| *** openstackgerrit has quit IRC | 02:36 | |
| *** openstackgerrit has joined #openstack-watcher | 02:42 | |
| *** openstackgerrit has quit IRC | 02:56 | |
| *** thorst_ has joined #openstack-watcher | 03:07 | |
| *** openstackgerrit has joined #openstack-watcher | 03:18 | |
| *** thorst_ has quit IRC | 03:38 | |
| *** thorst_ has joined #openstack-watcher | 03:39 | |
| *** thorst_ has quit IRC | 03:48 | |
| *** vtech has quit IRC | 04:26 | |
| *** thorst_ has joined #openstack-watcher | 04:45 | |
| *** thorst_ has quit IRC | 04:52 | |
| *** thorst_ has joined #openstack-watcher | 05:50 | |
| *** thorst_ has quit IRC | 05:57 | |
| *** ecelik has joined #openstack-watcher | 06:08 | |
| *** thorst_ has joined #openstack-watcher | 06:55 | |
| *** thorst_ has quit IRC | 07:02 | |
| *** vtech has joined #openstack-watcher | 07:10 | |
| *** vincentfrancoise has joined #openstack-watcher | 07:27 | |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Update Watcher documentation https://review.openstack.org/320493 | 07:40 |
|---|---|---|
| *** alexchadin has joined #openstack-watcher | 07:54 | |
| alexchadin | vincentfrancoise: good morning | 07:54 |
| vincentfrancoise | alexchadin: morning | 07:54 |
| alexchadin | vincentfrancoise: I tested algorithm on bare metal | 07:55 |
| vincentfrancoise | sweet! How is it performing? | 07:56 |
| alexchadin | 92% of overall time is on ceilometer queries | 07:56 |
| vincentfrancoise | lol | 07:56 |
| alexchadin | Disgusting | 07:56 |
| alexchadin | Ceilometer is really slow | 07:57 |
| alexchadin | But I didn't optimize Mongo and so on | 07:57 |
| alexchadin | Our senior engineer says it will help to get 60-70%, no more | 07:58 |
| vincentfrancoise | So I guess we really have to test monasca/ceilosca | 07:58 |
| *** thorst_ has joined #openstack-watcher | 08:00 | |
| alexchadin | vincentfrancoise: Do you use devstack to develop watcher? | 08:02 |
| *** thorst_ has quit IRC | 08:07 | |
| openstackgerrit | David TARDIVEL proposed openstack/python-watcherclient: Use goal name as strategy list filter https://review.openstack.org/321999 | 08:08 |
| *** alexchadin has quit IRC | 08:12 | |
| *** alexchadin has joined #openstack-watcher | 08:14 | |
| openstackgerrit | Vincent Françoise proposed openstack/python-watcherclient: Update Watcher CLI documentation https://review.openstack.org/320451 | 08:19 |
| vincentfrancoise | alexchadin: I started reviewing your code on periodic audits | 08:39 |
| alexchadin | vincentfrancoise: good! I'm already reviewing yours | 08:40 |
| vincentfrancoise | I have some questions on how you did it with the next_launch field | 08:40 |
| alexchadin | vincentfrancoise: feel free to ask | 08:40 |
| vincentfrancoise | in https://review.openstack.org/#/c/321411/1/watcher/decision_engine/audit/default.py | 08:40 |
| vincentfrancoise | You are actually making requests to the DB everytime to execute the jobs | 08:41 |
| vincentfrancoise | So it feels quite computationally expensive | 08:42 |
| alexchadin | 61 line? | 08:42 |
| vincentfrancoise | yep | 08:42 |
| alexchadin | but launch_audits_periodically function is called once per minute | 08:43 |
| vincentfrancoise | I would better see an in-memory catalogue of jobs that is loaded at once | 08:43 |
| vincentfrancoise | ah | 08:43 |
| alexchadin | we can specify frequency of calling this function | 08:44 |
| alexchadin | in conf file | 08:44 |
| vincentfrancoise | you could then wrap the execute() in a lambda that sleeps before executing | 08:45 |
| vincentfrancoise | What bugs me is the next launch field in DB | 08:45 |
| vincentfrancoise | IMHO, this bit of information should not be stored | 08:46 |
| vincentfrancoise | But if you don't, many problems arise | 08:46 |
| vincentfrancoise | because you have to keep in-sync the an in-memory catalogue of jobs with their database counterpart | 08:48 |
| alexchadin | hm | 08:51 |
| vincentfrancoise | I was thinking about the case where the service is restarted | 08:52 |
| vincentfrancoise | the next_launch date could be in the past | 08:53 |
| vincentfrancoise | and also in the case where someone cancels the periodic audit | 08:53 |
| vincentfrancoise | we would need to notify the job to stop | 08:53 |
| *** alexchadin has quit IRC | 08:54 | |
| *** alexchadin has joined #openstack-watcher | 08:55 | |
| alexchadin | I'm here, sorry | 08:55 |
| *** ecelik has quit IRC | 08:56 | |
| alexchadin | we would need to notify the job to stop - is it your last message? | 08:56 |
| vincentfrancoise | yep | 08:56 |
| openstackgerrit | Edwin Zhai proposed openstack/watcher: Enable strategy parameters https://review.openstack.org/297556 | 08:57 |
| alexchadin | if someone cancels the periodic audit | 08:57 |
| alexchadin | function just won't be called again | 08:58 |
| vincentfrancoise | gzhai2: Hi, FYI I started reviewing your code but I didn't finish yet | 08:58 |
| alexchadin | and next_launch field won't be updated | 08:59 |
| alexchadin | until we activate periodic calls again | 09:00 |
| vincentfrancoise | not sure I see where the job doesn't get called | 09:02 |
| vincentfrancoise | in launch_audits_periodically() | 09:02 |
| alexchadin | periodic_task.periodic_task | 09:02 |
| vincentfrancoise | you don't verify the state of the audit | 09:02 |
| vincentfrancoise | So how would the periodic task know if you can cancelled it or not ? | 09:04 |
| alexchadin | Ah | 09:04 |
| alexchadin | sorry | 09:04 |
| *** thorst_ has joined #openstack-watcher | 09:05 | |
| alexchadin | I got what you mean | 09:05 |
| alexchadin | We were talking of different things:) | 09:05 |
| alexchadin | Okay, I will add verifying of cancelled audits | 09:05 |
| vincentfrancoise | cool :) | 09:06 |
| vincentfrancoise | for the next_launch field | 09:07 |
| vincentfrancoise | can you do something like self.executor.submit(lamdba: time.sleep(DELTA);self.execute, audit.uuid, context) | 09:09 |
| vincentfrancoise | where DELTA is kind of your next launch | 09:10 |
| vincentfrancoise | ? | 09:10 |
| alexchadin | let me think, pls | 09:11 |
| alexchadin | okay | 09:12 |
| *** thorst_ has quit IRC | 09:12 | |
| alexchadin | you want to rid of next_launch, right? | 09:13 |
| vincentfrancoise | yes | 09:13 |
| alexchadin | and just leave period field | 09:13 |
| vincentfrancoise | exactly | 09:13 |
| vincentfrancoise | also, you said that the period is 60 seconds right? | 09:14 |
| alexchadin | By default | 09:14 |
| alexchadin | We can specify this parameter | 09:15 |
| alexchadin | hm | 09:15 |
| alexchadin | wait | 09:15 |
| alexchadin | you speak about audit's period? | 09:15 |
| vincentfrancoise | no the "refresh rate" of the periodic task | 09:15 |
| alexchadin | yes, 60 seconds | 09:15 |
| vincentfrancoise | so I have a question | 09:17 |
| vincentfrancoise | what if someone cancels the audit job | 09:17 |
| vincentfrancoise | but the period of the job is 30 seconds | 09:17 |
| vincentfrancoise | the time.sleep will still run the audit | 09:18 |
| vincentfrancoise | so we need to add a condition to the audit handler | 09:19 |
| vincentfrancoise | before https://review.openstack.org/#/c/321411/1/watcher/decision_engine/audit/default.py line 119 | 09:20 |
| vincentfrancoise | which checks the state of the audit | 09:20 |
| vincentfrancoise | so that only ONGOING audits should be executed | 09:20 |
| vincentfrancoise | or actually audits that are not CANCELLED | 09:21 |
| vincentfrancoise | well if we look at https://factory.b-com.com/www/watcher/doc/watcher/architecture.html#audit-state-machine | 09:22 |
| alexchadin | so that only ONGOING audits should be executed - but we set ONGOING state on 119 line | 09:23 |
| vincentfrancoise | only PENDING and ONGOING should be allowed | 09:23 |
| vincentfrancoise | where PENDING audits should be updated to ONGOING and already ONGOING should stay ONGOING | 09:23 |
| alexchadin | let's sum it overall | 09:26 |
| alexchadin | We launch audit | 09:26 |
| vincentfrancoise | ok | 09:26 |
| alexchadin | only if it is in PENDING state or ONGOING state | 09:26 |
| vincentfrancoise | yes | 09:27 |
| alexchadin | good | 09:27 |
| vincentfrancoise | the problem with that | 09:27 |
| vincentfrancoise | is that we will query cancelled audits over and over again to see that they are cancelled | 09:28 |
| vincentfrancoise | so add a filter on the state maybe | 09:28 |
| alexchadin | I have just thought of it | 09:29 |
| alexchadin | thank you | 09:29 |
| vincentfrancoise | {'type': 'CONTINUOUS', 'state': ('PENDING', 'ONGOING')} | 09:29 |
| vincentfrancoise | the sql IN operation is not handled currently | 09:29 |
| vincentfrancoise | by the objects filters | 09:30 |
| alexchadin | I'll make these changes and commit them ASAP | 09:31 |
| vincentfrancoise | cool :0 | 09:31 |
| vincentfrancoise | :) | 09:31 |
| vincentfrancoise | thanks for answering my concerns :) | 09:32 |
| vincentfrancoise | I'll link the conversation log into the review for the others | 09:32 |
| *** alexchadin has quit IRC | 09:38 | |
| openstackgerrit | Edwin Zhai proposed openstack/python-watcherclient: Enable strategy parameter https://review.openstack.org/297557 | 09:41 |
| *** alexchadin has joined #openstack-watcher | 09:48 | |
| *** alexchadin has quit IRC | 09:54 | |
| *** alexchadin has joined #openstack-watcher | 09:56 | |
| openstackgerrit | Merged openstack/watcher: Updated from global requirements https://review.openstack.org/319893 | 10:05 |
| alexchadin | vincentfrancoise: how will I know that continuous audit is already waiting for execute? | 10:07 |
| vincentfrancoise | because there is already a job for that audit in the pipe | 10:08 |
| *** thorst_ has joined #openstack-watcher | 10:09 | |
| alexchadin | how may I check the content of the pipe? | 10:09 |
| vincentfrancoise | good question | 10:09 |
| *** ecelik has joined #openstack-watcher | 10:11 | |
| vincentfrancoise | I guess we need to monitor the jobs we put in the queue | 10:15 |
| gzhai2 | vincentfrancoise: Thanks for review my code. Just uploaded new one to fit current infrastructure well. | 10:15 |
| vincentfrancoise | gzhai2: I didn't finish the reviews because I have a question about your code which I cannot answer yet | 10:16 |
| vincentfrancoise | gzhai2: how do you make sure that the parameters that are given by the operator are compliant with the constraints? | 10:17 |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Add goal name as filter for strategy list cmd https://review.openstack.org/322054 | 10:17 |
| *** thorst_ has quit IRC | 10:18 | |
| gzhai2 | vincentfrancoise: if not comply with the contraints, exception raised and audit's state become 'FAILED' | 10:21 |
| gzhai2 | operator can check the STATE for it. | 10:22 |
| gzhai2 | vincentfrancoise：vincentfrancoise: Another alternative is change the input value to default if constrain broken. But this silent failure is not good. | 10:32 |
| alexchadin | vincentfrancoise: we also need to check SUCCEEDED type | 10:40 |
| alexchadin | vincentfrancoise: when we get list of audits | 10:41 |
| *** alexchadin has quit IRC | 10:45 | |
| *** openstackgerrit has quit IRC | 10:47 | |
| *** openstackgerrit has joined #openstack-watcher | 10:48 | |
| *** thorst_ has joined #openstack-watcher | 11:16 | |
| *** thorst_ has quit IRC | 11:25 | |
| *** alexchadin has joined #openstack-watcher | 11:35 | |
| *** alexchadin has quit IRC | 11:36 | |
| *** thorst_ has joined #openstack-watcher | 11:42 | |
| *** alexchadin has joined #openstack-watcher | 11:46 | |
| *** openstackgerrit has quit IRC | 11:47 | |
| *** openstackgerrit has joined #openstack-watcher | 11:48 | |
| *** vtech has quit IRC | 12:04 | |
| *** vtech has joined #openstack-watcher | 12:07 | |
| *** acabot_ has joined #openstack-watcher | 12:11 | |
| openstackgerrit | Merged openstack/watcher: Added cold VM migration support https://review.openstack.org/299927 | 12:33 |
| openstackgerrit | Alexander Chadin proposed openstack/watcher-specs: Add Watcher Continuosly Optimization spec https://review.openstack.org/315551 | 12:37 |
| *** openstackgerrit has quit IRC | 12:48 | |
| *** openstackgerrit has joined #openstack-watcher | 12:48 | |
| alexchadin | jed56: hello? | 12:55 |
| jed56 | alexchadin: hello | 12:55 |
| jed56 | how are you ? | 12:55 |
| alexchadin | I'm good, how are you? | 12:56 |
| jed56 | good :-) | 12:57 |
| alexchadin | :-) | 12:57 |
| alexchadin | I found out that hardware.cpu.util is standard metric | 12:57 |
| alexchadin | I added it with SNMP support | 12:58 |
| jed56 | ah sorry, can please send to me the link ? | 12:58 |
| jed56 | you | 12:58 |
| alexchadin | https://github.com/openstack/ceilometer/blob/master/ceilometer/hardware/pollsters/data/snmp.yaml | 13:00 |
| alexchadin | take a look | 13:00 |
| alexchadin | I use it cause it comes with node_1 name | 13:01 |
| alexchadin | not node_1_node_1 as it comes with compute.node.cpu.percent | 13:01 |
| *** edleafe is now known as figleaf | 13:02 | |
| jed56 | Okay! you don't have take into account my comment | 13:02 |
| jed56 | I think the documentation of ceilometer is deprecated | 13:02 |
| jed56 | I didn't take the time to look at the code | 13:02 |
| alexchadin | No problem! | 13:03 |
| jed56 | alexchadin: do you think that you will have the time to address my other comments today ? | 13:03 |
| acabot_ | jwcroppe_ : I just +2 https://review.openstack.org/315551, if you can have a look and W+1 if OK | 13:04 |
| jed56 | we are releasing watcher on monday | 13:04 |
| alexchadin | I hope so | 13:04 |
| vincentfrancoise | acabot_: I just -1'd it | 13:04 |
| alexchadin | wait a sec | 13:04 |
| jed56 | we wants to have your strategy in this release | 13:04 |
| acabot_ | vincentfrancoise : ok ;-) | 13:05 |
| acabot_ | vincentfrancoise : ok I agree about the state machine | 13:06 |
| openstackgerrit | Alexander Chadin proposed openstack/watcher-specs: Add Watcher Continuosly Optimization spec https://review.openstack.org/315551 | 13:07 |
| vincentfrancoise | alexchadin: I +1'd it ;) | 13:09 |
| alexchadin | :) | 13:09 |
| alexchadin | I'm working on strategy | 13:09 |
| alexchadin | will be commited very soon | 13:09 |
| openstackgerrit | Alexander Chadin proposed openstack/watcher: Add Overload standard deviation strategy https://review.openstack.org/297590 | 13:16 |
| *** hvprash has joined #openstack-watcher | 13:17 | |
| *** hvprash_ has joined #openstack-watcher | 13:18 | |
| *** acabot_ has quit IRC | 13:18 | |
| *** hvprash has quit IRC | 13:21 | |
| *** alexchadin has quit IRC | 13:21 | |
| openstackgerrit | Alexander Chadin proposed openstack/watcher: Add continuously optimization https://review.openstack.org/321411 | 13:22 |
| *** alexchadin has joined #openstack-watcher | 13:22 | |
| *** alexchadin has quit IRC | 13:22 | |
| *** acabot_ has joined #openstack-watcher | 13:25 | |
| *** acabot_ has quit IRC | 13:28 | |
| *** hvprash_ has quit IRC | 13:29 | |
| *** hvprash has joined #openstack-watcher | 13:30 | |
| *** ecelik has quit IRC | 13:51 | |
| *** hvprash has quit IRC | 13:56 | |
| *** hvprash has joined #openstack-watcher | 13:56 | |
| *** acabot_ has joined #openstack-watcher | 14:07 | |
| *** acabot_ has quit IRC | 14:14 | |
| *** openstackgerrit has quit IRC | 14:18 | |
| *** openstackgerrit has joined #openstack-watcher | 14:18 | |
| *** lrensing has joined #openstack-watcher | 14:22 | |
| openstackgerrit | Jinquan Ni proposed openstack/watcher: Destination selectors plugin rst https://review.openstack.org/319176 | 14:43 |
| *** acabot_ has joined #openstack-watcher | 14:48 | |
| *** acabot_ has quit IRC | 14:51 | |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added EfficacyIndicator object https://review.openstack.org/318178 | 15:21 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added efficacy indicators to /action_plans https://review.openstack.org/318179 | 15:21 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added EfficacyIndicator model in DB https://review.openstack.org/318176 | 15:21 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added efficacy specification to /goals https://review.openstack.org/318177 | 15:21 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Documentation on goal and efficacy https://review.openstack.org/319275 | 15:21 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Decoupled Goal from Strategy https://review.openstack.org/318175 | 15:21 |
| openstackgerrit | Vladimir Ostroverkhov proposed openstack/python-watcherclient: Add support continuously-optimization https://review.openstack.org/321421 | 15:30 |
| *** vmahe has quit IRC | 15:42 | |
| openstackgerrit | Merged openstack/python-watcherclient: Replaced UUID of goal and strategy with name https://review.openstack.org/311985 | 15:50 |
| openstackgerrit | Merged openstack/python-watcherclient: Use goal name as strategy list filter https://review.openstack.org/321999 | 15:51 |
| openstackgerrit | Merged openstack/python-watcherclient: Update Watcher CLI documentation https://review.openstack.org/320451 | 15:51 |
| openstackgerrit | Merged openstack/watcher: Add goal_name & strategy_name in /audit_templates https://review.openstack.org/311733 | 15:51 |
| openstackgerrit | Merged openstack/watcher: Add goal name as filter for strategy list cmd https://review.openstack.org/322054 | 15:51 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Watcher plugins table in Guru meditation reports https://review.openstack.org/320529 | 15:55 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Enabled config parameters to plugins https://review.openstack.org/314036 | 15:55 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Documentation for plugins-parameters https://review.openstack.org/314037 | 15:55 |
| openstackgerrit | Merged openstack/watcher: Update Watcher documentation https://review.openstack.org/320493 | 15:58 |
| *** vtech has quit IRC | 16:01 | |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added EfficacyIndicator object https://review.openstack.org/318178 | 16:03 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added efficacy indicators to /action_plans https://review.openstack.org/318179 | 16:03 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added EfficacyIndicator model in DB https://review.openstack.org/318176 | 16:03 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Added efficacy specification to /goals https://review.openstack.org/318177 | 16:03 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Documentation on goal and efficacy https://review.openstack.org/319275 | 16:03 |
| openstackgerrit | Vincent Françoise proposed openstack/watcher: Decoupled Goal from Strategy https://review.openstack.org/318175 | 16:03 |
| *** vincentfrancoise has quit IRC | 16:18 | |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Enabled config parameters to plugins https://review.openstack.org/314036 | 16:25 |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Watcher plugins table in Guru meditation reports https://review.openstack.org/320529 | 16:25 |
| openstackgerrit | David TARDIVEL proposed openstack/watcher: Documentation for plugins-parameters https://review.openstack.org/314037 | 16:26 |
| *** hvprash_ has joined #openstack-watcher | 16:34 | |
| *** hvprash has quit IRC | 16:37 | |
| *** thorst_ has quit IRC | 16:42 | |
| *** thorst_ has joined #openstack-watcher | 16:44 | |
| *** hvprash_ has quit IRC | 16:46 | |
| *** hvprash has joined #openstack-watcher | 16:46 | |
| *** thorst_ has quit IRC | 16:48 | |
| *** alexchadin has joined #openstack-watcher | 17:08 | |
| *** vtech has joined #openstack-watcher | 17:08 | |
| alexchadin | jed56: I commited new PS for strategy | 17:08 |
| *** hvprash has quit IRC | 17:25 | |
| *** hvprash has joined #openstack-watcher | 17:26 | |
| *** hvprash_ has joined #openstack-watcher | 17:27 | |
| *** alexchadin has quit IRC | 17:29 | |
| *** hvprash has quit IRC | 17:30 | |
| *** wootehfoot has joined #openstack-watcher | 17:46 | |
| *** alexchadin has joined #openstack-watcher | 17:49 | |
| *** thorst_ has joined #openstack-watcher | 17:49 | |
| *** thorst_ has quit IRC | 17:50 | |
| *** thorst_ has joined #openstack-watcher | 17:50 | |
| *** wootehfoot has quit IRC | 17:50 | |
| *** alexchadin has quit IRC | 17:54 | |
| *** vtech has quit IRC | 17:57 | |
| *** wootehfoot has joined #openstack-watcher | 17:58 | |
| *** vtech has joined #openstack-watcher | 18:14 | |
| *** lrensing has quit IRC | 18:17 | |
| *** vtech has quit IRC | 18:19 | |
| *** hvprash has joined #openstack-watcher | 18:20 | |
| *** hvprash_ has quit IRC | 18:23 | |
| *** vtech has joined #openstack-watcher | 18:29 | |
| jwcroppe_ | acabot_: +W'd | 18:30 |
| openstackgerrit | Merged openstack/watcher-specs: Add Watcher Continuosly Optimization spec https://review.openstack.org/315551 | 18:33 |
| *** hvprash_ has joined #openstack-watcher | 18:34 | |
| *** hvprash has quit IRC | 18:37 | |
| *** wootehfoot has quit IRC | 18:51 | |
| *** wootehfoot has joined #openstack-watcher | 19:00 | |
| *** hvprash has joined #openstack-watcher | 19:33 | |
| *** hvprash__ has joined #openstack-watcher | 19:35 | |
| *** hvprash_ has quit IRC | 19:36 | |
| *** hvprash has quit IRC | 19:38 | |
| *** hvprash__ has quit IRC | 19:41 | |
| *** openstack has joined #openstack-watcher | 21:42 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!