugin projects. | 09:30 |
| *** brunograz has joined #openstack-watcher | 09:55 | |
| alexchadin | hi vincentfrancoise | 10:49 |
| alexchadin | Looks like I've found bug | 10:49 |
| alexchadin | in CLI | 10:49 |
| alexchadin | http://paste.openstack.org/show/530808/ | 10:50 |
| alexchadin | exception module doesn't really has ConnectionRefused error | 10:50 |
| alexchadin | vincentfrancoise: but it has CommunicationError https://github.com/openstack/python-watcherclient/blob/master/watcherclient/exceptions.py#L21 | 10:51 |
| *** alexchadin has quit IRC | 10:55 | |
| *** alexchadin has joined #openstack-watcher | 11:02 | |
| *** diga has quit IRC | 11:04 | |
| *** alexchadin has quit IRC | 11:26 | |
| *** alexchadin has joined #openstack-watcher | 11:26 | |
| *** alexchadin has quit IRC | 11:28 | |
| *** alexchadin has joined #openstack-watcher | 11:31 | |
| *** thorst has joined #openstack-watcher | 11:41 | |
| *** hanrong1 has left #openstack-watcher | 11:49 | |
| *** alexchadin has quit IRC | 12:15 | |
| *** alexchadin has joined #openstack-watcher | 12:16 | |
| openstackgerrit | Merged openstack/watcher-dashboard: Add support continuously-optimization https://review.openstack.org/327095 | 12:32 |
| *** hvprash has joined #openstack-watcher | 12:36 | |
| *** hvprash has quit IRC | 12:40 | |
| *** hvprash has joined #openstack-watcher | 12:41 | |
| *** hvprash has quit IRC | 12:45 | |
| *** jed56 has joined #openstack-watcher | 12:47 | |
| openstackgerrit | Alexandr Stavitskiy proposed openstack/python-watcherclient: Fix for error importing of exception class https://review.openstack.org/340935 | 13:12 |
| *** esberglu has joined #openstack-watcher | 13:18 | |
| openstackgerrit | David TARDIVEL proposed openstack/watcher-dashboard: [WIP]Add policies for API access control to watcher-dashboard https://review.openstack.org/339082 | 13:19 |
| openstackgerrit | David TARDIVEL proposed openstack/watcher-dashboard: Add policies for API access control to watcher-dashboard https://review.openstack.org/339082 | 13:20 |
| tkaczynski | hi guys | 13:20 |
| vincentfrancoise | hi | 13:21 |
| tkaczynski | I tried to use the latest watcher repo and it looks like decision engine is not starting (using latest master) | 13:21 |
| tkaczynski | https://www.irccloud.com/pastebin/0jRIyX9E/ | 13:22 |
| tkaczynski | looks like this goal is indeed missing, because when I run "watcher goal list" I get: | 13:22 |
| tkaczynski | https://www.irccloud.com/pastebin/HOCh4xoo/ | 13:22 |
| vincentfrancoise | you have to re-run a "sudo pip install -e ." inside you /opt/stack/watcher to install the airflow entrypoint | 13:23 |
| tkaczynski | ?? ^^ | 13:24 |
| tkaczynski | to me the problem is with "_" missing in the goal name. the name contains space | 13:24 |
| vincentfrancoise | true !!! I didn't see that one... | 13:25 |
| tkaczynski | how is that possible? | 13:25 |
| vincentfrancoise | I guess he never tested it properly... | 13:26 |
| tkaczynski | the goal implementation returns a name value with space... | 13:26 |
| dtardivel | tkaczynski: I will push a fix ASAP | 13:27 |
| tkaczynski | just wondering - no test failed because of that? | 13:27 |
| vincentfrancoise | we do a lookup on the goal name so if he doesn't write it correctly it won't work | 13:27 |
| vincentfrancoise | yeah that's clearly a missing test here | 13:27 |
| tkaczynski | ok, thanks guys :) please let me know when this is fixed, I'll rebase | 13:28 |
| vincentfrancoise | if I remember it correctly we test on a dummy case but I didn't write a test that would do a test matrix with all existing strategies and try to find their related goal | 13:28 |
| tkaczynski | sure, at least this time I've found a real problem, no something with my env :) | 13:29 |
| tkaczynski | *not | 13:29 |
| vincentfrancoise | yes, thanks for spotting it :) | 13:29 |
| tkaczynski | well, my environment is not working now. thanks for fixing :) | 13:29 |
| tkaczynski | regarding tests: I believe there is a integration test, which is starting watcher in real environment, right? no unit test is even needed, because this problem fails the whole decision engine service (and devstack) | 13:32 |
| vincentfrancoise | yes but it takes a lot of time to set up integration tests for strategies so if we sufficently can raise our confidence with a unit test then I would go for this option | 13:33 |
| vincentfrancoise | although I agree all strategies should definitely have some integration tests | 13:34 |
| tkaczynski | I would say just smoke test would be the minimum. devstack should at least start with watcher from master | 13:34 |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Bad goal name for airflow_optimization https://review.openstack.org/340946 | 13:34 |
| vincentfrancoise | true | 13:36 |
| vincentfrancoise | can you create a bug about it then? maybe someone other than me will accept doing this :p | 13:36 |
| tkaczynski | I don't know how to create bug :P | 13:37 |
| *** hvprash has joined #openstack-watcher | 13:42 | |
| vincentfrancoise | https://bugs.launchpad.net/watcher/+filebug | 13:42 |
| *** hvprash has quit IRC | 13:44 | |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Bad goal and strategy name for Airflow Optimization https://review.openstack.org/340946 | 13:47 |
| tkaczynski | https://bugs.launchpad.net/watcher/+bug/1602273 | 13:47 |
| openstack | Launchpad bug 1602273 in watcher "Latest Watcher taken from master should never crash" [Undecided,New] | 13:47 |
| tkaczynski | dtardivel: I see that I've opened a pandora box ;) | 13:48 |
| dtardivel | tkaczynski: yes :) | 13:49 |
| *** brunograz has quit IRC | 13:49 | |
| *** brunograz has joined #openstack-watcher | 13:50 | |
| dtardivel | We have also a pb with our unitary tests which should have detected this pb | 13:50 |
| tkaczynski | good that this wasn't released and tagged :) | 13:51 |
| vincentfrancoise | for sure ^ | 13:52 |
| *** michaelgugino has joined #openstack-watcher | 13:53 | |
| *** hvprash has joined #openstack-watcher | 13:55 | |
| *** hvprash_ has joined #openstack-watcher | 13:56 | |
| *** hvprash_ has quit IRC | 14:00 | |
| *** hvprash__ has joined #openstack-watcher | 14:00 | |
| *** hvprash has quit IRC | 14:00 | |
| *** hvprash__ has quit IRC | 14:02 | |
| *** hvprash has joined #openstack-watcher | 14:02 | |
| *** jinquan has quit IRC | 14:03 | |
| *** jinquan has joined #openstack-watcher | 14:05 | |
| *** hvprash_ has joined #openstack-watcher | 14:13 | |
| *** hvprash has quit IRC | 14:16 | |
| *** jinquan1 has joined #openstack-watcher | 14:25 | |
| *** jinquan1 has quit IRC | 14:41 | |
| *** jinquan1 has joined #openstack-watcher | 14:44 | |
| *** jed56 has quit IRC | 14:55 | |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Update unitaty tests to verify strategy and goal name https://review.openstack.org/340985 | 14:56 |
| *** alexchadin has quit IRC | 15:13 | |
| dtardivel | vi edleafe x | 15:13 |
| *** dronshaw has joined #openstack-watcher | 15:13 | |
| edleafe | dtardivel: ? | 15:18 |
| *** bauzas is now known as bauzas_notabot | 15:40 | |
| *** mestery has quit IRC | 15:49 | |
| *** mestery has joined #openstack-watcher | 15:50 | |
| *** brunograz has quit IRC | 15:59 | |
| *** jinquan1 has quit IRC | 16:03 | |
| *** bauzas_notabot is now known as bauzas | 16:15 | |
| openstackgerrit | Yatin Kumbhare proposed openstack/watcher: Add Python 3.5 classifier and venv https://review.openstack.org/341079 | 16:35 |
| *** diga has joined #openstack-watcher | 16:38 | |
| *** thorst is now known as thorst_afk | 16:42 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-watcherclient: Updated from global requirements https://review.openstack.org/340064 | 16:42 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/watcher: Updated from global requirements https://review.openstack.org/339952 | 16:44 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/watcher-dashboard: Updated from global requirements https://review.openstack.org/340176 | 16:44 |
| *** michaelgugino has quit IRC | 16:49 | |
| *** hvprash has joined #openstack-watcher | 16:58 | |
| *** hvprash_ has quit IRC | 16:58 | |
| *** hvprash_ has joined #openstack-watcher | 17:05 | |
| *** hvprash has quit IRC | 17:09 | |
| openstackgerrit | Michael Gugino proposed openstack/watcher: WIP: Implement goal_id, strategy_id and host_aggregate into Audit api https://review.openstack.org/330165 | 17:36 |
| *** thorst_afk is now known as thorst | 17:45 | |
| openstackgerrit | Michael Gugino proposed openstack/watcher: WIP: Implement goal_id, strategy_id and host_aggregate into Audit api https://review.openstack.org/330165 | 18:00 |
| *** hvprash_ has quit IRC | 18:05 | |
| *** hvprash has joined #openstack-watcher | 18:06 | |
| *** wootehfoot has joined #openstack-watcher | 18:10 | |
| *** hvprash has quit IRC | 18:12 | |
| *** hvprash has joined #openstack-watcher | 18:12 | |
| *** wootehfoot has quit IRC | 18:12 | |
| *** hvprash_ has joined #openstack-watcher | 18:13 | |
| *** hvprash__ has joined #openstack-watcher | 18:15 | |
| *** hvprash has quit IRC | 18:17 | |
| *** hvprash_ has quit IRC | 18:18 | |
| *** hvprash has joined #openstack-watcher | 18:19 | |
| *** hvprash__ has quit IRC | 18:22 | |
| *** hvprash has quit IRC | 18:28 | |
| *** hvprash has joined #openstack-watcher | 18:28 | |
| *** wootehfoot has joined #openstack-watcher | 18:31 | |
| *** hvprash has quit IRC | 19:08 | |
| *** hvprash has joined #openstack-watcher | 19:10 | |
| *** hvprash_ has joined #openstack-watcher | 19:11 | |
| *** hvprash has quit IRC | 19:12 | |
| *** hvprash has joined #openstack-watcher | 19:12 | |
| *** hvprash_ has quit IRC | 19:16 | |
| *** hvprash has quit IRC | 19:24 | |
| *** hvprash has joined #openstack-watcher | 19:25 | |
| *** hvprash_ has joined #openstack-watcher | 19:31 | |
| *** hvprash has quit IRC | 19:34 | |
| *** hvprash_ has quit IRC | 19:39 | |
| *** hvprash_ has joined #openstack-watcher | 19:39 | |
| *** hvprash_ has quit IRC | 19:40 | |
| *** hvprash_ has joined #openstack-watcher | 19:44 | |
| sballe_ | edleafe: Did acabot mentioned that we might like to go out for dinner or have lunch with the nova team next week. Is that a possibility? | 19:47 |
| edleafe | sballe_: No, he didn't | 19:48 |
| edleafe | sballe_: You buying? :) | 19:48 |
| sballe_ | Intel is paying for lunch :) as far as I know | 19:49 |
| sballe_ | so maybe lunch is better :) | 19:49 |
| openstackgerrit | Susanne Balle proposed openstack/watcher-specs: Added blueprint priorities for Newton https://review.openstack.org/341163 | 20:05 |
| openstackgerrit | Michael Gugino proposed openstack/watcher: WIP: Implement goal_id, strategy_id and host_aggregate into Audit api https://review.openstack.org/330165 | 20:08 |
| openstackgerrit | Susanne Balle proposed openstack/watcher-specs: Blueprint priorities for Newton https://review.openstack.org/341163 | 20:23 |
| *** wootehfoot has quit IRC | 20:53 | |
| *** thorst has quit IRC | 21:16 | |
| *** thorst has joined #openstack-watcher | 21:17 | |
| *** thorst_ has joined #openstack-watcher | 21:18 | |
| *** thorst__ has joined #openstack-watcher | 21:20 | |
| *** thorst has quit IRC | 21:21 | |
| *** thorst_ has quit IRC | 21:23 | |
| *** thorst__ has quit IRC | 21:24 | |
| *** dtardivel has quit IRC | 21:57 | |
| *** esberglu has quit IRC | 22:06 | |
| *** thorst has joined #openstack-watcher | 22:25 | |
| *** thorst_ has joined #openstack-watcher | 22:29 | |
| *** thorst has quit IRC | 22:30 | |
| *** thorst_ has quit IRC | 22:34 | |
| *** hvprash_ has quit IRC | 22:44 | |
| *** hvprash has joined #openstack-watcher | 22:44 | |
| *** hvprash_ has joined #openstack-watcher | 22:45 | |
| *** hvprash_ has quit IRC | 22:48 | |
| *** hvprash has quit IRC | 22:49 | |
| *** thorst has joined #openstack-watcher | 23:51 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!