Thursday, 2015-07-30

« Wednesday, 2015-07-29 Index Friday, 2015-07-31 »
*** adurbin_ has quit IRC00:01
*** _cjones_ has quit IRC00:21
*** thorst1 has joined #openstack-ceilometer00:22
*** thorst1 has quit IRC00:24
openstackgerritgordon chung proposed openstack/ceilometer: Revert "Revert "remove instance:<flavor> meter""  https://review.openstack.org/19250300:30
*** edmondsw has quit IRC00:34
*** vishwanathj has joined #openstack-ceilometer00:41
*** vishwanathj has quit IRC00:46
*** vishwanathj has joined #openstack-ceilometer00:46
*** jwcroppe has joined #openstack-ceilometer00:49
*** vishwanathj has quit IRC00:55
*** h00327910__ has joined #openstack-ceilometer00:59
*** rbak has joined #openstack-ceilometer01:02
*** rbak has quit IRC01:36
*** llu_linux is now known as llu01:41
*** Kennan has quit IRC02:02
*** Kennan has joined #openstack-ceilometer02:02
*** ankita_wagh has joined #openstack-ceilometer02:05
*** ankita_wagh has quit IRC02:10
*** ankita_wagh has joined #openstack-ceilometer02:10
openstackgerritliusheng proposed openstack/aodh: Add alembic support for aodh  https://review.openstack.org/20645902:40
*** changbl has quit IRC02:47
*** liusheng has joined #openstack-ceilometer03:00
*** changbl has joined #openstack-ceilometer03:06
*** julim has quit IRC03:32
*** jwcroppe has quit IRC03:33
*** jwcroppe has joined #openstack-ceilometer03:34
*** jwcroppe has quit IRC03:38
*** deepthi has joined #openstack-ceilometer03:55
*** yprokule has joined #openstack-ceilometer04:25
*** htruta_ has quit IRC04:26
*** fawadkhaliq has joined #openstack-ceilometer04:27
*** fawadkhaliq has quit IRC04:52
openstackgerritMehdi Abaakouk (sileht) proposed openstack/gnocchi: gate: fixup gabbi live tests  https://review.openstack.org/20651004:53
openstackgerritMehdi Abaakouk (sileht) proposed openstack/aodh: api: fix alarm deletion and update  https://review.openstack.org/20714004:54
openstackgerritMehdi Abaakouk (sileht) proposed openstack/aodh: tests: ensure gabbi live test fail  https://review.openstack.org/20729104:54
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: tests: add an integration test  https://review.openstack.org/20713705:02
*** fawadkhaliq has joined #openstack-ceilometer05:05
*** Ephur has joined #openstack-ceilometer05:27
*** Ephur has quit IRC05:33
*** deepthi has quit IRC05:33
*** lexholden has quit IRC05:40
*** lsmola has joined #openstack-ceilometer05:54
*** Kennan2 has joined #openstack-ceilometer06:03
*** Kennan has quit IRC06:03
*** smyz709 has joined #openstack-ceilometer06:23
* sileht sigh, gate is so slow....06:26
*** smyz709 has quit IRC06:29
*** deepthi has joined #openstack-ceilometer06:31
*** david-lyle has quit IRC06:43
*** david-lyle has joined #openstack-ceilometer06:49
*** jiaxi_ has joined #openstack-ceilometer06:49
jiaxi_Hello06:50
jiaxi_What is the function of coordination.py that live in the main directory ?06:50
*** _nadya_ has joined #openstack-ceilometer06:52
*** ankita_w_ has joined #openstack-ceilometer06:57
*** ankita_wagh has quit IRC06:57
*** Ala has joined #openstack-ceilometer07:04
*** belmoreira has joined #openstack-ceilometer07:10
*** _nadya_ has quit IRC07:16
*** _nadya_ has joined #openstack-ceilometer07:16
*** _nadya_ has quit IRC07:20
*** vishwanathj has joined #openstack-ceilometer07:23
*** safchain has joined #openstack-ceilometer07:25
* sileht sigh, gate is dead, again07:26
*** vishwanathj has quit IRC07:28
*** vishwanathj has joined #openstack-ceilometer07:29
*** ankita_w_ has quit IRC07:29
*** jwcroppe has joined #openstack-ceilometer07:29
*** ankita_wagh has joined #openstack-ceilometer07:29
*** ankita_wagh has quit IRC07:33
*** cdent has joined #openstack-ceilometer07:34
*** guy_rodrigue has joined #openstack-ceilometer07:36
*** guy_rodrigue has left #openstack-ceilometer07:36
-openstackstatus- NOTICE: Our CI system is broken again today, jobs are not getting processed at all.07:39
*** ChanServ changes topic to "Our CI system is broken again today, jobs are not getting processed at all."07:39
-openstackstatus- NOTICE: CI system is broken and very far behind. Please do not approve any changes for a while.07:49
*** ChanServ changes topic to "CI system is broken and very far behind. Please do not approve any changes for a while."07:49
*** tshtilma has joined #openstack-ceilometer07:50
openstackgerritChris Dent proposed openstack/ceilometer: Drop out-of-time-sequence rate of change samples  https://review.openstack.org/20715307:53
openstackgerritMehdi Abaakouk (sileht) proposed openstack/aodh: tests: ensure gabbi live test fail  https://review.openstack.org/20729107:54
openstackgerritMerged openstack/aodh: Refactor alarm scenario tests (RuleThreshold)  https://review.openstack.org/20443908:29
openstackgerritMerged openstack/aodh: Imported Translations from Transifex  https://review.openstack.org/20690708:29
*** afazekas has joined #openstack-ceilometer08:33
*** jwcroppe has quit IRC08:37
*** jwcroppe has joined #openstack-ceilometer08:38
*** jwcroppe has quit IRC08:42
*** yassine has joined #openstack-ceilometer08:43
*** jiaxi_ has quit IRC08:44
openstackgerritliusheng proposed openstack/ceilometer: Make the partition coordinator log more readable  https://review.openstack.org/20733808:44
*** fawadkhaliq has quit IRC08:53
*** ChanServ changes topic to "OpenStack Telemetry (Ceilometer) | https://wiki.openstack.org/wiki/Ceilometer"08:58
-openstackstatus- NOTICE: CI is back online but has a huge backlog. Please be patient and if possible delay approving changes until it has caught up.08:58
openstackgerritJulien Danjou proposed openstack/gnocchi: InfluxDB storage driver  https://review.openstack.org/16540709:12
*** jwcroppe has joined #openstack-ceilometer09:20
*** jwcroppe has quit IRC09:21
*** jwcroppe has joined #openstack-ceilometer09:21
*** jwcroppe_ has joined #openstack-ceilometer09:23
*** jwcroppe_ has quit IRC09:23
*** jwcroppe_ has joined #openstack-ceilometer09:24
*** fawadkhaliq has joined #openstack-ceilometer09:25
*** fawadkhaliq has quit IRC09:26
*** fawadkhaliq has joined #openstack-ceilometer09:26
*** jwcroppe has quit IRC09:27
*** jwcroppe_ has quit IRC09:28
*** ig0r_ has joined #openstack-ceilometer09:32
*** jwcroppe has joined #openstack-ceilometer09:38
*** jwcroppe has quit IRC09:40
*** _nadya_ has joined #openstack-ceilometer09:45
*** alexpilotti has joined #openstack-ceilometer10:01
*** deepthi has quit IRC10:10
*** exploreshaifali has joined #openstack-ceilometer10:11
*** kbyrne has quit IRC10:21
*** kbyrne has joined #openstack-ceilometer10:23
*** jwcroppe has joined #openstack-ceilometer10:32
*** _nadya_ has quit IRC10:55
*** _nadya_ has joined #openstack-ceilometer10:55
*** _nadya_ has quit IRC11:00
*** jwcroppe has quit IRC11:02
*** dikonoor has joined #openstack-ceilometer11:03
*** rbak has joined #openstack-ceilometer11:03
openstackgerritliusheng proposed openstack/aodh: Keep alarm other attrs constantly after evaluating an alarm  https://review.openstack.org/20573211:24
*** rbak has quit IRC11:30
openstackgerritIlya Tyaptin proposed openstack/ceilometer: [MongoDB] Refactor indexes for meter and resources  https://review.openstack.org/20740911:32
*** pauldix has joined #openstack-ceilometer11:32
*** fabian4 has joined #openstack-ceilometer11:36
*** pauldix has quit IRC11:39
cdentmariadb-slow.log just filled up the disk on one of my openstack test machines11:40
cdentthat's irony or something11:40
*** thorst1 has joined #openstack-ceilometer11:44
*** gordc has joined #openstack-ceilometer11:54
gordcanyone notice any blockers or have a reason why i should not tag liberty-211:59
gordcthe gate seems dead so i'm probably going to have to call it11:59
gordccdent: i'm leaving your batch item for l-312:02
*** exploreshaifali has quit IRC12:02
gordccdent: cool if we close bp after that merges?12:02
*** fawadk has joined #openstack-ceilometer12:02
*** fawadkhaliq has quit IRC12:03
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Convert identity, sahara and volume to meters yaml  https://review.openstack.org/20303112:05
gordcildikov: sileht: jd__ : llu: speak now or forever hold your peace ^... or hold for a few weeks.12:06
*** fawadk has quit IRC12:10
silehtgordc, the gate will stop you, so ... ;)12:11
*** _nadya_ has joined #openstack-ceilometer12:11
gordcsileht: lol yeah... i give you guys false hope.12:12
gordcwhen i say hold for a few weeks i mean until gate clears.12:12
*** _nadya_ has quit IRC12:18
*** jaosorior has joined #openstack-ceilometer12:19
*** _nadya_ has joined #openstack-ceilometer12:19
*** edmondsw has joined #openstack-ceilometer12:19
*** _nadya_ has quit IRC12:20
*** jwcroppe has joined #openstack-ceilometer12:23
*** jasonamyers has joined #openstack-ceilometer12:26
jd__gordc: go ahead12:34
*** liusheng has quit IRC12:37
*** liusheng has joined #openstack-ceilometer12:37
openstackgerritMehdi Abaakouk (sileht) proposed openstack/gnocchi: gate: fixup gabbi live tests  https://review.openstack.org/20651012:38
gordcjd__: cool cool. in process of tagging.12:38
*** fabian4 has quit IRC12:45
openstackgerritMerged openstack/aodh: Keep alarm other attrs constantly after evaluating an alarm  https://review.openstack.org/20573212:48
cdentgordc i'm happy12:50
gordclol, it took 14hrs for prad to find out his patch failed gate. i guess we have to start testing locally.12:51
jasonamyershaha12:51
* gordc debates whether to recheck or abandon a failed gate check.12:53
*** pauldix has joined #openstack-ceilometer12:55
*** htruta has joined #openstack-ceilometer12:55
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: gnocchi: retry with a new token on 401  https://review.openstack.org/20656012:58
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: gnocchi: reduce the number of patch to gnocchi API  https://review.openstack.org/20258912:58
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Adds batch dispatcher  https://review.openstack.org/20334912:58
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: gnocchi: concurrently post resource/measures  https://review.openstack.org/20355312:58
silehtif people have some review times: https://review.openstack.org/#/c/207140/ and https://review.openstack.org/#/c/207137/12:59
silehtI would test that into a gate job now12:59
sileht(the futur gate job: https://review.openstack.org/#/c/207298/)12:59
cdentgordc: I've just had a friendly remember from QA that especially in the case of notification related test failures gratitous rechecks aren't very helpful, better to rewrite the failing test so it provides more effective debugging. I said I'd pass it along. So there ya go.13:00
gordccdent: what have we been rechecking?13:01
gordci sometimes recheck just to make sure i don't see any random failures.13:01
*** safchain has quit IRC13:04
gordccdent: is this in regards to current gate backup? ie. don't do anything right now?13:04
cdentI rechecked something, but I think the point is that there is no such thing as a random failure when notification-related test fails13:04
cdentthere is a failure which the test should be able to point at13:04
cdentand eventually lead to a better test and may eventualy lead to a real fix13:05
cdentfor example this case: 75776440413:05
cdentwhoops13:05
cdentthis case: http://logs.openstack.org/48/206148/4/check/gate-ceilometer-python27/e0283e1/console.html#_2015-07-28_17_59_48_33513:05
gordcbank pin?13:05
cdentno, some sort of userid for something other13:05
cdentjust the username though13:05
cdentanyway, in that failure the error message is completely useful13:06
gordcsarcasm?13:06
cdentand provides no information that would allow for any diagnostic investigation13:06
cdentyes13:06
cdentor I might be unable to type and meant useless13:06
gordci really have no idea why that failed. that check has no relation to timing.13:07
cdentindeed13:08
cdentand the message helps not one wit13:09
gordcok. i thought you were talking about tempest... and i just assume tempest has a 15% fail rate built in13:09
* cdent thinks tempest is a bad of pooh and pays about zero attention to it13:09
cdentdammit! bag!13:10
*** pauldix has quit IRC13:10
*** julim has joined #openstack-ceilometer13:10
gordchope your qa connections don't read taht.13:10
cdentseriously, one of the reasons I'm so keen on in-tree-functional and integration testing is because tempest is junk13:10
cdentI think there would be some agreement from some people13:10
*** ig0r_ has quit IRC13:10
gordci actually understand the failure you linked... not why, but what is failing... probably because i wrote it. :)13:11
cdent:)13:11
cmysternope we havn't read that ;)13:11
*** fawadkhaliq has joined #openstack-ceilometer13:11
*** pauldix has joined #openstack-ceilometer13:11
* cdent gives cmyster a cookie13:11
gordcdammit! bridge burnt13:12
gordccdent: you were are only connect!13:12
gordcour&13:12
cdentI think I have given you my typing disease gordc13:13
gordcyeah. or just can't type... let's go with "blame it on you."13:13
cdenteverything is my fault13:13
cdentexcept that test error message13:14
cdentthat's on you13:14
gordcmeh.13:14
gordci think i could fix it... but it involves duplicating code...13:14
* cdent does not believe in DRY in tests13:14
*** safchain has joined #openstack-ceilometer13:14
*** exploreshaifali has joined #openstack-ceilometer13:15
gordc"fix it" == "limit chances of randomness"13:15
cdentI'd be very curious to see what that is. Or even just to hear about it.13:15
cdent(It's nice so much code is moving lately, but it sure makes for a lot to review)13:16
*** fawadkhaliq has quit IRC13:16
gordccdent: what is failing is that hmac got called apparently when it's not suppose to be.13:16
*** ig0r_ has joined #openstack-ceilometer13:16
cdenthow could that be non-deterministic?13:17
cdentseems kind of straightforward13:17
gordcwe patch it at the beginning of test but it really can be done later.13:17
gordcthat said, i don't know what in the steps before is calling it... nothing should be13:17
cdentthere's that should word13:17
gordcit could be the tests are running segregated.13:17
gordcaren't*13:18
gordci'll change it for now. that was the one and only time i've seen that failure.13:19
*** pauldix has quit IRC13:20
cdentgordc: I've found this style: mock.method.assert_called_with a bit more readable and reliable and informative (since it requires args)13:20
gordccdent: how do i do the opposite? method is never called.13:21
cdentassert_not_called13:22
*** pauldix has joined #openstack-ceilometer13:22
gordcwith anything.13:22
*** rbrooker has joined #openstack-ceilometer13:23
*** pauldix has quit IRC13:23
gordcadmittedly, testing negative set is not really proving much...13:24
*** pauldix has joined #openstack-ceilometer13:26
gordci'm going to drop that test and just test the method.13:26
*** kbyrne has quit IRC13:26
*** jwcroppe has quit IRC13:26
cdentdoes that bring us any closer to understanding why that one failure happened?13:27
*** jwcroppe has joined #openstack-ceilometer13:27
*** kbyrne has joined #openstack-ceilometer13:27
thorst1Hey.  I'm doing some testing with the powervm compute inspector...but I'm having trouble getting the inspector to load.  While debugging, I came across this change:  https://review.openstack.org/#/c/203319/   Wondering if its OK for me to open a bug and update that to include powervm?13:27
gordc*shrugs* i'm testing that a native python method isn't called. i'm assuming there's any number of things that could call it.13:28
*** pauldix has quit IRC13:29
gordccdent: https://review.openstack.org/#/c/16785313:29
gordccontext^13:29
cdentthorst1: interesting. your trouble suggests that that way of controlling inspectors is kind of limiting13:30
gordcthorst1: sure thing. probably doesn't need a bug but either or13:30
cdentyour external plugin ought to be the way of the future13:30
thorst1yeah, if OK without bug I can just propose an update.13:30
cdentbut it should also require a core change in order to be usable13:30
*** ig0r_ has quit IRC13:31
thorst1but yeah...with an external plugin, I thought the intent was that it wouldn't need changes to core ceilometer?13:31
thorst1I'm not sure that this is in fact the issue...just seems limiting to only have the in-tree ones listed there.13:31
gordcthorst1: cdent: i'm sort of leaning towards that the list is what ceilometer considers as validated drivers.13:31
gordcbut that's just me being political.13:32
*** jwcroppe has quit IRC13:32
gordcaside from that. it doesn't really add much value.13:32
cdentthorst1: it does look like that not allowing powervm would break your ability to load stuff13:33
thorst1yeah...it limits you to ONLY those choices13:33
cdentgordc: are we in the business of validating things or enabling things?13:33
thorst1I think13:33
cdentthorst1: that appears to be the case, yeah13:33
gordccdent: both.13:34
gordcto a certain extent13:35
* cdent wonders if gordc is a crypto-fascist13:35
*** tshtilma has quit IRC13:35
thorst1so I guess the question is...with the external plugin future, should it be a list of choices?  If so, I could add powervm to it.  If not, I could update it so the validated choices are in the help text.13:36
gordc... is there a difference between fascist and facist? because if so, i've been using it wildly inappropriately.13:36
cdenta "facist" would be a "face-ist" and probably not what you mean13:37
cdentthorst1: I think that change should be reverted, but I'm not sure I'm in the majority on this, thus my current badgering of gordc13:38
gordcthorst1: i think we need quick debate. whether to add to it choices or drop choices completely.13:38
thorst1seems like a good approach.  I can work around it for now, maybe something for the ceilometer meeting in a little over an hour?13:39
gordcsure.13:39
gordcadding to agenda13:39
cdentthorst1: that's incredibly sensible of you, I'm not sure that's allowed ;)13:39
thorst1lol13:40
gordcwe usually just grab bats and see who last person standing is.13:40
thorst1ooo, that'd be a fun booth in Tokyo13:40
cdentgordc is only ptl because he has a lot of baseball experience so knows how to use a bat13:40
gordctrue story.13:40
gordci'm the only north american so really i was only one with a bat.13:41
thorst1I went to the batting cages a few months back...I think I managed contact with 2 or 3 of the pitches...13:41
gordclucky for you, you might be in running for ptl.13:41
thorst1eek13:42
gordci think going to batting cages and not getting hit by a ball from machine or foul is a win.13:43
*** fabian4 has joined #openstack-ceilometer13:50
*** fabian4 has quit IRC13:51
*** jaypipes has joined #openstack-ceilometer13:53
*** exploreshaifali has quit IRC14:00
cdentjd__, sileht: is there a reason there is not history option GETting one resource?14:04
cdentyou can either get history when doign a get_all14:04
cdentor do POST of a complex search14:04
jd__cdent: I don't think there is a reason14:04
cdentI keep wanting to do it, so I guess I'll add it?14:04
*** prad has joined #openstack-ceilometer14:05
jd__cdent: sure go ahead14:07
jd__gordc: boss I'll be AFK during the meeting14:07
gordcjd__: cool cool14:08
* cdent makes a bug as a reminder since I'm not going to do it _right_ now14:08
jd__cdent: good idea14:08
*** jiaxi_ has joined #openstack-ceilometer14:10
jiaxi_Hello,everyone14:12
gordcjiaxi_: hey14:14
*** pauldix has joined #openstack-ceilometer14:14
jiaxi_Jekins is ill ?14:14
*** Kennan2 has quit IRC14:16
gordcjiaxi_: yep. i don't have a link to confirm what is wrong though.14:16
*** Kennan has joined #openstack-ceilometer14:16
gordcyou can ask in openstack-infra but best to not bother them and let them work it out.14:17
jiaxi_gordc: You are right. They are working on it14:17
gordcf... possibly another memory leak in notification agent14:25
cdentgordc: :(14:26
gordci'm going up 10MB per min14:26
gordcat 5s polling14:26
gordci think the constant CPU load might be related to pipeline.14:28
gordcdefault one worker it will hover around 10% CPU load.14:28
gordcwith event disable, one worker, it will hover around 6%14:29
*** _nadya_ has joined #openstack-ceilometer14:32
*** ddaskal has joined #openstack-ceilometer14:32
*** ig0r_ has joined #openstack-ceilometer14:32
*** rbrooker has quit IRC14:38
cdentgordc: i'm on a 10s polling with 4 notification-agents (from master) and seeing stable memory use14:39
cdentI assume you are on one your branches?14:40
gordcmaster14:41
cdent(floating around 230/50 drs/rss)14:41
*** ccrouch has joined #openstack-ceilometer14:42
gordcrss started at ~80M, went to 113M.14:42
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Enable entry points for new declarative meters  https://review.openstack.org/20303214:43
cdentgordc: how long have you been running?14:44
gordci've been jumping back and forth testing random branches so that might be it.14:44
gordccdent: 5 days?14:44
cdentyou've had the same notification agent process alive for 5 days?14:45
gordcadmittedly i did see collector spike at some point and i just ignored and restarted and it was fine.14:45
gordcoh no. i've had my machine on for 5 since last reboot14:45
cdenta messed up thing about python is that it never gives memory back14:46
cdentso if at some point your process gets large, it will stay large14:46
cdentpython will reuse that empty memory but not return it to the OS14:46
* gordc has faith in magic garbage collect14:46
cdentso there can be pretty significant ramp up period while the process grows as it explores all its nooks and grannies before it hits a stable period14:47
cdentmy notification-agentshave been up for nearly 4 hours14:47
cdentand they and the collectors have stable (but what I would consider high) memory use14:48
gordcwill reboot and leave system up later today.14:48
gordcwill focus on this idle CPU load item for now.14:48
cdentgordc: we're you round when I was talking to sileht about that?14:48
gordccdent: do you see that?14:49
gordccdent: possibly? something about heartbeat?14:49
cdentthe listeners sit in a tight epoll_wait tight loop which can cause some cpu usage14:49
gordcyeah i think i was there for that.14:49
cdentif you turn off the listeners the processes sti truly idle14:49
gordcor i came back and read.14:50
gordci added some comments to bug14:50
gordcafter the fact14:50
gordchttps://bugs.launchpad.net/ceilometer/+bug/147813514:50
openstackLaunchpad bug 1478135 in Ceilometer "notification agent never idles" [High,New]14:50
* cdent is not getting all the bug emails he thought he was signed up for14:50
cdenthmm, interesting14:51
cdenti shall have to try my fiddling again14:51
cdent(I'm not sure if I was looking post sample receipt)14:52
* cdent makes meeting coffee14:52
*** afazekas has quit IRC14:55
* gordc thinks it's pipeline.14:57
gordcif i disable pipelines but keep listeners it's fine. will look after meeting.14:57
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Separates event listener from sample listener  https://review.openstack.org/20749314:59
gordcfriendly reminder, we have meeting now.15:01
*** rbrooker has joined #openstack-ceilometer15:02
*** ankita_wagh has joined #openstack-ceilometer15:02
*** ig0r_ has quit IRC15:03
*** primechuck has joined #openstack-ceilometer15:03
*** eglynn has joined #openstack-ceilometer15:03
*** ig0r__ has joined #openstack-ceilometer15:04
*** logan2 has quit IRC15:08
*** ig0r__ has quit IRC15:09
*** ig0r_ has joined #openstack-ceilometer15:09
*** fawadkhaliq has joined #openstack-ceilometer15:11
*** ig0r__ has joined #openstack-ceilometer15:13
*** ig0r__ has quit IRC15:15
*** ig0r_ has quit IRC15:15
*** lakshmiS has joined #openstack-ceilometer15:19
*** vishwanathj has quit IRC15:21
*** jiaxi_ has quit IRC15:22
*** haypo has joined #openstack-ceilometer15:33
haypojd__, sileht : hi. i ran "tox -e py34" in aodh. remaining python 3 issues are non trivial (no obvious python 3 issue)15:34
haypoany idea to debug them? for example, a content type is None on a request object15:34
*** jruano has joined #openstack-ceilometer15:40
*** rbrooker has quit IRC15:45
*** alexpilotti has quit IRC15:48
cdentgordc: in typical fashion I don't understand your latest comment on the notification agent load bug15:52
*** safchain has quit IRC15:52
cdentI feel like maybe we need couples therapy or something15:52
*** ig0r_ has joined #openstack-ceilometer15:53
*** lakshmiS has quit IRC15:53
gordc... do i get a doll so i can point to where you hurt me?15:53
cdentyour the one with the baseball bat15:53
*** lakshmiS has joined #openstack-ceilometer15:54
gordccdent: true true15:54
cdentanyway I wanted to replicate what you tried but since I couldn't understand what you tried I didn't15:54
gordcbasically i turned off store_events which essentially doesn't build the event_pipeline15:54
gordcthat dropped CPU load15:54
cdentso even when you weren't sending any events, having the even pipeline present was causing load?15:55
gordcseems like it15:55
gordcor there's a difference between having store_events on and store_events off.15:55
*** jaosorior has quit IRC15:55
*** chinmay_g has quit IRC15:55
*** h00327910__ has quit IRC15:55
gordcnext i disabled setup_pipeline and skipped it in appropriate places but still keeping listeners on15:56
gordcso basically i had listeners enabled but they just tossed each notification before pipeline stage15:57
gordcwhen i did that, it would have expected cpu load when notifications arrived... but would idle at zero when nothing15:57
cdentare you savvy to strace and lsof?15:58
gordccdent: i need to google what lsof is15:59
*** rbrooker has joined #openstack-ceilometer15:59
gordcis that a good enough answer? :)15:59
cdentit tells you what filedescriptors in a process are associated with what files, sockets, etc15:59
cdentwhich is handy when you're using strace, which will will show you the syscalls of a running process15:59
cdentso if you see that is epoll_wait on filedescriptor 3 you can go to lsof and find out what that is16:00
cdentand usually know slightly more16:00
cdenthaypo: I seem to remember that as a bug in pecan or webtest or webob16:01
haypocdent: the traceback points to webtest16:01
haypo  File "/home/haypo/prog/openstack/aodh/.tox/py34/lib/python3.4/site-packages/webtest/response.py", line 499, in json16:01
haypo    if not self.content_type.endswith(('+json', '/json')):16:01
haypocdent: do you mean that the bug can be in webtest, not in aodh or how aodh uses webtest?16:02
cdenthaypo: _maybe_16:02
cdentI'm trying to refresh my memory16:02
haypocdent: the content_type of the response is None16:02
*** vishwanathj has joined #openstack-ceilometer16:04
*** thumpba has joined #openstack-ceilometer16:04
*** vishwana_ has joined #openstack-ceilometer16:05
gordccdent: after 15hr you need a recheck. tempest karma.16:05
gordcnever criticise the gods16:05
* cdent sighs16:05
cdentthat's my whole life16:05
gordclol16:06
cdentwhy be alive unless it is to challenge the gods16:06
*** rbrooker has quit IRC16:06
cdenthaypo: webtest never assigns self.content_type, but its super webob.Response does16:08
haypocdent: we told me that it's not a good thing to write a robot to automatically recheck16:08
cdent:)16:08
haypogordc: ^^16:08
*** vishwanathj has quit IRC16:08
*** lsmola has quit IRC16:09
*** rbrooker has joined #openstack-ceilometer16:09
gordci should change to full time recheck robot. i will track down all rechecks16:09
lakshmiSquestion on glance notifications. For searchlight project we will be suggesting glance team to add new data(tags) to notifications. will it impact ceilometer if its see new fields in glance notifications?16:10
gordclakshmiS: new fields? no. it will only show up in metadata and raw fields16:11
gordcif you want to do something with the new fields, you'll need to declare it in ceilometer.16:11
lakshmiSgordc: thats good news. thanks16:11
openstackgerritMehdi Abaakouk (sileht) proposed openstack/gnocchi: api: dogpile cache for getting resource  https://review.openstack.org/20752116:11
openstackgerritMehdi Abaakouk (sileht) proposed openstack/gnocchi: Don't update resource if nothing has changed  https://review.openstack.org/20752216:11
gordclakshmiS: np16:11
*** exploreshaifali has joined #openstack-ceilometer16:14
*** rbrooker has quit IRC16:14
*** lakshmiS has quit IRC16:15
*** _nadya_ has quit IRC16:17
cdenthaypo: it's probably WSME16:17
*** ankita_w_ has joined #openstack-ceilometer16:18
*** logan2 has joined #openstack-ceilometer16:19
gordccdent: ha!16:20
praddoes the intel ci do anything useful today? i rarely see it pass16:21
*** ankita_wagh has quit IRC16:22
*** ankita_w_ has quit IRC16:23
*** _cjones_ has joined #openstack-ceilometer16:23
cdentprad that failure is usually the result of network problems the intel folk are aware16:23
pradah ok16:24
*** chinmay_g has joined #openstack-ceilometer16:24
*** tshtilma has joined #openstack-ceilometer16:24
*** jaosorior has joined #openstack-ceilometer16:27
openstackgerritMehdi Abaakouk (sileht) proposed openstack/gnocchi: api: dogpile cache for getting resource  https://review.openstack.org/20752116:29
openstackgerritMehdi Abaakouk (sileht) proposed openstack/gnocchi: Don't update resource if nothing has changed  https://review.openstack.org/20752216:29
*** deepthi has joined #openstack-ceilometer16:32
* cdent is never going to catch up with the rate of sileht and jd__ patches16:41
*** deepthi has quit IRC16:45
*** alexpilotti has joined #openstack-ceilometer16:46
*** dikonoor has quit IRC16:49
cdentsileht: is your aodh,ceilo,gnocchi,heat stuff committed anywhere?16:51
*** gordc has quit IRC16:57
haypocdent: holy crap, i think i found it16:58
haypoif (state['status_code'] / 100) not in (2, 3):16:58
*** belmoreira has quit IRC16:58
*** vishwana_ has quit IRC17:00
cdentwow17:00
cdentnice find17:00
cdenthaypo++17:01
haypocdent: it took me one hour of pdb17:01
cdentouch17:01
haypo(and print)17:01
haypoceilometer has the same bug17:01
haypowhich probably explains a lot of things17:01
*** ankita_wagh has joined #openstack-ceilometer17:02
cdentdoing math on status codes is bad mojo17:02
*** IBerezovskiy has quit IRC17:02
haypocdent: well, why not17:03
cdentbecause though they are numeric they are not numbers17:03
openstackgerritVictor Stinner proposed openstack/aodh: Port remaining tests to Python 3  https://review.openstack.org/20754717:03
* haypo invokes the daemons of Python 317:04
haypojd__: one line to port aodh to Python 3 :-D17:04
cdenthaypo I dropped a quick "delete the entire py34 env" comment17:06
cdentI can't _wait_ until everythign is py3 that will make me very happy17:07
*** _nadya_ has joined #openstack-ceilometer17:07
openstackgerritVictor Stinner proposed openstack/aodh: Port remaining tests to Python 3  https://review.openstack.org/20754717:08
*** h00327910__ has joined #openstack-ceilometer17:09
haypocdent: what do you mean by "everything?"17:09
cdentopenstack17:09
haypocdent: ok. in this case, you must vote for my talk https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/5683 and advertise it :)17:10
cdentis today the last day of voting?17:11
haypocdent: i have no idea17:13
cdentI think I saw something like that.17:13
cdentanyway, I voted, because I in face "would love to see!"17:13
cdentfact, not face17:13
openstackgerritVictor Stinner proposed openstack/ceilometer: Fix WSGI replacement_start_response() on Python 3  https://review.openstack.org/20755117:15
haypocdent: cool, thanks ;)17:15
haypoand here --^ is the fix for ceilometer17:16
haypobye17:16
* cdent waves17:16
*** [1]tshtilma has joined #openstack-ceilometer17:25
*** yprokule has quit IRC17:26
*** tshtilma has quit IRC17:27
*** [1]tshtilma is now known as tshtilma17:27
*** ddieterly has joined #openstack-ceilometer17:31
*** _nadya_ has quit IRC17:32
*** _nadya_ has joined #openstack-ceilometer17:33
*** _nadya_ has quit IRC17:33
*** _nadya_ has joined #openstack-ceilometer17:33
*** boris-42 has joined #openstack-ceilometer17:37
*** _nadya_ has quit IRC17:38
silehtcdent, infra: https://review.openstack.org/#/c/207298/ and gabbi tests: https://review.openstack.org/#/c/207137/17:40
cdentawesome, thanks17:40
*** gordc has joined #openstack-ceilometer17:41
*** tshtilma has quit IRC17:41
*** Ala has quit IRC17:42
cdentsileht: I guess you decided not use <@ for the heat template?17:44
silehtcdent, the template is finally short17:45
silehtbbl17:46
openstackgerritMehdi Abaakouk (sileht) proposed openstack/gnocchi: Don't update resource if nothing has changed  https://review.openstack.org/20752217:50
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: tests: add an integration test  https://review.openstack.org/20713717:52
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: tests: add an integration test  https://review.openstack.org/20713717:53
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Convert identity, sahara and volume to meters yaml  https://review.openstack.org/20303117:55
*** alexpilotti has quit IRC17:56
*** tshtilma has joined #openstack-ceilometer17:59
*** Ephur has joined #openstack-ceilometer17:59
*** gordc has quit IRC18:07
*** gordc has joined #openstack-ceilometer18:10
openstackgerritIlya Tyaptin proposed openstack/ceilometer: [MongoDB] Cast Int64 values to float  https://review.openstack.org/20756718:11
*** jasonamyers has quit IRC18:17
*** ig0r__ has joined #openstack-ceilometer18:25
*** eglynn has quit IRC18:26
*** ddaskal has quit IRC18:26
*** thumpba_ has joined #openstack-ceilometer18:30
*** gordc has quit IRC18:34
*** kbyrne has quit IRC18:34
*** ildikov has quit IRC18:34
*** thumpba has quit IRC18:34
*** ildikov has joined #openstack-ceilometer18:34
*** kbyrne has joined #openstack-ceilometer18:35
openstackgerritIlya Tyaptin proposed openstack/ceilometer: [MongoDB] Refactor indexes for meter and resources  https://review.openstack.org/20740918:37
*** ankita_wagh has quit IRC18:43
*** changbl has quit IRC18:43
*** exploreshaifali has quit IRC18:46
*** ig0r_ has quit IRC18:50
*** eglynn has joined #openstack-ceilometer18:53
*** changbl has joined #openstack-ceilometer18:56
*** tshtilma has quit IRC18:57
*** tshtilma has joined #openstack-ceilometer19:01
*** ankita_wagh has joined #openstack-ceilometer19:02
*** ankita_wagh has quit IRC19:03
*** ig0r_ has joined #openstack-ceilometer19:04
*** ddieterly has quit IRC19:06
*** tshtilma has quit IRC19:07
*** eglynn has quit IRC19:09
*** ityaptin has quit IRC19:09
*** ankita_wagh has joined #openstack-ceilometer19:10
*** gordc has joined #openstack-ceilometer19:25
*** julim has quit IRC19:27
*** fawadkhaliq has quit IRC19:42
*** fawadkhaliq has joined #openstack-ceilometer19:43
*** openstackgerrit has quit IRC19:46
*** openstackgerrit has joined #openstack-ceilometer19:46
openstackgerritDrew Thorstensen proposed openstack/ceilometer: Remove the compute inspector choice restriction  https://review.openstack.org/20759219:50
*** cdent has quit IRC19:50
*** ig0r_ has quit IRC19:56
*** julim has joined #openstack-ceilometer19:57
*** belmoreira has joined #openstack-ceilometer19:57
thorst1I've noticed in the PowerVM compute inspector...whenever a sample is collected I end up with multiple cpu_util entries.  I think the behavior is occurring because I've implemented both the CPUStats and CPUUtilStats (looks like all other inspectors only do one).  It looks like something is taking the CPUStats and translating that to CPUUtilStats...but I can't find that logic.  Does someone know where that is offhand?19:57
gordcthorst1: in the pipeline we compute a cpu_util value the cpu time.20:00
gordchttps://github.com/openstack/ceilometer/blob/master/etc/ceilometer/pipeline.yaml#L42-L5120:01
thorst1awesome20:01
gordcyou'd probably want to disble that in your setup20:01
thorst1yeah.  PowerVM does 'partial' processors.  So you could get like .35 processors.  It can mess with that calculation I think20:01
thorst1or at least, that's where I'm seeing the issues.  :-)20:02
gordci'd imagine. the rate of change transformer makes some assumption and is fairly basic20:02
thorst1OK - I'll take a peak around and figure out how I could turn this off.  Thank you for the quick reference!20:05
gordcthorst1: you can delete this section: https://github.com/openstack/ceilometer/blob/master/etc/ceilometer/pipeline.yaml#L9-L1420:06
gordcthat should be good enough.20:06
thorst1so its a yaml deletion only?  I'm thinking from say that deployment side...how to automate the deployment for it.  Conf file changes are easy.20:08
gordcthe pipeline.yaml file is meant to be edited, just that most people dont. by default, it gives you everything and according to a typical libvirt setup.20:09
gordcif your driver already calculates cpu_util, that pipeline (the part i delete) should not be in your file.20:10
thorst1alright.  So this can definitely be something we document.  We're going to get started on that stuff.20:10
thorst1maybe I can include a 'default for powervm' yaml file in my stackforge project for now20:11
gordcyeah. that can work. i don't see us changing pipeline file in this cycle... maybe next... probably not.20:12
thorst1yeah, totally understand20:12
thorst1thanks so much for the info.  Saved me a lot of time  :-D20:12
gordci think from deployement pov it could probably be done using puppet or something.20:12
gordcnp20:12
*** openstackgerrit has quit IRC20:16
*** openstackgerrit has joined #openstack-ceilometer20:16
* gordc gives up trying to debug notification agent20:41
*** mgagne has joined #openstack-ceilometer20:44
*** openstackgerrit has quit IRC20:46
*** openstackgerrit has joined #openstack-ceilometer20:46
*** jruano has quit IRC20:49
*** ddaskal has joined #openstack-ceilometer20:52
*** rbak has joined #openstack-ceilometer20:56
*** thumpba_ has quit IRC21:00
*** openstack has joined #openstack-ceilometer21:08
*** prad has quit IRC21:12
openstackgerritMike Dorman proposed openstack/ceilometer: Fix kafka broker 'pop from empty list' error  https://review.openstack.org/20763821:38
*** gordc_ is now known as gordc21:39
*** yassine has joined #openstack-ceilometer21:44
*** ig0r__ has quit IRC21:51
*** thorst1 has joined #openstack-ceilometer21:51
*** rbak has quit IRC21:57
*** gordc has quit IRC21:59
*** jwcroppe has joined #openstack-ceilometer22:02
*** thorst1 has quit IRC22:22
*** pauldix has quit IRC22:37
*** belmoreira has quit IRC22:37
*** ddaskal has quit IRC22:38
*** primechuck has quit IRC22:39
*** harlowja has quit IRC22:44
*** jaosorior has quit IRC22:44
*** harlowja has joined #openstack-ceilometer22:44
*** eglynn has joined #openstack-ceilometer22:58
*** boris-42 has quit IRC23:00
*** eglynn has quit IRC23:06
openstackgerritVictor Stinner proposed openstack/ceilometer: Fix more tests on Python 3  https://review.openstack.org/20766123:23
*** _cjones_ has quit IRC23:24
*** htruta has quit IRC23:26
*** htruta has joined #openstack-ceilometer23:28
*** yassine has quit IRC23:31
*** htruta is now known as htruta_23:31
*** htruta_ has quit IRC23:32
*** htruta has joined #openstack-ceilometer23:32
*** jaypipes has quit IRC23:39
« Wednesday, 2015-07-29 Index Friday, 2015-07-31 »

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