Thursday, 2015-06-25

*** david-ly_ has joined #openstack-ceilometer00:06
*** jwcroppe has quit IRC00:07
*** jwcroppe has joined #openstack-ceilometer00:07
*** david-lyle has quit IRC00:08
*** ankita_w_ has quit IRC00:09
*** jwcroppe has quit IRC00:10
*** Longgeek has joined #openstack-ceilometer00:12
*** jwcroppe has joined #openstack-ceilometer00:13
*** edmondsw has joined #openstack-ceilometer00:16
*** edmondsw_ has joined #openstack-ceilometer00:16
*** edmondsw_ has quit IRC00:17
*** pauldix has joined #openstack-ceilometer00:20
*** Longgeek has quit IRC00:23
*** pauldix has quit IRC00:29
*** rwsu has quit IRC00:38
*** thorst has joined #openstack-ceilometer00:39
*** boris-42 has quit IRC00:42
*** ddieterly has joined #openstack-ceilometer00:43
*** jfluhmann has quit IRC00:47
*** jwcroppe_ has joined #openstack-ceilometer00:47
openstackgerritliusheng proposed openstack/ceilometer: Add support for posting samples to notification-agent via API
*** edmondsw has quit IRC00:47
*** r-daneel has quit IRC00:48
*** jwcroppe has quit IRC00:49
*** dscottsf has joined #openstack-ceilometer00:50
*** dscottsf has quit IRC00:53
*** thorst has quit IRC01:00
*** ankita_wagh has joined #openstack-ceilometer01:04
*** jfluhmann has joined #openstack-ceilometer01:12
openstackgerritliusheng proposed openstack/ceilometer: Add support for posting samples to notification-agent via API
*** david-ly_ is now known as david-lyle01:36
*** _cjones_ has quit IRC01:41
*** dontalton has quit IRC01:56
openstackgerritRyota MIBU proposed openstack/ceilometer: Add 'event' type and 'event_rule' to alarm API
*** ddieterly has quit IRC02:07
*** ddieterly has joined #openstack-ceilometer02:09
*** jruano has joined #openstack-ceilometer02:11
*** ddieterly has quit IRC02:14
*** ankita_wagh has quit IRC02:35
openstackgerritgordon chung proposed openstack/ceilometer: remove unused notifier
*** ddieterly has joined #openstack-ceilometer02:52
*** jruano has quit IRC03:15
*** Longgeek has joined #openstack-ceilometer03:20
*** Longgeek has quit IRC03:21
*** Longgeek has joined #openstack-ceilometer03:21
*** jruano has joined #openstack-ceilometer03:34
openstackgerritMerged openstack/ceilometer: Add support for posting samples to notification-agent via API
*** rjaiswal has quit IRC03:44
*** ankita_wagh has joined #openstack-ceilometer03:50
*** deepthi has joined #openstack-ceilometer03:55
*** ddieterly has quit IRC04:10
*** jruano has quit IRC04:13
*** jruano has joined #openstack-ceilometer04:34
openstackgerritliusheng proposed openstack/ceilometer: Add documentation about the usage of api-no-pipline
*** Longgeek has quit IRC04:37
*** Longgeek has joined #openstack-ceilometer04:37
openstackgerritliusheng proposed openstack/ceilometer: Add documentation about the usage of api-no-pipeline
*** Longgeek has quit IRC04:44
*** jfluhmann has quit IRC04:51
*** yatin has joined #openstack-ceilometer05:04
*** jruano has quit IRC05:05
*** ddieterly has joined #openstack-ceilometer05:10
*** ddieterly has quit IRC05:15
*** yprokule has joined #openstack-ceilometer05:17
*** Longgeek has joined #openstack-ceilometer05:17
*** ildikov has quit IRC05:26
*** ddieterly has joined #openstack-ceilometer06:11
*** ildikov has joined #openstack-ceilometer06:13
*** changbl_ has quit IRC06:15
*** ddieterly has quit IRC06:16
*** jaosorior has joined #openstack-ceilometer06:41
*** ankita_wagh has quit IRC06:42
*** liusheng_phone has joined #openstack-ceilometer06:46
*** belmoreira has joined #openstack-ceilometer06:53
*** exploreshaifali has joined #openstack-ceilometer06:53
*** yprokule has quit IRC06:57
*** _nadya_ has joined #openstack-ceilometer06:58
*** yprokule has joined #openstack-ceilometer06:59
*** _nadya_ has quit IRC07:00
*** _nadya_ has joined #openstack-ceilometer07:00
*** _nadya_ has quit IRC07:05
*** safchain has joined #openstack-ceilometer07:09
*** fabian4 has quit IRC07:11
*** ddieterly has joined #openstack-ceilometer07:12
*** gzhai1 has joined #openstack-ceilometer07:14
*** ddieterly has quit IRC07:17
*** lan has quit IRC07:20
*** shardy has joined #openstack-ceilometer07:20
*** lan has joined #openstack-ceilometer07:21
*** fabian4 has joined #openstack-ceilometer07:24
*** gzhai1 has quit IRC07:25
*** lsmola has joined #openstack-ceilometer07:28
*** cdent has joined #openstack-ceilometer07:29
*** ankita_wagh has joined #openstack-ceilometer07:36
*** fabian4 has quit IRC07:36
*** jfluhmann has joined #openstack-ceilometer07:37
*** j_t has joined #openstack-ceilometer07:44
j_tOnce I'm monitoring a VMs cpu, memory, etc etc. If I find VMs that aren't "doing anything", for awhile.. is there some other openstack system I can use to delete them? or at least say send an email... and it gets deleted in a month. give the user a link to click to cancel the process. in a week it gets shut down, in 3 more weeks it gets deleted07:45
j_tWe run a private cloud, and people spin up Vms and they forget about them07:45
*** Ala has joined #openstack-ceilometer07:45
j_tso it would be nice to clean up unused crap in an automated way07:45
j_tdoes anything like this exist?07:45
*** fabian4 has joined #openstack-ceilometer07:48
cdentj_t: sounds like maybe ceilometer's alarming functionality or the heat project (which uses ceilometer's data) might be able to help you07:50
j_tcdent: hmm, so I'm guessing this is something that has to be built from scratch?07:59
*** gzhai1 has joined #openstack-ceilometer08:00
j_tleveraging the ceilometer/heat stuff08:00
cdentthat's a good question. I bet there are pre-assembled solutions out there, but I don't know about them.08:00
cdentmuch of the openstack stuff is bits of lego to put together in various ways08:00
cdentIf you're on the operators mailing list you might ask there: there are often people who have shared the tools they've assembled08:01
j_tyep, but it's mixed south american legos, with the chinese knock offs too, and the expensive american legit ones08:01
j_tso the pieces don't always fit08:01
j_tso you have to use a hammer, with some glue and duct tape too08:01
j_tI did find this old krufty perl project: from paypal08:02
j_tactually the last time I checked that, there were no recent commits08:03
j_tIt didn't look like it worked08:03
j_tOne of those "let's be opensourcey and throw our internal thing that is specific to our environment over the wall" situations08:04
j_tand no updates for years08:04
j_tbut now when I look at it, there was a recent update a month ago08:04
j_tI wonder if that works.08:04
cdentgood luck, if you need to use the hammer, start with light taps before swinging wildly08:04
*** deepthi has quit IRC08:05
j_tyep, especially when it comes to deleting08:05
j_tI'll delete devtest VMs first before running this in production.08:05
*** ankita_wagh has quit IRC08:09
*** ddieterly has joined #openstack-ceilometer08:13
*** exploreshaifali has quit IRC08:16
openstackgerritChris Dent proposed openstack/gnocchi: Ensure location header urls account for script_name
*** ddieterly has quit IRC08:18
*** _nadya_ has joined #openstack-ceilometer08:33
*** yassine_ has joined #openstack-ceilometer08:36
*** _nadya_ has quit IRC08:38
*** _nadya_ has joined #openstack-ceilometer08:38
*** _nadya_ has quit IRC08:43
openstackgerritLianhao Lu proposed openstack/ceilometer: (DO NOT MERGE)
*** deepthi has joined #openstack-ceilometer08:52
openstackgerritMerged openstack/gnocchi: Ensure live gabbi tests run in gate
openstackgerritMerged openstack/gnocchi: Don't reclone the repo we already did
openstackgerritChris Dent proposed openstack/gnocchi: Ensure location header urls account for script_name
*** deepthi has quit IRC09:02
openstackgerritChris Dent proposed openstack/gnocchi: Allow gnocchi API at /gnocchi
openstackgerritChris Dent proposed openstack/gnocchi: Allow gnocchi API at /metric
*** liusheng_phone has quit IRC09:09
*** shardy has quit IRC09:14
*** ddieterly has joined #openstack-ceilometer09:14
*** yprokule has quit IRC09:14
therveHey there09:15
*** shardy has joined #openstack-ceilometer09:15
therveWe're seeing some test failures in integration tests with ceilometer09:15
therveThere is a suspicious commit yesterday, anyone has an idea what it does?09:15
*** ddieterly has quit IRC09:18
*** deepthi has joined #openstack-ceilometer09:18
therveEspecially that new direct flag, not sure what it does09:20
*** yprokule has joined #openstack-ceilometer09:26
llusileht: I dig more into the oslo.messaging pool issue09:29
llusileht: it seems that the pool works fine09:30
llusileht: it's other issues on my local environment and my patches, as well as bug#1467776, combined together caused the jenkins failure09:31
llutherve: what's the failure?09:31
thervellu, bug #146869709:32
openstackbug 1468697 in heat "Test ceilometer alarm scenario is failing on gates" [Medium,Confirmed] - Assigned to Kairat Kushaev (kkushaev)09:32
thervellu, We're posting a sample, but it doesn't trigger an alarm notification as we expect09:32
therveAnd as it used to do I may say09:32
llutherve: ah09:32
lluyou also need to have notification-agent running in order to have the POST sample working09:33
therveIt seems to be running as expected09:34
lluhmm, then this would indicate some issues in the patch series,n,z09:37
lluliusheng: are you there?09:38
lluliusheng: could you look at bug #146869709:38
openstackbug 1468697 in heat "Test ceilometer alarm scenario is failing on gates" [Medium,Confirmed] - Assigned to Kairat Kushaev (kkushaev)09:38
liushengllu: yes09:38
liushengllu: ok, looking09:38
openstackgerritThomas Herve proposed openstack/ceilometer: WIP: test revert
*** alexpilotti has joined #openstack-ceilometer09:45
liushengllu: I guess the reason is in devstack senario, the configure option "notification_driver = messagingv2" is not configured which is needed for api-no-pipeline09:54
liushengllu: I have upload documentation about the usage
jd__llu: sileht is in vacation09:57
jd__and I need him for the Ceph driver :(09:57
liushengllu: I am working on ceilometerclient to add flag 'direct', the failed tests can work with 'drect' flag09:57
liushengllu: or fix it in devstack script09:58
*** edmondsw has joined #openstack-ceilometer10:06
*** edmondsw has quit IRC10:06
*** edmondsw has joined #openstack-ceilometer10:07
openstackgerritMerged openstack/ceilometer: remove unused notifier
*** Kennan2 has joined #openstack-ceilometer10:14
*** Kennan has quit IRC10:14
*** Kennan2 is now known as Kennan10:17
*** exploreshaifali has joined #openstack-ceilometer10:20
*** gautamd has joined #openstack-ceilometer10:25
*** jfluhmann has quit IRC10:34
*** arun_ has joined #openstack-ceilometer10:40
*** arun_ has left #openstack-ceilometer10:40
*** deepthi has quit IRC10:42
openstackgerritAleksey Ripinen proposed openstack/ceilometer: Samples with last timestamp in sqlalchemy backend
openstackgerritAleksey Ripinen proposed openstack/ceilometer: Cover impl_db2 by tests
openstackgerritAleksey Ripinen proposed openstack/ceilometer: Add complex query for statistic requests
openstackgerritAleksey Ripinen proposed openstack/ceilometer: Query statistics function for sql
openstackgerritliusheng proposed openstack/python-ceilometerclient: Add ceilometerclient support for api-no-pipeline
therveliusheng, llu: So I confirmed that this commit is the culprit. What should we do? Revert?11:08
liushengtherve: I'm trying to fix it in devstack script, patch will be uploaded soon11:10
therveliusheng, Do you think that's enough?11:10
therveIt sounds like a backward incompatible change11:10
liushengtherve: no, I think it should be add a configuration in ceilometer conf as default for api-no-pipeline feature11:11
therveWell it's breaking heat gate, so we'd like to have a quick fix11:13
*** ddieterly has joined #openstack-ceilometer11:15
*** david-lyle has quit IRC11:17
therveIf only one value for notification_driver is supported, maybe there shouldn't be an option at all11:19
*** gautamd has quit IRC11:20
*** ddieterly has quit IRC11:20
*** david-lyle has joined #openstack-ceilometer11:21
liushengtherve: :(, actually, it is ok without the configuration but that need heat test add the 'direct' flag and need waiting ceilometerclient's patch merged11:25
therveliusheng, It should be the other way around to keep compatibility11:25
liushengtherve: I have uploaded a patch in devstack for this11:26
therveliusheng, Why is the patch in devstack? Should you take care of this in ceilometer code base itself?11:27
thervePeople not using devstack will have the same issue as heat11:27
*** kairat_kushaev has joined #openstack-ceilometer11:27
liushengtherve: the notification_driver is also add in devstack for other projects (nova, neutron, cinder...) for ceilometer itself, it only can provide documentation to remind deployers to config this option11:30
therveSorry it doesn't make sense to me11:31
*** prad has joined #openstack-ceilometer11:32
therveliusheng, If I'm running the kilo version of ceilometer, and then upgrade to liberty, I expect the behavior to stay the same without having to change a configuration option11:32
liushengtherve: for testing scenario, the feature suggest posting sample with 'direct=True' flag, if so, the configure option isn't needed. very sorry for this, if we cannot wait for others merged, please revert it11:37
*** EmilienM|off is now known as EmilienM11:38
openstackgerritliusheng proposed openstack/python-ceilometerclient: Add ceilometerclient support for api-no-pipeline
*** AJaeger has joined #openstack-ceilometer12:01
AJaegerHi ceilometer cores, it's time for another translation import IMHO - could somebody +A, please?12:01
cdentAJaeger: done12:02
AJaegerthanks, cdent !12:02
cdentThat's the third time I've +2 that review :)12:02
AJaegercdent: Since it's a automatic import, other projects don't use a second core. A single "+2A" is enough for those. Might be something you want to discuss...12:03
cdentthat's a good idea12:04
cdentI'm a relatively new core and gordc gave me strict instructions to not +A anything that didn't already have a +212:04
cdentbut that gordc, he's just a tyrant, I'm going to ignore him henceforth12:04
AJaegerDON'T! ;)12:04
AJaegerThe translation import is an exception - and the only one I'm aware off - for many projects12:05
cdentI was kidding!12:05
AJaegercdent: I know12:05
AJaegerso, figure it out with your wise leader ;)12:06
cdentI gordc were actively listening I'd feel far better about all the teasing of him I want to do12:07
AJaegercdent: I'll wave good-bye again. Have a great day!12:08
* cdent waves12:08
*** AJaeger has left #openstack-ceilometer12:08
openstackgerritChris Dent proposed openstack/gnocchi: Allow gnocchi API at /metric
*** htruta has joined #openstack-ceilometer12:12
gordci have no control12:15
gordci grant thee permission to +2A translations12:16
*** ddieterly has joined #openstack-ceilometer12:16
gordcand also reqiurements12:16
gordcand also anything i post.12:16
*** ddieterly has quit IRC12:20
*** exploreshaifali has quit IRC12:32
*** david-ly_ has joined #openstack-ceilometer12:35
*** david-lyle has quit IRC12:39
*** _nadya_ has joined #openstack-ceilometer12:41
*** yatin has quit IRC12:48
*** liusheng_phone has joined #openstack-ceilometer12:50
*** _nadya_ has quit IRC12:51
*** _nadya_ has joined #openstack-ceilometer12:52
*** Ctina has joined #openstack-ceilometer12:56
openstackgerritgordon chung proposed openstack/ceilometer: catch warnings from error tests
*** julim has joined #openstack-ceilometer13:01
openstackgerritgordon chung proposed openstack/ceilometer: fix kafka tests from flooding logs
openstackgerritgordon chung proposed openstack/ceilometer: test error log - catch dummy error
thervegordc, Hey, if we can get your feedback on that alarm issue that'd be cool13:04
openstackgerritChris Dent proposed openstack/gnocchi: Allow gnocchi API at /metric
gordctherve: sure. you have a link? i might have missed it13:04
thervegordc, Links you mean :)13:05
thervebug #1468697 is the test issue we're having in Heat13:05
openstackbug 1468697 in heat "Test ceilometer alarm scenario is failing on gates" [Medium,Confirmed] - Assigned to Kairat Kushaev (kkushaev)13:05
*** belmoreira has quit IRC13:05
therveIt seems I6a1587e4aae055cc559286367632249865008654 would be the culprit13:05
therveSo I think it's introducing a regression, and apparently changing some options would fix it13:06
thervegordc, for the discussion13:07
gordckk, reading13:07
*** afazekas has joined #openstack-ceilometer13:08
gordctherve: at quick glance, i'd be surprised if that patch is affecting your alarms.13:08
thervegordc, I checked with and it seems to do the trick13:10
therveYou think that option is set properly?13:10
gordcoh. you're posting samples, i just read up13:11
*** jruano has joined #openstack-ceilometer13:11
*** husanux1 has joined #openstack-ceilometer13:13
*** husanux1 has quit IRC13:13
gordctherve: we can revert it for now.13:14
gordcso it unblocks you... i think the devstack change might be correct though.13:14
thervegordc, You don't think it's a backward incompatible change though?13:15
therveI'm okay to wait a bit, if the devstack change can land today.13:16
therveI'm just not convinced it's the right solution13:16
gordctherve: it'll probably take longer to land in devstack13:16
gordcwant to post a proper revert? or i can13:17
*** shardy has quit IRC13:17
therveI can, let me do this right now13:17
gordckk i'll  +2A13:17
gordctherve: i should probably mention there's a button in gerrit to revert it (assuming it works).13:20
openstackgerritThomas Herve proposed openstack/ceilometer: Revert "Add support for posting samples.."
thervegordc, ^^ :)13:21
*** exploreshaifali has joined #openstack-ceilometer13:21
gordcdid you hit revert button actually?  just so it's easier to track?13:22
*** shardy has joined #openstack-ceilometer13:22
thervegordc, Ah no I didn't13:24
therveLet's try that13:24
gordctherve: yeah, that'd be better. thanks!13:24
openstackgerritThomas Herve proposed openstack/ceilometer: Revert "Add support for posting samples to notification-agent via API"
*** husanux9 has joined #openstack-ceilometer13:25
thervegordc, There you go13:25
gordctherve: thanks! +A. we'll work on fixing that in the meantime.13:28
therveThank you!13:28
*** husanux9 has quit IRC13:29
*** liusheng has quit IRC13:35
*** liusheng has joined #openstack-ceilometer13:35
*** kun_huang has joined #openstack-ceilometer13:38
gordcliusheng: added a comment to your devsatck patch.13:40
*** changbl_ has joined #openstack-ceilometer13:40
gordcliusheng: we probably don't need it. i think we just configured notifier wrong. (maybe we need to add a test case in test_notification for api?)13:40
liushenggordc, thanks, but the notification_driver option is needed to get notifier13:42
liushengthe option has a [] as default13:42
*** shardy has quit IRC13:42
gordcliusheng: we already have our own value:
*** shardy has joined #openstack-ceilometer13:44
*** julim has quit IRC13:44
liushenggordc, I got it, yes, I didn't notice this option, I will test it, thanks13:45
*** _nadya_ has quit IRC13:46
*** _nadya_ has joined #openstack-ceilometer13:47
gordcliusheng: cool cool13:48
* gordc makes a mental note on how to create notifier.13:48
cdentgordc you should share all mental notes of that ilk in some non-mental way13:51
*** _nadya_ has quit IRC13:51
*** htruta has quit IRC13:51
gordccdent: 75% of it is not safe for work... but i guess safe for internet.13:54
liushenggordc, I have noticed nova, cinder, heat... need this option manully to send notifications( to notification-agent, so I assume ceilometer also need this, my mistake13:54
cdentall the better gordc13:54
gordcliusheng: i feel like our option is some side effect to pre-oslo.messaging days. we might want to use notificaiton_driver eventually.13:55
gordcthat said, i don't know what happens when you have both set.13:56
gordccdent: don't encourage me. a lot of effort goes into containing the beast.13:56
cdentFREE THE BEAST13:56
* cdent aws13:57
liushenggordc, yeah, I wonder why notification_driver has a default value (messaging) than a empty list13:57
liushengs /has/hasn't/13:57
gordcliusheng: you'll need to ask sileht :)13:57
liushenggordc, :)13:58
gordccdent: i haven't saved enough yet to be able to live unemployed.13:58
*** exploreshaifali has quit IRC13:58
cdentokay, I'll let you off this time13:58
gordc3 years? i'll open a bank account -- shoe box is getting full anyways13:59
*** julim has joined #openstack-ceilometer14:04
*** llu has quit IRC14:07
*** llu has joined #openstack-ceilometer14:08
*** r-daneel has joined #openstack-ceilometer14:09
*** rbak has joined #openstack-ceilometer14:16
*** liusheng has quit IRC14:20
*** _nadya_ has joined #openstack-ceilometer14:29
*** shardy has quit IRC14:29
*** shardy has joined #openstack-ceilometer14:30
*** htruta has joined #openstack-ceilometer14:33
*** eglynn has joined #openstack-ceilometer14:33
*** pauldix has joined #openstack-ceilometer14:43
openstackgerritAlexandre Viau proposed openstack/python-ceilometerclient: Typo: Show an sample => Show a sample
*** alextricity has joined #openstack-ceilometer14:47
*** pauldix has quit IRC14:48
alextricityHey guys. I'm working on getting ceilometer up on Kilo, but am having trouble with adding the following configs for configuring cinder to send notifications14:48
alextricitycontrol_exchange = cinder14:48
alextricitynotification_driver = messagingv214:48
alextricityMy cinder volume creation hangs if I add those directives.14:49
alextricityHowever, if I remove them, cinder starts working again :/14:49
alextricityI've checked the logs..and unfortunately I wasn't able to find anything significant. Can anybody help me out?14:49
openstackgerritMerged openstack/ceilometermiddleware: Add test which asserts non-empty target_id
*** david-ly_ is now known as david-lyle15:01
_nadya_alextricity: hi! Maybe anything interesting in the logs when Cinder doesn't work?15:02
gordcheads up. we have a meeting over at openstack-meeting15:02
*** shardy has quit IRC15:08
*** shardy has joined #openstack-ceilometer15:08
*** rbak has quit IRC15:09
*** idegtiarov_ has joined #openstack-ceilometer15:09
*** rbak has joined #openstack-ceilometer15:13
*** shardy has quit IRC15:13
*** shardy has joined #openstack-ceilometer15:13
*** pauldix has joined #openstack-ceilometer15:15
*** edmondsw has quit IRC15:16
*** edmondsw has joined #openstack-ceilometer15:25
openstackgerritRoman Vasilets proposed openstack/python-ceilometerclient: Change default values from [] to None
openstackgerritRoman Vasilets proposed openstack/python-ceilometerclient: Change default values from [] to None
*** Ala has quit IRC15:58
*** idegtiarov_ has quit IRC15:59
*** Longgeek has quit IRC15:59
*** Longgeek has joined #openstack-ceilometer16:00
gordc_nadya_: why were you using notifier? we can add it back... it just didn't seem to be connected to anything when i was looking at it16:00
gordcoh, you just used it as a dump for tests... leaving it off now :)16:02
_nadya_gordc: no, I don't think it's needed to get it back. It was the easiest way to reach plugin_base's  to_samples_and_publish16:03
_nadya_gordc: I think I will manage using several mocks16:03
ildikovgordc: _nadya_: sorry, on another meeting and I still cannot talk and write at the same time... :(16:04
ildikovso as I see the issue is that as we don't generate data, but receive it and because of the OpenStack madness we are periodically in trouble to properly consume notifications16:05
ildikovbut as we are on the consumer side, we are not able to handle the root cause16:06
_nadya_gordc: it's just funny that it was in the project for years and removed at the moment I started to use it :)16:06
gordcildikov: yeah. but then you need to propose a schema. and ppl for some reason have allegiance to own schema16:07
ildikovgordc: Nova is about to move to versioned obejcts for notifications, I think that could be a starting point, at least they have now an intention to change16:08
ildikovgordc: it can be raised on the cross-project level/meeting that it is still an issue let's solve it anyhow16:09
cdentI've similarly heard that versioned objects could help the schematization process16:11
* cdent waits to be convinced :16:11
gordcbut version object only guarantees that the notifications work amongst themselves no? if the content was never there, versionobjects doesn't solve that.16:11
ildikovI just wanted to point out that at least they are aiming for a schema and versioning16:12
ildikovbut they don't have the spec and roadmap yet16:12
gordcildikov: is it a schema or just versioning?16:12
*** htruta has quit IRC16:12
ildikovso this could be the time to collaborate a bit as if they change it anyway, then it could be changed in a way that's somehow good for us too16:12
ildikovgordc: as much as I undrestood they would like to have schema too, but they don't really have any material that would describe the plans yet16:13
_nadya_I think that is_valid() addition doesn't block an action 'raise the question'16:13
ildikovI'm just saying that if we start to drop things now kind of silently that will be confusing from user perspective16:15
gordcildikov: i see. yeah, i think our issue is schema right now... or in the case _nadya_ is addresssing, the middleware was giving unusable data for certain points16:15
ildikovbut maybe I'm caring too much about users :)16:15
gordc_nadya_: did we ever figure out why it was missing a resource_id in payload16:15
_nadya_the initial bug is here . I can add handling to HBase of course but Mongo and HBase will act differently :(16:16
openstackLaunchpad bug 1464265 in Ceilometer "hbase backend not collecting all Swift metric" [Undecided,In progress] - Assigned to Nadya Privalova (nprivalova)16:16
gordcsql should fail. i'm pretty sure i made resource_id mandatory field16:16
_nadya_gordc: I tried to break it but no success. Maybe the issue happened on old version16:17
gordc_nadya_: yeah, i was looking at code, i couldn't figure out a scenario where it'd be empty16:17
_nadya_gordc: so sql fails, HBase fails but Mongo lives16:18
*** liusheng_phone has quit IRC16:19
*** afazekas has quit IRC16:23
gordc_nadya_: can we not just create a unique index in mongo16:24
gordc_nadya_: i guess main concern i have is we're adding a check on every single datapoint now... this check is not needed most of the time... and is only useful in the case where we get unusable data16:25
_nadya_so in fact move validation on storage lvl?16:25
gordc_nadya_: that's my thought. i don't know if anyone else has a differing opinion16:25
gordciiuc, we'll never be able to recover these datapoints.16:26
_nadya_gordc: could you please add it in review :)? I will ping someone else, just gather the thoughts16:27
_nadya_gordc: and will see16:27
openstackgerritChris Dent proposed openstack/ceilometer: Make ceilometer work correctly when hosted with a SCRIPT_NAME
cdentIsn't the only way to get people to send good notifications to not accept them?16:30
gordc_nadya_: cool cool16:31
ildikovcdent: the only with this question is that those who send will not be the ones who face with the issue of missing samples16:31
gordccdent: yeah, you'd think so. but then they say it's your fault and walk away16:31
ildikovonly *issue16:31
cdentI agree with both of you :)16:31
* cdent locates dinner16:32
* cdent waves16:32
ildikovcdent: gordc: but I think we would need to be more pushy on this, then just let them walk away16:32
*** cdent has quit IRC16:32
*** jaosorior has quit IRC16:35
*** yatin has joined #openstack-ceilometer16:36
gordcildikov: if we present an idea at cross project, what is the proposal?16:36
ildikovgordc: as for the unusable data from the middleware, can we address that?16:36
gordcyou must include a resource_id (at a specific location) in your payload?16:36
gordcildikov: we can... i actually don't know why it's missing resource_id16:37
ildikovif we can define what we consider as required field for "sane" samples, then it's a starting point16:37
gordcildikov: i made this
ildikovgordc: so I think we should validate those things that we actually can16:37
gordccadf actually validates fields. i think we just let it slide with empty string16:37
ildikovcadf is good, but we have it only fo raudit data and not for everything16:39
ildikovwhy I'mpushing this is that we are trying to encourage every project to send notifications so we don't need to poll that much16:39
*** htruta has joined #openstack-ceilometer16:43
*** shardy has quit IRC16:45
gordcoh. i wasn't trying to sell cadf. i was just mentioning it is because the ceilometermiddleware uses it.16:45
*** shardy has joined #openstack-ceilometer16:45
ildikov... but we don't give a guideline nor enforce to have required fields and schema etc to be able to validate16:45
*** eglynn has quit IRC16:45
gordcand that id that's missing in middleware shouldn't be missing which is why i'm confused16:46
_nadya_sometimes resource_id is not needed in notifications. we construct it ourselves16:46
gordcildikov: so i think every cycle for past 2 years, someone has tried to suggest a schema16:46
gordcildikov: the general response is... 'maybe there's another alternative... i don't have a real example of an alternative, but maybe there is'16:47
ildikovgordc: I would buy it, but I'm already on your side :)16:47
_nadya_and in the bug it's not missed actually. It's empty String16:47
gordcwhat notification would we consume that doesn't have a resource_id?16:48
gordc_nadya_: yeah. i don't know how... that mean that account doesn't have a name in swift?16:48
*** r-mibu has quit IRC16:48
*** r-mibu has joined #openstack-ceilometer16:48
gordcildikov: meters/metrics definitely have a resource id... if it's not there, the datapoint makes no sense.16:49
_nadya_gordc: yep. that's why cadf generates it16:49
gordcildikov: events might not have a resource_id, but then you wonder what that event is actually about and why do i care about it.16:49
gordc_nadya_: you can have accounts without names in swift?16:50
gordcmaybe we're using the wrong value as our resource_id in our middleware.16:50
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Update alarm history only if change in alarm property
*** jwcroppe_ has quit IRC16:51
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Implementation of dynamically reloadable pipeline
openstackgerritRohit Jaiswal proposed openstack/ceilometer-specs: Spec for DBaaS(Trove) notification consumption
*** jwcroppe has joined #openstack-ceilometer16:51
_nadya_gordc: don't think so. but if your reseller_prefix is wrong you may fetch empty line16:52
*** rwsu has joined #openstack-ceilometer16:52
_nadya_gordc: in ipmi we do resource_id = '%(nodeid)s-%(sensorid)s'16:54
ildikovgordc: yeap, data without resource_id is mostly not useful16:55
_nadya_in swift resource_id=message['payload']['target']['id'] . So the message may not contain 'resource_id' field16:55
*** jwcroppe has quit IRC16:56
*** _cjones_ has joined #openstack-ceilometer16:58
gordc_nadya_: well in cadf, (since that msg is in cadf) it's a mandatory field17:01
gordc_nadya_: if we can't get the value, it's either a bug in swift, or a issue with our middleware i think17:01
_nadya_# Resource.id17:03
_nadya_        setattr(self, RESOURCE_KEYNAME_ID, id or identifier.generate_uuid())17:03
_nadya_gordc: it's cadf code17:04
*** htruta has quit IRC17:04
*** dscottsf has joined #openstack-ceilometer17:05
_nadya_gordc: maybe just not send notification from our middleware if id is empty?17:06
_nadya_gordc: solve problem with Swift only17:06
*** dscottsf has quit IRC17:06
gordc_nadya_: so if CADF generates an id if missing, how is it empty when it reaches the storage?17:09
_nadya_gordc: I don't know :) I have only one thought: maybe the guy who filed the bug used very old version without cadf17:12
_nadya_gordc: but in description (or a letter) he daid it's Kilo17:13
gordc_nadya_: i see... yeah we still had the old middleware in kilo package.17:14
gordc_nadya_: i'll comment on your patch.17:15
*** _cjones_ has quit IRC17:17
*** _cjones_ has joined #openstack-ceilometer17:17
*** shardy has quit IRC17:25
*** jfluhmann has joined #openstack-ceilometer17:25
*** shardy has joined #openstack-ceilometer17:26
*** ankita_wagh has joined #openstack-ceilometer17:28
*** rjaiswal has joined #openstack-ceilometer17:31
*** Ctina_ has joined #openstack-ceilometer17:31
*** Ctina has quit IRC17:35
*** Ctina_ has quit IRC17:36
*** ddieterly has joined #openstack-ceilometer17:39
*** jwcroppe has joined #openstack-ceilometer17:41
*** yassine_ has quit IRC17:45
kun_huangguys, using ceilometer client, create a sample successfully and get [] from running samples.list()17:45
kun_huangwhat's the possible reasons17:45
gordckun_huang: what does your pipeline look like? it's not explicit but even when you post to api, it gets filtered through api before going to db17:48
*** yatin has quit IRC17:48
kun_huanggordc: pipeline = request_id authtoken api-server17:49
kun_huangfrom api-paste.ini17:49
*** afazekas has joined #openstack-ceilometer17:49
kun_huangwhat I print from sample.create(...) is:17:51
kun_huang[<OldSample {u'counter_name': u'cpu_util', u'user_id': u'e7c841a59f5e45a2b95e1b4a8755dfd0', u'resource_id': u'rally_resource_QtHTpY9wcb', u'timestamp': u'2015-06-23T01:38:36.116869', u'resource_metadata': {}, u'source': u'ccd9013decc9485aaf69780436a1e59a:openstack', u'counter_unit': u'instance', u'counter_volume': 1.0, u'project_id':17:51
kun_huangu'ccd9013decc9485aaf69780436a1e59a', u'message_id': u'90bae1f8-1948-11e5-9639-000c29c6eec1', u'counter_type': u'gauge'}>]17:51
gordckun_huang: oh i meant your pipeline.yaml17:51
gordckun_huang: here's the bug i'm mentioning
openstackLaunchpad bug 1432077 in Ceilometer "Ceilometer client fails to create samples with new (unsupported) meter/counter name" [Medium,Fix released] - Assigned to ZhiQiang Fan (aji-zqfan)17:51
*** yprokule has quit IRC17:51
kun_huangthis one17:52
*** boris-42 has joined #openstack-ceilometer17:54
*** afazekas has quit IRC17:55
kun_huanggordc: btw rally calls ceilometer with counter_type, not meter_type. is that serious?17:57
*** idegtiarov_ has joined #openstack-ceilometer18:00
gordckun_huang: meter_type?18:01
kun_huangthe parameter of the command line "ceilometer sample-create"18:02
*** dontalton has joined #openstack-ceilometer18:02
*** husanu4 has joined #openstack-ceilometer18:02
*** ildikov has quit IRC18:07
*** jwcroppe_ has joined #openstack-ceilometer18:08
*** jwcroppe_ has quit IRC18:09
gordckun_huang: sorry, just finishing lunch18:09
gordckun_huang: i'm not actually familiar with exact names of param.. i'd need to check18:09
*** jwcroppe_ has joined #openstack-ceilometer18:09
*** jwcroppe has quit IRC18:11
gordckun_huang: hmmm. you're right, the client expects meter-type not counter-type18:11
*** husanu4 has quit IRC18:14
*** lsmola has quit IRC18:19
*** julim has quit IRC18:28
*** julim has joined #openstack-ceilometer18:28
*** jruano has quit IRC18:33
*** _nadya_ has quit IRC18:38
*** _nadya_ has joined #openstack-ceilometer18:38
*** _nadya_ has quit IRC18:43
openstackLaunchpad bug 1468697 in heat "Test ceilometer alarm scenario is failing on gates" [Medium,Confirmed] - Assigned to Kairat Kushaev (kkushaev)18:43
kun_huangthis issue18:43
*** dontalton has quit IRC18:44
kun_huangnot only rally gate is blocked, and heat's18:44
*** j_t has left #openstack-ceilometer18:49
*** julim has quit IRC18:52
*** rbrooker has joined #openstack-ceilometer18:52
*** r-daneel has quit IRC18:54
*** r-daneel has joined #openstack-ceilometer18:55
*** jwcroppe_ has quit IRC18:55
*** jwcroppe has joined #openstack-ceilometer18:56
*** jwcroppe has quit IRC18:57
*** jwcroppe has joined #openstack-ceilometer18:57
*** julim has joined #openstack-ceilometer18:59
*** ddieterly has quit IRC19:02
*** julim has quit IRC19:02
*** julim has joined #openstack-ceilometer19:03
*** ddieterly has joined #openstack-ceilometer19:03
*** gzhai1 has quit IRC19:08
*** jruano has joined #openstack-ceilometer19:08
*** gzhai1 has joined #openstack-ceilometer19:09
*** htruta has joined #openstack-ceilometer19:13
*** ildikov has joined #openstack-ceilometer19:22
*** shardy has quit IRC19:42
*** Ephur has joined #openstack-ceilometer19:42
*** shardy has joined #openstack-ceilometer19:43
*** julim has quit IRC19:44
*** Ephur has quit IRC19:47
*** safchain has quit IRC19:48
boris-42gordc: hi19:52
boris-42kun_huang: hi19:52
boris-42gordc: kun_huang any progress regarding to finxing gates?19:52
gordcboris-42: kun_huang: we're waiting on:
gordcboris-42: there are bigger breaks in the gate that blocking the revert19:59
openstackgerritgordon chung proposed openstack/ceilometer: Revert "Add support for posting samples to notification-agent via API"
*** nsaje_ has joined #openstack-ceilometer20:03
*** nsaje__ has joined #openstack-ceilometer20:04
boris-42gordc: oh20:05
*** nsaje has quit IRC20:06
*** nsaje_ has quit IRC20:07
ankita_waghgordc : So if I choose Kafka, will the meters published in the form of flattened json20:08
ankita_waghkafka as a publisher *20:08
gordcankita_wagh: so yeah, when the publishers publish, the data will be serialised20:08
gordcankita_wagh: let me check on that. i don't actually use the kafka publish20:08
gordci'd assume it's in some format that is accept to kafka broker20:08
ankita_waghI also need to add some fields before pushing into kafka20:09
ankita_waghHow can I do that20:09
gordcmeter_message_from_counter basically serialised a sample20:10
gordcto add fields you'll probably need to build your own transformer20:11
gordcthere was a similar patch a while back
gordcyou could probably copy it... it was never merged20:11
ankita_waghgot it... thanku20:12
*** Longgeek has quit IRC20:13
*** nsaje___ has joined #openstack-ceilometer20:15
*** nsaje__ has quit IRC20:18
*** ViswaV has joined #openstack-ceilometer20:19
*** nsaje____ has joined #openstack-ceilometer20:20
*** nsaje___ has quit IRC20:22
*** nsaje has joined #openstack-ceilometer20:22
gordcankita_wagh: i should mention, if you find that transformer useful, you can push to have it merged.20:24
*** nsaje____ has quit IRC20:25
*** julim has joined #openstack-ceilometer20:47
*** stevebaker has joined #openstack-ceilometer20:50
stevebakerhey, I'm wondering if a ceilometer commit has caused a regression in check-heat-dsvm-functional-mysq20:52
stevebakerwe have a voting job which is failing every run on the ceilometer alarm integration test
*** idegtiarov_ has quit IRC21:01
*** ankita_wagh has quit IRC21:03
*** prad has quit IRC21:05
openstackLaunchpad bug 1468907 in heat "heat_integrationtests CeilometerAlarmTest.test_alarm FAILED" [Critical,Triaged]21:07
*** ddieterly has quit IRC21:11
*** ankita_wagh has joined #openstack-ceilometer21:16
*** Longgeek has joined #openstack-ceilometer21:21
*** xmltok_ has joined #openstack-ceilometer21:25
gordc stevebaker sorry commented on bug.21:25
*** xmltok_ has left #openstack-ceilometer21:25
gordcstevebaker: it's being reverted... problem is there are bigger issues in gate which are blocking the revert21:26
stevebakerok, I'll mark mine as a dupe21:26
gordcstevebaker: cool cool21:27
stevebakergordc: is this a different issue?
gordcstevebaker: yes that looks different21:29
gordcwe recently made our id fields smaller.21:29
stevebakerah, I'm pretty sure we've used the arn format since forever. it is big21:30
gordcwe expect uuids... right now we still give quite a big buffer of 255 i believe21:30
gordcstevebaker: i'm not familiar with arn format... how big is that? we can revert the id fix but i'd rather not if the failure is because odd test data21:32
stevebakerthe non-uuid bit in the middle could be up to 25521:33
stevebakerwe should probably just switch to uuid, which is the last element21:33
gordcoh... :( yeah, didn't think anyone was pushing ids that large21:34
gordcthe reason we made it smaller was because it's impossible to build indexes when all our ids are varchar(1024)21:34
stevebakerI'll attach the bug to ceilo and heat, but likely the fix will be to heat21:35
gordcstevebaker: yes, that'd be ideal. let me know if that's not possible21:35
gordcwe just merged it so it can be reverted... (it's in liberty-1 so i might need to figure out migration rules there)21:36
*** shardy has quit IRC21:40
*** Longgeek has quit IRC21:40
stevebakerI've raised
openstackLaunchpad bug 1468916 in heat "stack arn is too big for shorter ceilometer resource_id column" [High,Triaged]21:40
*** julim has quit IRC21:51
gordcstevebaker: cool cool. just added a comment to it.21:52
openstackgerritChris Dent proposed openstack/ceilometer: Make ceilometer work correctly when hosted with a SCRIPT_NAME
*** jwcroppe_ has joined #openstack-ceilometer22:05
*** jwcroppe has quit IRC22:07
*** edmondsw has quit IRC22:27
*** eglynn has joined #openstack-ceilometer22:31
*** dontalton has joined #openstack-ceilometer22:32
*** jruano has quit IRC22:35
*** ddieterly has joined #openstack-ceilometer22:37
*** Longgeek has joined #openstack-ceilometer22:37
*** ddieterly has quit IRC22:38
*** Longgeek has quit IRC22:42
openstackgerritMerged openstack/python-ceilometerclient: Updated from global requirements
*** dontalton has quit IRC22:45
*** dhellmann has quit IRC22:46
*** dhellmann has joined #openstack-ceilometer22:46
*** jfluhmann has quit IRC23:18
*** rbak has quit IRC23:21
*** rbrooker has quit IRC23:22
*** rbak has joined #openstack-ceilometer23:31
*** rbak has quit IRC23:34
*** ddieterly has joined #openstack-ceilometer23:43

Generated by 2.14.0 by Marius Gedminas - find it at!