Friday, 2017-01-27

*** hidekazu has joined #openstack-watcher00:02
hidekazumorning00:03
*** harlowja has quit IRC00:04
yuanyingmorning00:05
*** thorst_ has joined #openstack-watcher00:45
*** thorst_ has quit IRC00:55
*** thorst_ has joined #openstack-watcher01:27
*** thorst_ has quit IRC02:18
*** thorst_ has joined #openstack-watcher02:20
*** absubram has joined #openstack-watcher02:43
*** sanfern has quit IRC02:47
*** absubram has quit IRC02:48
*** absubram has joined #openstack-watcher02:49
*** yuanying has quit IRC02:56
*** thorst_ has quit IRC03:40
*** sanfern has joined #openstack-watcher04:14
*** absubram has quit IRC04:55
*** hidekazu has quit IRC05:05
*** adisky_ has joined #openstack-watcher05:11
*** thorst_ has joined #openstack-watcher05:56
*** thorst_ has quit IRC06:01
*** diga has joined #openstack-watcher07:09
*** diga has quit IRC07:12
*** diga has joined #openstack-watcher07:12
*** vmahe has joined #openstack-watcher07:17
*** diga has quit IRC07:26
*** adisky_ has quit IRC07:29
*** alexchadin has joined #openstack-watcher07:54
*** thorst_ has joined #openstack-watcher07:57
*** thorst_ has quit IRC08:01
*** dtardivel has joined #openstack-watcher08:05
*** diga has joined #openstack-watcher08:11
*** hidekazu has joined #openstack-watcher08:16
*** vmahe_ has joined #openstack-watcher08:36
*** adisky_ has joined #openstack-watcher08:36
acabotalexchadin : morning08:40
alexchadinacabot: morning08:40
*** vincentfrancoise has joined #openstack-watcher08:40
*** vmahe has left #openstack-watcher08:40
acabotalexchadin : do you know if you would be able to attend the Atlanta PTG ? I mean if you become our PTL08:41
alexchadinacabot: Unfortunately, I've missed Travel Support08:42
acabotalexchadin : ok would you be able to be with us via hangout ?08:42
acabotUTC-5 hours08:43
alexchadinacabot: I've already registered for Boston. Yeah, of course I will be.08:43
alexchadinacabot: It will be about 22-23 hours for me, but it's okay08:43
acabotalexchadin : so it will be easier for you to attend morning session right ?08:44
alexchadinacabot: absolutely right08:44
acabotalexchadin : I will take this into account when building the agenda08:44
alexchadinacabot: thank you08:45
acabotalexchadin : could you please add your name on https://etherpad.openstack.org/p/pike-watcher-ptg and add your prefered time for hangout sessions ?08:45
acabotalexchadin : thx08:46
alexchadinacabot: Done08:48
vincentfrancoisehidekazu: Hi08:49
hidekazuvincentfrancoise: Hi08:53
vincentfrancoisehidekazu: did you discuss with your colleagues about yesterday's use case?08:54
hidekazuvincentfrancoise: i discussed with colleague today.08:54
hidekazuvincentfrancoise: we want continuous audit to suspend, which means propose no actionplan.08:55
vincentfrancoisehidekazu: and then you want to be able to resume right?08:55
hidekazuvincentfrancoise: and then re-run.08:55
hidekazuvincentfrancoise: yes.08:56
hidekazuvincentfrancoise: we achived by state change. once cancelled, then ongoing..08:57
hidekazuvincentfrancoise: state change is surprising way,08:58
vincentfrancoisehidekazu: ok so that's going to require a blueprint because the best way of doing it would probably mean adding a new PAUSED state or something like this.08:58
vincentfrancoisehidekazu: if you look at http://docs.openstack.org/developer/watcher/architecture.html#audit-state-machine08:59
hidekazuyes, i know of course.09:00
vincentfrancoisehidekazu: you'll see that what we do not really allow switching from CANCELLED to ONGOING09:00
vincentfrancoisehidekazu: although we may not check whether or not the transition is actually allowed09:00
hidekazuvincentfrancoise: so state check should be introduced, right?09:01
vincentfrancoisehidekazu: so to update this state machine, we need to write a blueprint explaining all the impacts I believe09:01
vincentfrancoisehidekazu: there's already some transitions that are verified I think09:02
hidekazuvincentfrancoise: our company are creating application includes watcher plugin.09:02
vincentfrancoisehidekazu: after quick check, we seem to actually only check the transitions for action plans09:03
vincentfrancoisehidekazu: see https://github.com/openstack/watcher/blob/master/watcher/api/controllers/v1/action_plan.py#L491-L51309:03
hidekazuvincentfrancoise: so i will analyze why our company need to create app other than watcher , and propose some.09:03
vincentfrancoisehidekazu: so IMHO, we should solve this issue in 2 steps09:04
vincentfrancoisehidekazu: 1 is to block the update of an audit altogether to prevent any broken state as you mentioned with the period that can't be changed09:05
vincentfrancoisehidekazu: the 2nd step would be to submit a BP that re-enable PATCH on audits but with thorough state transition checks + (or separately) introduces the ability to somehow pause an audit09:06
vincentfrancoisehidekazu: the 1st step could be done in ocata-rc1 although the 2nd step would have to wait for the Pike cycle09:07
alexchadinhi guys09:08
vincentfrancoisehidekazu: BTW, are your applications that use Watcher in production?09:09
vincentfrancoisealexchadin: morning09:09
hidekazuvincentfrancoise: No, PoC09:10
vincentfrancoisehidekazu: ok :)09:10
hidekazuvincentfrancoise: under development. target is the end of march.09:10
hidekazuvincentfrancoise: i will discuss at ptg09:11
hidekazuvicentfrancoise: if alexchadin permit :)09:11
vincentfrancoisehidekazu: ah nice, you go to the PTG?09:11
hidekazuvincentfrancoise: yes.09:11
hidekazuvincentfrancoise: so from Dec company work bother me.09:12
alexchadinvincentfrancoise: IMHO we need to persist our continuosly audits09:12
vincentfrancoisehidekazu: BTW, following our discussion on HA, I added some discussions points about it09:12
alexchadinvincentfrancoise: since if watcher is in down state, restart will reset all next times to run09:13
vincentfrancoisealexchadin: that could be solution, I didn't give much thought to it though09:14
hidekazualexchadin: what is down state?09:14
vincentfrancoisealexchadin: but for resilience purposes, we should now make sure that Watcher is as stateless as possible09:15
alexchadinseems that we don't have proper mechanism to control changes states09:15
hidekazualexchadin: service down?09:15
alexchadinin states*09:15
alexchadinhidekazu: yeap09:15
vincentfrancoisealexchadin: which mainly applies to the decision engine09:15
vincentfrancoiseour main objective was to first make it all work with as little moving parts/external services as possible09:16
vincentfrancoisethat's why many things were computed and stored in-memory09:17
vincentfrancoisebut this should change eventually09:18
hidekazuvincentfrancoise: placement api comes.09:18
vincentfrancoisehidekazu: I know, this is really good news and I've put it in the etherpad for the PTG too09:19
hidekazuvincentfrancoise: i know only existence.09:19
alexchadinvincentfrancoise: what if administrator wants to set continuos audit to run it everyday at 3:00 am? If decision engine restarted, next-run will be reseted.09:19
vincentfrancoisealexchadin: but what if when you resume the DE, it's 3:30am, should you still run it?09:21
vincentfrancoiseor should we wait for 3:00 am of the next day?09:21
alexchadinvincentfrancoise: good question09:21
alexchadinI think it should run next dat09:21
alexchadinday09:21
*** efoley has joined #openstack-watcher09:22
hidekazualexchadin: +109:22
vincentfrancoisealexchadin: that really depends on everyone's use case09:22
alexchadinseems that we need one more blueprint :D09:22
alexchadinvincentfrancoise: at least next-run field should be stored in DB09:23
hidekazui may add bp about suspend and resume.09:23
vincentfrancoiseoverall, we should also get away from period in seconds (if possible) to instead rely on the cron syntax as well09:23
alexchadinvincentfrancoise: didn't someone already propose it?09:24
vincentfrancoisealexchadin: maybe, I don't know09:25
vincentfrancoisealexchadin: it's difficult to both follow everything and write code :p09:25
alexchadincron syntax is good idea09:25
alexchadinvincentfrancoise: +109:25
vincentfrancoisealexchadin: I don't see a thing on cron syntax for CONTINOUS audits09:26
vincentfrancoiseI'll submit one straight away ;)09:27
alexchadinvincentfrancoise: you mean BP?09:28
vincentfrancoisealexchadin: yup09:28
alexchadinvincentfrancoise: cool09:28
hidekazuvincentfrancoise: suspend & resume needs another bp, right?09:29
vincentfrancoisealexchadin: it's a "not too hard" BP with a low-ish priority which means that it can also be assigned to newcomers09:29
vincentfrancoisehidekazu: yes09:29
hidekazuvincentfrancoise: ok.09:30
vincentfrancoisehidekazu: the smaller a BP is, the more chances people will decide to implement them09:30
hidekazuvincentfrancoise: i see.09:31
hidekazuvincentfrancoise: alexchadin: i will propose storage optimization also.09:32
hidekazuchould i propose bp for pike?09:32
alexchadinhidekazu: You have some thoughts about it?09:32
vincentfrancoisehidekazu: the Pike roadmap will be decided at the PTG I believe so submit it and it will be dispatched afterwards09:33
hidekazualexchadin: yes, but not have code yet.09:33
hidekazuvincentfranoise: so submit before ptg.09:34
vincentfrancoisehidekazu: yes09:34
vincentfrancoisehidekazu: add also add it to the discussions in the etherpad09:35
vincentfrancoisehidekazu: https://etherpad.openstack.org/p/pike-watcher-ptg09:35
hidekazuvincentfrancoise: alexchadin already added question about storage.09:35
hidekazualexchadin: do you have concrete idea?09:36
vincentfrancoisehidekazu: ok so we're good then (add you name next to it too so it gives it more weight)09:36
hidekazui will submit bps about suspend&resume and storage optimization in 2 weeks.09:38
alexchadinhidekazu: not yet09:39
hidekazualexchadin: i just want to avoid conflict with you.09:39
alexchadinhidekazu: feel free to share your thoughts about this topic09:40
hidekazualexchadin: ok, thanks.09:40
alexchadinhvprash: Hi09:40
hidekazualexchadin: btw, i happy you will become ptl.09:41
alexchadinhidekazu: It is not exactly yet, but thank you for your support :)09:42
openstackgerritDeepak Jon proposed openstack/puppet-watcher: Add oslo::db to dependency chain  https://review.openstack.org/42613609:43
vincentfrancoisehttps://blueprints.launchpad.net/watcher/+spec/cron-based-continuous-audits09:45
alexchadincool09:46
alexchadinvincentfrancoise: don'y you know, is there someone who already implements https://blueprints.launchpad.net/watcher/+spec/audit-tag-vm-metadata ?09:46
alexchadinvincentfrancoise: I'm looking for BP to take09:46
vincentfrancoisealexchadin: this BP is still under discussion so don't go for it yet09:47
vincentfrancoisealexchadin: I have discussions about it with hvprash and it definitely needs some more work09:47
alexchadinvincentfrancoise: shouldn't we add it to PTG topics?09:48
vincentfrancoisealexchadin: we already discussed it in Barcelona, it's more about implementation details at this stage so I'm not sure09:49
vincentfrancoisealexchadin: the pending questions are mainly about security because VM tags are seen by tenants so we should make sure they don't see more than they should09:50
alexchadinvincentfrancoise: yeah, I see your comments09:50
*** hidekazu has quit IRC09:51
vincentfrancoisealexchadin: a fairly easy BP that can be useful too is https://blueprints.launchpad.net/watcher/+spec/action-versioned-notifications-api09:52
*** gabor_antal has joined #openstack-watcher09:52
openstackgerritMerged openstack/watcher: Documentation update  https://review.openstack.org/42418709:53
*** thorst_ has joined #openstack-watcher09:58
*** absubram has joined #openstack-watcher10:02
*** thorst_ has quit IRC10:02
alexchadinvincentfrancoise: add me as contributor to this BP10:06
*** absubram has quit IRC10:07
*** alexchadin has quit IRC10:26
*** alexchadin has joined #openstack-watcher10:37
*** pkoniszewski has quit IRC10:41
*** pkoniszewski has joined #openstack-watcher10:42
*** vmahe_ has quit IRC10:44
*** alexchadin has quit IRC11:18
*** alexchadin has joined #openstack-watcher11:19
*** alexchadin has quit IRC11:19
*** alexchadin has joined #openstack-watcher11:19
*** alexchadin has quit IRC11:19
*** alexchadin has joined #openstack-watcher11:20
*** alexchadin has quit IRC11:20
*** alexchadin has joined #openstack-watcher11:21
*** alexchadin has quit IRC11:21
*** alexchadin has joined #openstack-watcher11:22
*** alexchadin has quit IRC11:22
*** alexchadin has joined #openstack-watcher11:22
*** alexchadin has quit IRC11:23
*** alexchadin has joined #openstack-watcher11:23
*** alexchadin has quit IRC11:23
*** thorst_ has joined #openstack-watcher11:59
*** thorst_ has quit IRC12:03
openstackgerritDeepak Jon proposed openstack/puppet-watcher: Add oslo::db to dependency chain  https://review.openstack.org/42613612:09
*** dtardivel has quit IRC12:14
*** diga has quit IRC12:16
*** alexchadin has joined #openstack-watcher12:17
*** alexchadin has quit IRC12:42
*** thorst__ has joined #openstack-watcher12:43
*** sanfern has quit IRC12:47
*** alexchadin has joined #openstack-watcher13:02
*** thorst__ is now known as thorst_13:42
*** edleafe is now known as figleaf13:54
openstackgerritVincent Françoise proposed openstack/watcher: Idiomatic loop for calculate_num_migrations  https://review.openstack.org/42624014:00
*** absubram has joined #openstack-watcher14:14
*** absubram has quit IRC14:18
*** absubram has joined #openstack-watcher14:24
*** alexchadin has quit IRC14:28
*** sanfern has joined #openstack-watcher14:31
*** efoley has quit IRC14:37
*** efoley has joined #openstack-watcher14:38
*** efoley has quit IRC14:39
*** efoley has joined #openstack-watcher14:39
*** efoley_ has joined #openstack-watcher15:05
*** absubram has quit IRC15:06
*** efoley has quit IRC15:09
*** dtardivel has joined #openstack-watcher15:21
openstackgerritMerged openstack/watcher: Idiomatic loop for calculate_num_migrations  https://review.openstack.org/42624015:28
dtardivelsballe_: Hi15:29
sballe_yes?15:29
dtardivelhow are you ?15:30
sballe_good how about you?15:30
openstackgerritMerged openstack/watcher: Fix multinode tempest test failure  https://review.openstack.org/41954415:31
dtardivelsballe_: fine. could you please review https://review.openstack.org/#/c/425582/1 & https://review.openstack.org/#/c/425605/ and add workflow+1 ?15:31
dtardivelsballe_: thx :)15:33
sballe_yes will do15:33
dtardivelsballe_: and this one https://review.openstack.org/#/c/425261/ as well :)15:34
sballe_ok :-)15:34
sballe_dtardivel: done15:43
openstackgerritMerged openstack/watcher-specs: Add specs Pike folder  https://review.openstack.org/42558215:46
openstackgerritMerged openstack/watcher-specs: Move Ocata specs as implemented  https://review.openstack.org/42560515:46
*** absubram has joined #openstack-watcher15:49
*** efoley__ has joined #openstack-watcher16:06
*** efoley_ has quit IRC16:09
*** efoley__ has quit IRC16:12
*** absubram has quit IRC16:47
*** vincentfrancoise has quit IRC16:55
*** harlowja has joined #openstack-watcher18:02
*** dtardivel has quit IRC18:04
*** wootehfoot has joined #openstack-watcher18:56
*** absubram has joined #openstack-watcher19:10
*** absubram has quit IRC19:15
*** absubram has joined #openstack-watcher19:18
*** ss4 has joined #openstack-watcher21:09
*** ss4 has quit IRC21:11
*** wootehfoot has quit IRC21:13
*** wootehfoot has joined #openstack-watcher21:18
*** absubram has quit IRC21:33
*** absubram has joined #openstack-watcher21:55
*** wootehfoot has quit IRC22:01
*** absubram has quit IRC22:01
*** wootehfoot has joined #openstack-watcher22:03
*** thorst_ has quit IRC22:16
*** danpawli1 has quit IRC22:30
*** danpawlik has joined #openstack-watcher23:04
*** figleaf is now known as edleafe23:09

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!