Wednesday, 2014-04-23

openstackgerritA change was merged to openstack/python-ceilometerclient: Correct testcase content
*** _nadya_ has quit IRC00:25
*** matsuhashi has joined #openstack-ceilometer00:29
*** _cjones_ has quit IRC00:50
*** _cjones_ has joined #openstack-ceilometer00:50
*** alexpilotti has quit IRC00:54
*** _cjones_ has quit IRC00:55
*** matsuhashi has quit IRC00:59
*** matsuhashi has joined #openstack-ceilometer01:01
*** yfujioka has quit IRC01:05
*** fnaval has joined #openstack-ceilometer01:10
*** _cjones_ has joined #openstack-ceilometer01:21
*** flwang has quit IRC01:32
*** _cjones_ has quit IRC01:32
*** flwang has joined #openstack-ceilometer01:39
*** nosnos has joined #openstack-ceilometer01:41
*** flwang has quit IRC01:41
*** flwang has joined #openstack-ceilometer02:15
openstackgerritZhiQiang Fan proposed a change to openstack/ceilometer: Disable reverse dns lookup
*** flwang has quit IRC02:21
*** flwang has joined #openstack-ceilometer02:24
*** fnaval has quit IRC02:41
*** fnaval has joined #openstack-ceilometer02:41
*** fnaval has quit IRC02:46
*** fnaval has joined #openstack-ceilometer03:03
*** openstackgerrit has quit IRC03:04
*** openstackgerrit has joined #openstack-ceilometer03:04
*** matsuhashi has quit IRC03:14
*** matsuhashi has joined #openstack-ceilometer03:15
*** matsuhashi has quit IRC03:19
*** _cjones_ has joined #openstack-ceilometer03:29
*** nosnos has quit IRC03:29
*** changbl has joined #openstack-ceilometer03:32
*** _cjones_ has quit IRC03:34
*** rwsu has quit IRC03:50
*** matsuhashi has joined #openstack-ceilometer04:19
*** nosnos has joined #openstack-ceilometer04:22
*** fnaval has quit IRC04:49
*** fnaval has joined #openstack-ceilometer04:50
*** fnaval has quit IRC04:54
*** ildikov has quit IRC05:01
*** shakamunyi has quit IRC05:14
*** ilyashakhat has quit IRC05:15
*** ilyashakhat has joined #openstack-ceilometer05:16
*** nati_ueno has quit IRC05:16
*** ildikov has joined #openstack-ceilometer05:23
*** matsuhashi has quit IRC05:37
*** matsuhashi has joined #openstack-ceilometer05:37
*** matsuhashi has quit IRC05:43
*** matsuhashi has joined #openstack-ceilometer05:44
*** matsuhas_ has joined #openstack-ceilometer05:50
*** matsuhashi has quit IRC05:53
*** matsuhas_ has quit IRC05:58
*** matsuhashi has joined #openstack-ceilometer05:58
openstackgerritOpenStack Proposal Bot proposed a change to openstack/ceilometer: Imported Translations from Transifex
*** _nadya_ has joined #openstack-ceilometer06:09
openstackgerritliusheng proposed a change to openstack/ceilometer: Support direct alarm_evaluator db access
*** nati_ueno has joined #openstack-ceilometer06:19
openstackgerritliusheng proposed a change to openstack/ceilometer: Support direct alarm_evaluator db access
*** _nadya_ has quit IRC06:26
*** _nadya_ has joined #openstack-ceilometer06:31
*** _nadya_ has quit IRC06:46
*** lsmola has joined #openstack-ceilometer06:48
*** _nadya_ has joined #openstack-ceilometer06:51
*** idegtiarov has joined #openstack-ceilometer07:03
*** nati_ueno has quit IRC07:16
*** _nadya_ has quit IRC07:20
*** nacim has joined #openstack-ceilometer07:23
*** sdake_ has quit IRC07:25
*** _nadya_ has joined #openstack-ceilometer07:26
*** ildikov has quit IRC07:38
*** ildikov has joined #openstack-ceilometer07:43
*** _nadya_ has quit IRC07:45
openstackgerritliusheng proposed a change to openstack/ceilometer: Support direct alarm_evaluator db access
*** nati_ueno has joined #openstack-ceilometer07:55
*** tian has joined #openstack-ceilometer07:59
*** nacim has quit IRC08:11
*** nacim has joined #openstack-ceilometer08:15
*** Alexei_987 has quit IRC08:27
*** matsuhashi has quit IRC08:27
*** matsuhashi has joined #openstack-ceilometer08:34
openstackgerritIonut Artarisi proposed a change to openstack/ceilometer: reconnect to mongodb on connection failure
*** bada has quit IRC09:00
*** Alexei_987 has joined #openstack-ceilometer09:14
*** flwang has quit IRC09:25
*** eglynn has joined #openstack-ceilometer09:26
*** ildikov_ has joined #openstack-ceilometer09:28
*** ildikov has quit IRC09:31
openstackgerritAlexei Kornienko proposed a change to openstack/ceilometer: Made get_capabilities a classmethod instead of object method
openstackgerritDina Belova proposed a change to openstack/ceilometer: Describe storage backends in the collector installation guide
openstackgerritliusheng proposed a change to openstack/ceilometer: Support direct alarm_evaluator db access
*** alexpilotti has joined #openstack-ceilometer09:46
openstackgerritAlexei Kornienko proposed a change to openstack/ceilometer: Replaced CONF object with url in storage engine creation
openstackgerritSylvain Afchain proposed a change to openstack/ceilometer: Opencontrail network statistics driver
*** matsuhashi has quit IRC10:16
openstackgerritDina Belova proposed a change to openstack/ceilometer: Change 200->404 for the history of non-existent alarm
*** nacim has quit IRC10:21
*** matsuhashi has joined #openstack-ceilometer10:23
jd__eglynn: I've filled the form :)10:28
eglynnjd__: that was quick :) thank you sir!10:28
jd__I know how painful it is to organize that10:29
jd__I know you're getting tons of email with various more or less interesting stuff for the next summit10:30
jd__it's like organizing a party but without the interesting part!10:30
eglynnjd__: yeah, someone else gets to choose the beer ;)10:32
DinaBelova:D well, such big event are always about crazy party organizing :)10:34
*** _nadya_ has joined #openstack-ceilometer10:35
*** _nadya_ has quit IRC10:39
*** nacim has joined #openstack-ceilometer10:40
*** _nadya_ has joined #openstack-ceilometer10:44
*** Alexei_987 has quit IRC10:56
*** matsuhashi has quit IRC11:04
DinaBelovaeglynn, btw, may you take a look on the ? As discussed with ildikov_, there is no clear idea if that's the thing to fix really. I've caught it from the Launchpad (as it was triaged one), but after some discussions there is no solid opinion about this problem.11:22
eglynnDinaBelova: thanks, looking ...11:24
ildikov_eglynn: hi11:25
ildikov_eglynn: I will have another question too, when the bug topic is closed :)11:26
eglynnildikov_: hey ... shoot11:26
ildikov_eglynn: so I saw the doc about the sessions and there is the doc related one, which I registered under the Documentation topic11:27
ildikov_eglynn: so it should be already under the right track or did I miss something?11:27
ildikov_eglynn: I'm a bit overloaded nowadays, so it can easily happen :S :)11:28
eglynnildikov_: yep, you submitted to the right track originally :)11:29
eglynnildikov_: ... but I included it in the ceilo speadsheet anyway because we have to be sure it gets scheduled one way or the other11:29
eglynnildikov_: ... so if annegentle doesn't have space for it on the docco track11:29
eglynnildikov_: ... then we'll have to squeeze in somehow, seeing as it's one of the TC commitments11:29
eglynn... squeeze *it in11:30
ildikov_eglynn: do we know how many sessions are in the Doc track?11:32
eglynnildikov_: ... docco has only 5 proposals for 4 slots11:33
eglynnildikov_: ... (smaller over-subscription ratio than most other projects)11:33
eglynnildikov_: ... so I'm still hopeful it can be accomodated there11:33
ildikov_eglynn: hopefully in the worst case scenario we can get a shared session there11:34
eglynnildikov_: ... yep, fingers crossed :)11:34
ildikov_eglynn: mine too :)11:34
eglynnildikov_: ... I'll reach out to anne about it today, at a time more convenient for her TZ11:34
ildikov_eglynn: cool, thanks, please update me with the news11:35
eglynnildikov_: ... will do!11:35
ildikov_eglynn: ... cool, many thanks11:37
ildikov_eglynn: now get back to the original bug, I'm the bad cop there too :)11:37
eglynnildikov_: ... how uncharacteristic of you! ;)11:38
ildikov_eglynn: ... FYI, currently not my statistics are the worst :)11:38
eglynnildikov_: ... you must be going soft ;)11:39
ildikov_eglynn: ... I blame the Easter Bunny ;)11:45
ildikov_eglynn: do you think that we need an error message there?11:54
ildikov_eglynn: for instance it will return an error also, when the user do not have permission to see the alarm history11:54
eglynnildikov_: ... an explicit error msg in response to alarm history query, for never-existing alarms?11:55
ildikov_eglynn: ... yes11:55
ildikov_eglynn: ... I do not see why it would be more useful, than the empty list11:55
ildikov_eglynn: ... also in case of the future TTL value never existing is a question11:56
eglynnildikov_: ... isn't an error msg provided by Dina's patch?11:56
eglynn line 94?11:56
ildikov_eglynn: ... there is also a config option to turn off alarm history, in which case if we tell that the alarm never existed is totally misleading and not correct11:57
ildikov_eglynn: .. in my view the bug itself is not valid11:57
ildikov_eglynn: ... so if everyone sais that the bug itself is correct, than it will be hard to return correct error messag(es)11:58
eglynnildikov_: ... but doesn't the error msg currentlu just say that history wasn't found?11:58
ildikov_eglynn: ... maybe I haven't seen the latest version of the patch11:59
eglynnildikov_: ... yeah it's "No history found for the alarm alarm_id" in
ildikov_eglynn: ... but in this case, my question is that why we should report that back, when the result of the alarm history query is an empty list?11:59
ildikov_eglynn: ... my first question is that the bug itself, for which this patch is proposed is valid or not?12:00
ildikov_eglynn: ... this one I mean:
eglynnildikov_: hmmm, do we currently return 404 for a constrained statistics query with no hits?12:01
eglynnildikov_: ... IIRC we just return [] in that case12:01
ildikov_eglynn: ... I'm pretty sure we do not return error message for any empty query result12:01
ildikov_eglynn: ... and I also think that it is correct in that way, so the empty list is the best I think12:02
silehteglynn, ildikov_ I agree too12:04
ildikov_eglynn: ... the current error message in that patch does not have any added value compared it to the empty list, but if you write anything else in that error message it will not be correct for all cases12:04
eglynnildikov_, sileht: ... don't we already return 404s for non-existence in several places?12:04
* eglynn checks code ....12:04
eglynnsay for non-existent alarm12:05
silehteglynn, only when we are really sure12:05
ildikov_eglynn: ... this is my strict opinion, but I thought it would be better to discuss, if I may miss something here12:05
eglynn... or for resource12:05
eglynnsileht: ... and the "not sure" case occurs say after the TTL has kicked in?12:06
ildikov_eglynn: ... it's a different issue there12:06
eglynnildikov_: ... is the core issue the fact that alarm history retention can be turned off?12:07
silehteglynn, or when a normal user doesn't see items of the history because all changed are done by a admin12:07
ildikov_eglynn: the core issue is that the empty list describes the best that there is no alarm history, because of any reason12:08
eglynnildikov_: ok, I begin to see the light: keep it generic coz we can't be 100% sure why there's no history in any individual case12:09
eglynnBUT ... should we also apply that logic to a resource for which all samples have possibly been expired?12:10
eglynni.e. as a result of
*** nosnos has quit IRC12:10
ildikov_eglynn: ... it's a waste of time and resorces there to figure out all the different cases and send back the proper error message, if we even can do that in every case...12:11
*** promulo has quit IRC12:11
ildikov_eglynn: ... it's better to me to apply this for resources as well12:11
ildikov_eglynn: ... in the example that you've sent one particular resource was requested with it's id12:12
ildikov_eglynn: ... so in case of get all it works fine12:12
eglynnildikov_: yep, which is just asking for history for an individual alarm, no?12:13
eglynn... just *like12:13
ildikov_eglynn: ... the get_one is a special case there, if there isn't any resource than there isn't anything to respond12:13
ildikov_eglynn: ... no :)12:13
ildikov_eglynn: ... it would be the same as requesting one particular alarm12:13
ildikov_eglynn: ... but we can have alarm history for deleted alarms12:14
ildikov_eglynn: ... it's an unlucky situation how the endpoint URL looks like in case of alarm history12:15
ildikov_eglynn: ... but the query itself on that endpoint is a get_all request12:15
eglynnildikov_: sure about the deleted alarm history, that's deliberate ... the alarm is gone but the history still occurred12:16
ildikov_eglynn: ... I cannot see any benefit of having an error message here, but more problems because of it12:16
eglynnildikov_: ... ok, that's a fair point12:17
ildikov_eglynn: ... I'm not the expert of REST, so maybe someone who is, could mention some professional reasons and not just bad feelings like me :)12:17
eglynnildikov_: ... TBH I don't fully the distinction between the empty history case and requesting a un-found resource that may never have existed *or* may have existed but been evicted on sample expiry12:19
DinaBelovasorry, was afk, reading the logs about the 404 problems12:19
ildikov_eglynn, sileht: ... so as a summarization, I still think that the bug is invalid12:20
ildikov_eglynn: ... in the example there I do not know how we could return nothing on that endpoint, which should return one resource12:20
silehtildikov_, like you can see on the bug report history I have change my mind many times, but I prefer keep the current behavior12:21
ildikov_eglynn: ... in case of the alarm history, you request for all the stored alarm history, with a default query, where the alarm_id is something and the user can specify additional query fields12:21
ildikov_sileht: ... yes, I saw that, it is why I didn't mark it invalid and stated in the review also that it should be discussed12:22
ildikov_eglynn: ... so these two cases are different, as the behavior/purpose of these two endpoints are different12:22
* eglynn reads back over bug history ...12:23
ildikov_eglynn: ... for the get_one endpoint I do not know how to return nothing as there is no empty list there12:23
* DinaBelova trying to get final opinion12:23
* ildikov_ will be back in a minute12:24
eglynnDinaBelova: OK, I'm swayed by sileht & ildikov's arguments, particularly the final point on get_one versus get_all12:26
*** ildikov has joined #openstack-ceilometer12:26
eglynn(... i.e. there's being no analogue of [] to return from a get_one-style query)12:26
DinaBelovaildikov_, sileht, eglynn so about some final opinion about this moment - this exact moment is not the bug here12:26
*** _nadya_ has quit IRC12:27
DinaBelovathat's just a way of presenting the things to the user - so if no history was found, well - just [] will be returned and that's it12:27
eglynnDinaBelova: yes, that about sums it up12:27
DinaBelovaokay, in this case let's mark the bug as "not a bug" and I'll abandon the change not to be in the list of the changes)12:28
DinaBelovaildikov_, sileht, eglynn - ok with that?12:28
*** ildikov_ has quit IRC12:28
ildikoveglynn: ... kind of no analogue, so the response of get_one there is correct, but here (for alarm history or any other get_all request) we have a valid query response, which is an empty list12:28
DinaBelovaildikov, ok, got your idea, I think that's quite logical here12:28
silehtDinaBelova, ok for me12:29
eglynnDinaBelova: yes, please do that re. the launchpad housekeeping12:29
ildikovDinaBelova: I'm ok with that and if this will be the final decision, I'm sorry that you worked with this bug that much12:29
* DinaBelova abandoning12:29
ildikovDinaBelova: thanks, and sorry again :(12:29
DinaBelovaildikov, np - I think several hours is ok to make launchpad cleaner)))12:29
eglynnDinaBelova: ... nontheless, thanks for your attention to detail on this12:29
DinaBelovaand to decide if that's the bug or not))12:29
ildikovDinaBelova, eglynn, sileht: thanks all for the time for clarifying this12:30
* eglynn needs to eat after all that ;)12:30
ildikovDinaBelova: I can add a comment to the bug and link the log about this discussion, I think you can also set the bug to invalid, but I will if you cannot12:31
ildikoveglynn: ... bon apetite ;)12:31
*** eglynn is now known as eglynn-lunch12:31
ildikoveglynn-lunch: : ... you're the luckyer, I will have a half bloody meeting right now :)12:31
DinaBelovaI've abandoned the change and marked invalid with some comments :)12:33
ildikovDinaBelova: thanks :)12:34
silehtDinaBelova, thx12:35
*** _nadya_ has joined #openstack-ceilometer12:35
*** alexpilotti has quit IRC12:35
DinaBelovaildikov, sileht, eglynn-lunch (heh, funny to write *-lunch :)), btw - I have mentioned interesting moment while investigating the working process of the ceilometer rpc working >>> WARNING ceilometer.publisher.rpc [-] Failed to publish 5125 samples, queue them <<< - it looks like that while tests running messages are collected in some way and queued because of it, I'd like to investigate this moment :)12:39
DinaBelovaare you ok with it?12:39
*** _nadya_ has quit IRC12:40
*** alexpilotti has joined #openstack-ceilometer12:41
*** _nadya_ has joined #openstack-ceilometer12:44
silehtDinaBelova, yes you can take a look to the swift middleware tests12:44
silehtDinaBelova, it does this kinds of negative tests12:44
DinaBelovaokay, thanks12:44
*** _nadya_ has quit IRC12:44
*** fnaval has joined #openstack-ceilometer12:45
silehtDinaBelova, This behavior is disable by default , but for swift you can want to postpone publishing of samples if rabbitmq is not avialable to not block swift request (but you have to deal with the memory queue instead :p)12:46
*** jdob has joined #openstack-ceilometer12:48
*** promulo has joined #openstack-ceilometer13:08
*** prad has joined #openstack-ceilometer13:08
*** nekron99 has quit IRC13:18
*** nati_ueno has quit IRC13:21
*** gordc has joined #openstack-ceilometer13:25
*** promulo has quit IRC13:32
*** fnaval has quit IRC13:39
*** fnaval has joined #openstack-ceilometer13:39
*** joesavak has joined #openstack-ceilometer13:42
*** mengxd has joined #openstack-ceilometer13:42
*** fnaval has quit IRC13:44
ildikovgordc: hi13:48
gordcildikov: hi, whatsup?13:49
ildikovgordc: I commented on the ceilo client patch13:51
ildikovgordc: I think it would be better to revert the changes in that job, I asked eglynn-lunch about this earlier, but then this patch lost the priority as Jenkins needed some time to finish with the tests :)13:52
gordcildikov: hmm... do you know if gate changed because of this bug?13:54
*** fnaval has joined #openstack-ceilometer13:54
*** promulo has joined #openstack-ceilometer13:54
DinaBelovafolks, o/ I've noticed small problem with the ceilometer devstack - - I've proposed small fix for that13:54
gordcildikov: just wondering because if it was for another reason, they may change the non-voting later and we're broken again... i'll look at infra patches to see.13:54
ildikovgordc: the gate is set to non-voting because of that bug, I made the fix for, according to my knowledge13:56
gordcildikov: ah i see. that's fine with me then.13:56
ildikovgordc: not just ceilo client was affected and when I saw that the job is non-voting, it seemed to be better to revert that temporary fix13:58
ildikovgordc: cool, thanks13:58
*** fnaval has quit IRC13:59
*** thomasem has joined #openstack-ceilometer14:00
*** tongli has joined #openstack-ceilometer14:03
*** eglynn-lunch is now known as eglynn14:05
*** rwsu has joined #openstack-ceilometer14:14
*** flwang has joined #openstack-ceilometer14:23
*** julim has joined #openstack-ceilometer14:26
*** fnaval has joined #openstack-ceilometer14:27
*** Alexei_987 has joined #openstack-ceilometer14:28
gordcildikov: was there ever a plan to include groupby functionality in complex query?14:48
*** bada has joined #openstack-ceilometer14:48
ildikovgordc: we plan to extend complex query functionality for statistics14:49
gordcildikov: ok... but no plans to add groupby for complex query against samples?14:50
ildikovgordc: hmm, I'm not sure I get what you mean14:51
ildikovgordc: do you have an example?14:51
gordcildikov: hmm. let me take a look at use case... maybe it's covered by statistics.14:52
ildikovgordc: I have a session proposal for statistics, if you have a UC that is valid, doable, but not implemented yet14:54
ildikovgordc: I'm also open for complex query improvements, I just not clearly see how groupby could be used for samples query14:55
ildikovgordc: but maybe the reason for this is that we're close here to the end of the working day :)14:55
gordcildikov: yeah. that's why i'm asking. i had someone ask me about grouping by resource last week. can't remember exact details.14:56
gordcildikov: i'll ping you again if i remember and if its not covered by complex query for statistics.14:56
ildikovgordc: ok, cool, if you have a pastebin with an example or something like this, that would be the best, but anyway, ping me if you have more details :)14:57
*** shakamunyi has joined #openstack-ceilometer14:59
gordcildikov: ah got it... they wanted the latest sample of a specific meter for each resource... so right now they've been calling v2/meters/<meter_name>?q.field=resource_id&q.value=resourceId&limit=1 for each resourceId (multiple calls). i couldn't think of a way to do this one call... can you?15:02
ityaptineglynn, hi! May you take a look into our collector performance document, if you have time? Doc's url:
eglynnityaptin: will do!15:02
*** shakamunyi has quit IRC15:02
*** shakamunyi has joined #openstack-ceilometer15:03
*** idegtiarov has quit IRC15:04
*** cdent has joined #openstack-ceilometer15:05
ildikovgordc: hmm, give me a few sec15:05
*** mengxd has quit IRC15:05
ildikovgordc: I'm not an SQL expert, so it does not seem to be an easy question on the first place :)15:10
*** julim has quit IRC15:11
gordcildikov: agreed. maybe outside of the scope of our api.15:11
ildikovgordc: if it requires subquery, which it seems to do, it makes the problem complex, maybe more complex, than we would like it to be15:11
ildikovgordc: we were thinking about subqueries earlier, but we decided to skip it, because of the complexity it would bring into the system15:14
ildikovgordc: I'm not against talking a few words about this on the summit, if we will have a half session for CQ for statistics15:16
*** gordc has quit IRC15:16
ildikovgordc: there can be other opinions or someone with more SQL experience15:16
*** prad has quit IRC15:30
*** prad has joined #openstack-ceilometer15:32
*** gordc has joined #openstack-ceilometer15:33
*** _cjones_ has joined #openstack-ceilometer15:35
gordcildikov: sorry about that, internet dropped. i don't believe it requires a subquery... but i'm not sure it would work regardless because we'd need some mechanism to drop ungrouped columns.15:40
gordcwe can leave it off the table for no  i think.15:41
ildikovgordc: I saw that, I hoped you will read the logs as I got distracted a bit in the meantime :)15:44
ildikovgordc: I will think about it, but I cannot see a way currently without subquery or additional logic in the db driver15:45
gordcildikov: cool cool15:49
openstackgerritA change was merged to openstack/ceilometer: Disable reverse dns lookup
*** Ruetobas has quit IRC16:01
*** shakamunyi has quit IRC16:03
*** Ruetobas has joined #openstack-ceilometer16:03
*** ildikov has quit IRC16:03
*** dperaza has joined #openstack-ceilometer16:04
*** shakamunyi has joined #openstack-ceilometer16:05
*** Ruetobas has quit IRC16:08
DinaBelovaeglynn, btw, are you ok with current version of the ?16:10
*** jsavak has joined #openstack-ceilometer16:12
*** joesavak has quit IRC16:13
*** Ruetobas has joined #openstack-ceilometer16:13
*** joesavak has joined #openstack-ceilometer16:15
*** jsavak has quit IRC16:18
eglynnDinaBelova: thanks ... on a call now, will look when that finishes16:18
DinaBelovaeglynn, np, it's you who I should thank)16:18
*** Anju_ has joined #openstack-ceilometer16:22
*** changbl has quit IRC16:24
*** erecio has joined #openstack-ceilometer16:33
*** cdent has quit IRC16:34
*** cdent has joined #openstack-ceilometer16:36
*** fnaval has quit IRC16:39
*** fnaval has joined #openstack-ceilometer16:39
*** fnaval has quit IRC16:44
*** fnaval has joined #openstack-ceilometer16:45
eglynnityaptin: an interesting doc!16:50
eglynnityaptin: ... couple questions16:51
eglynnityaptin: ... what exactly does the "Delta time" on the x-axis of the single collector graph actually represent?16:52
openstackgerritOpenStack Proposal Bot proposed a change to openstack/ceilometer: Updated from global requirements
eglynn(i.e. delta between what & what?)16:52
eglynnityaptin: ... also any thoughts on explaining the saw-tooth pattern for hbase in the 3-collector case as the message per second is ramped up?16:53
openstackgerritgordon chung proposed a change to openstack/ceilometer: refactor sql backend for write speed
openstackgerritOpenStack Proposal Bot proposed a change to openstack/python-ceilometerclient: Updated from global requirements
*** nacim has quit IRC17:03
ityaptineglynn, o/. let me clarify these moments: 1. Delta time - time from test starting) 2. I think it's happens because connections pool is empty and samples wait next free connection.17:04
*** nati_ueno has joined #openstack-ceilometer17:06
eglynnityaptin: ... are the generated messages drip-fed into the test over time, or all generated together and injected from the get-go?17:07
ityaptineglynn: messages are generated one by one with improving rate from 50 m/s to 1000 m/s during the test17:09
eglynnityaptin: fair enough, thanks for the clarification17:10
eglynnityaptin: one other quick point, have you captured the mongo "Overflow sort stage buffered data usage ..." issue seen in a bug on LP?17:11
ityaptineglynn: collector + hbase cluster can process about 300 m/s and then rate exceed this number, messages accumulate in queue17:11
*** nati_uen_ has joined #openstack-ceilometer17:12
eglynnityaptin: (that mongo error could be indicative of an issue with our indexing ... along the lines of )17:12
eglynnityaptin: ... so it would be good for us to dig in and get to the bottom of it17:12
*** nati_ueno has quit IRC17:15
*** eglynn is now known as eglynn-afk17:15
ityaptinI have not captured this bug before.17:16
ityaptinI may take a look on this problem and try to catch it.17:19
*** shakamunyi has quit IRC17:32
*** shakayumi has joined #openstack-ceilometer17:32
*** eglynn-afk is now known as eglynn17:36
eglynnityaptin: thanks, that would be great!17:37
*** shakayumi has quit IRC17:40
*** lsmola has quit IRC17:51
*** cdent has quit IRC18:00
*** ildikov has joined #openstack-ceilometer18:02
*** promulo has quit IRC18:10
openstackgerritDina Belova proposed a change to openstack/ceilometer: Describe storage backends in the collector installation guide
DinaBelovaeglynn, gordc - have updated docs change ^^ Please take a look)18:14
eglynnDinaBelova: thanks, looking ...18:14
DinaBelovaI did not fix all comments, please take a look on my answers :)18:14
*** jaypipes has quit IRC18:15
*** sdake_ has joined #openstack-ceilometer18:18
*** shakamunyi has joined #openstack-ceilometer18:19
DinaBelovaeglynn, you here?18:45
*** promulo has joined #openstack-ceilometer18:45
eglynnDinaBelova: yep18:45
DinaBelovaI'm about your comment to the
DinaBelovas/Ceilometer services/central agent, notification agent, and API service/18:46
eglynnDinaBelova: darn typo!18:46
eglynnDinaBelova: s/central agent/collector/18:46
eglynnDinaBelova: sorry!18:46
DinaBelovaoh, thank you :)18:46
DinaBelovaI was having this face -> 0_o18:46
eglynnDinaBelova: I can well imagine :)18:47
*** _nadya_ has joined #openstack-ceilometer18:47
openstackgerritDina Belova proposed a change to openstack/ceilometer: Describe storage backends in the collector installation guide
DinaBelovaeglynn, ^^ :)18:56
eglynnDinaBelova: looking ...18:56
DinaBelovaildikov, you too :) ^^18:56
ildikovDinaBelova: I'm faster, already looking ;)18:57
openstackgerritA change was merged to openstack/ceilometer: cleanup virt pollster code
*** changbl has joined #openstack-ceilometer19:03
ildikovDinaBelova: for the first quick look it seems to be ok, I will wait for Jenkins to see the "built picture"19:04
DinaBelovaok, cool)19:04
ildikovDinaBelova: I mean that the links are working and how the final formatting looks, etc19:05
ildikovDinaBelova: so nothing serious :)19:05
*** nati_uen_ has quit IRC19:08
*** jdob_ has joined #openstack-ceilometer19:10
*** tongli has quit IRC19:12
*** nati_ueno has joined #openstack-ceilometer19:14
*** eglynn has quit IRC19:20
openstackgerritDina Belova proposed a change to openstack/ceilometer: Describe storage backends in the collector installation guide
*** jaypipes has joined #openstack-ceilometer19:30
*** nati_ueno has quit IRC19:30
*** julim has joined #openstack-ceilometer19:41
*** _nadya_ has quit IRC19:49
*** jsavak has joined #openstack-ceilometer20:01
*** joesavak has quit IRC20:04
*** nati_ueno has joined #openstack-ceilometer20:06
*** _nadya_ has joined #openstack-ceilometer20:09
*** nati_ueno has quit IRC20:11
*** _nadya_ has quit IRC20:21
*** nati_ueno has joined #openstack-ceilometer20:24
*** julim has quit IRC20:33
*** joesavak has joined #openstack-ceilometer20:37
*** prad_ has joined #openstack-ceilometer20:39
*** prad has quit IRC20:41
*** prad_ is now known as prad20:41
*** jsavak has quit IRC20:41
*** jdob_ has quit IRC20:44
*** yjiang5 is now known as yjiang5_away21:00
*** Anju_ has quit IRC21:01
*** erecio has quit IRC21:07
*** thomasem has quit IRC21:13
*** shakamunyi has quit IRC21:17
*** jdob has quit IRC21:17
openstackgerritRob Raymond proposed a change to openstack/ceilometer: Replace hard coded WSGI application creation
*** gordc has quit IRC21:32
*** joesavak has quit IRC21:41
*** changbl has quit IRC21:47
*** nati_ueno has quit IRC21:59
*** nati_ueno has joined #openstack-ceilometer22:15
*** shakamunyi has joined #openstack-ceilometer22:39
*** shakamunyi has quit IRC22:52
openstackgerritRob Raymond proposed a change to openstack/ceilometer: Replace hard coded WSGI application creation
*** shakamunyi has joined #openstack-ceilometer22:59
*** fnaval has quit IRC23:02
*** fnaval has joined #openstack-ceilometer23:18
openstackgerritA change was merged to openstack/ceilometer: Describe storage backends in the collector installation guide
openstackgerritA change was merged to openstack/ceilometer: Improve combination rule validation
*** jaypipes has quit IRC23:54

Generated by 2.14.0 by Marius Gedminas - find it at!