Friday, 2014-08-22

*** alexpilotti has joined #openstack-ceilometer00:11
*** rbak has joined #openstack-ceilometer00:21
*** rbak has quit IRC00:21
*** mikedillion has quit IRC00:33
*** ccrouch is now known as ccrouch-afk00:42
*** sbfox has joined #openstack-ceilometer00:49
*** promulo has quit IRC00:52
*** promulo has joined #openstack-ceilometer00:53
*** nosnos has joined #openstack-ceilometer00:59
*** harlowja is now known as harlowja_away01:00
*** deepthi has joined #openstack-ceilometer01:04
*** deepthi has quit IRC01:11
*** deepthi has joined #openstack-ceilometer01:25
*** fnaval has quit IRC01:26
*** Fabio has joined #openstack-ceilometer01:26
*** deepthi has quit IRC01:31
*** harlowja_away is now known as harlowja01:40
*** alexpilotti has quit IRC01:44
*** alexpilotti has joined #openstack-ceilometer01:47
*** alexpilotti has quit IRC01:48
*** zul has joined #openstack-ceilometer01:55
*** Fabio has quit IRC01:56
*** yasukun has joined #openstack-ceilometer01:58
*** nosnos has quit IRC02:03
*** nosnos has joined #openstack-ceilometer02:04
*** zul has quit IRC02:05
*** nosnos_ has joined #openstack-ceilometer02:08
*** nosnos has quit IRC02:08
*** thomasem has joined #openstack-ceilometer02:17
*** thomasem has quit IRC02:17
*** thomasem has joined #openstack-ceilometer02:18
*** fnaval has joined #openstack-ceilometer02:25
*** sbfox has quit IRC02:40
*** harlowja is now known as harlowja_away02:49
*** promulo has quit IRC02:52
*** nosnos_ has quit IRC02:56
*** deepthi has joined #openstack-ceilometer03:05
*** sbfox has joined #openstack-ceilometer03:11
openstackgerritOpenStack Proposal Bot proposed a change to openstack/ceilometer: Updated from global requirements
*** nosnos has joined #openstack-ceilometer03:40
yasukunHow can I add some meters into meter-list?03:52
*** thomasem has quit IRC03:52
*** jaypipes has joined #openstack-ceilometer04:05
*** yatin has joined #openstack-ceilometer04:18
*** idegtiarov has joined #openstack-ceilometer04:20
*** asalkeld has joined #openstack-ceilometer04:26
*** idegtiarov has quit IRC04:29
*** llu_linux has joined #openstack-ceilometer04:33
*** llu has quit IRC04:34
*** idegtiarov has joined #openstack-ceilometer04:45
*** fnaval has quit IRC04:47
*** yatin has quit IRC04:48
openstackgerritPrabhakar Kudva proposed a change to openstack/ceilometer: Added a new metric for predictive failure for host. Spec has been added in ceilometer-specs corresponding to this change.
openstackgerritPrabhakar Kudva proposed a change to openstack/ceilometer: Predictive Failure Alert Metric added Unit test added specs added to ceilometer-specs Fixed jenkins pep8 errors
openstackgerritPrabhakar Kudva proposed a change to openstack/ceilometer: Fixed some jenkins errors
*** shakamunyi has quit IRC05:03
*** deepthi has quit IRC05:27
*** yatin has joined #openstack-ceilometer05:28
*** deepthi has joined #openstack-ceilometer05:39
*** harlowja_away has quit IRC05:44
*** _nadya_ has joined #openstack-ceilometer05:55
*** idegtiarov has quit IRC06:02
openstackgerritOpenStack Proposal Bot proposed a change to openstack/ceilometer: Imported Translations from Transifex
*** _nadya_ has quit IRC06:23
*** chrisf has joined #openstack-ceilometer06:25
*** k4n0 has joined #openstack-ceilometer06:35
stevelleyasukun: if you add a sample, the meter will be added automatically06:50
stevelleyasukun: this will add a new meter called 'fake' for example06:52
*** _nadya_ has joined #openstack-ceilometer07:17
*** sbfox has quit IRC07:32
yasukunstevelle: thanks for responding. it seems usefull. But I meant some kinds of measurament (like memory.usage) can be seen meter-list in Havana version.07:35
yasukunstevelle: And a while ago, I heard from my friend that Havana version does not supoort memory.usage mesurament. Is it true?07:36
stevelleSorry, I don't know the answer on that.  I only started working with OpenStack in July this year.07:37
yasukunno problem. thank you.07:38
stevelleI am not seeing memory metrics in my devstack, running icehouse.07:40
*** sdake has joined #openstack-ceilometer07:41
*** sdake has joined #openstack-ceilometer07:41
yasukunstevelle: How about network.outgoing.packets? memory metrics can be seen if you are not using vsphere. (see
*** _nadya_ has quit IRC07:58
*** eoutin has joined #openstack-ceilometer08:15
*** Longgeek has joined #openstack-ceilometer08:23
*** Longgeek has quit IRC08:35
*** Longgeek has joined #openstack-ceilometer08:35
*** _nadya_ has joined #openstack-ceilometer08:35
eoutinHi guys, just wondering how to empty the ceilometer mongo db?08:37
eoutincause i'm still seeing metrics from old VMs08:37
eoutinis it safe to drop the db?08:37
*** claudiub has joined #openstack-ceilometer08:43
*** _nadya_ has quit IRC09:01
stevelleyasukun: I see network.outgoing.packets being recorded in icehouse09:04
stevelleagain, in devstack09:04
stevelleeoutin: how do you mean "safe" exactly?09:07
openstackgerritA change was merged to openstack/ceilometer: Switch to oslotest
eoutinstevelle, meaning that ceilometer will still be able to store data09:18
eoutinstevelle, cause I expect drop db to fully delete the database. I'd like to just flush/empty it09:19
stevelleeoutin: that seems like a question about mongo, more than ceilometer, based on what I undertand.  db.<collection>.remove({}) will truncate a table and retain indexes and is probably safest.09:19
eoutinstevelle, yea sure.. Ok thx for helping09:20
*** IvanBerezovskiy has joined #openstack-ceilometer09:27
*** _nadya_ has joined #openstack-ceilometer09:32
*** deepthi has quit IRC09:34
openstackgerritZhai, Edwin proposed a change to openstack/ceilometer: Add IPMI support
*** cdent has joined #openstack-ceilometer09:36
*** idegtiarov has joined #openstack-ceilometer09:53
DinaBelovajd__, eglynn-office, cdent, nsaje - here is the docco I told about yesterday - with ceilo testing on real lab (IO load, CPU load, etc.) -
DinaBelovait might be interesting for you :)09:54
* cdent looks09:54
* nsaje looks09:54
DinaBelovacdent, nsaje you're welcome :)09:54
*** deepthi has joined #openstack-ceilometer09:56
DinaBelovansaje, these english verbs make me crazy :D09:58
cdentThat's a lot of very interesting information. Was there anything that surprised you Di?10:01
cdentDinaBelova ^10:01
cdentI apparently can't type today.10:01
DinaBelovacdent, wow, what did happen with your fingers?10:02
nsajeDinaBelova: that's an amazing document and it'll be great to have!10:03
DinaBelovacdent, actually the most interesting thing was the answer to the question - if we speak about medium cloud with more-or-less many resources to poll - what would happen? Can ceilo poll once a minute? Can even more?10:03
DinaBelovacdent, it was not clear actually10:03
DinaBelovabecause it were so many talks about ceilo incompetence and disabilities10:04
* cdent nods10:04
DinaBelovaand also we wanted to approximate what kind of deployment we need if we want to use ceilo10:04
DinaBelovacan we run mongo on controllers or do we need separated nodes? bla-bla like that10:05
DinaBelovansaje, thank you sir :)10:05
cdentSeems like if you've got the resources, then the recommendation is to put mongo on its own nodes.10:05
DinaBelovacdent, exactly10:05
nsajeDinaBelova: one question: I haven't seen many load reports yet, but wouldn't it look better if CPU load was in the range of 0-100%? It's not as intuitive if you say 300% out of 1200% available, wouldn't it be better to simply say 40%?10:05
DinaBelovansaje, we used one cpu == 100%10:06
DinaBelovaand as we had 12CPU controllers it was 1200%10:06
nsajeDinaBelova: ok, as I said, I'm not familiar with load report terminology best practices :)10:06
nsajeDinaBelova: just my layman opinion10:07
DinaBelovathat's because in Russia CPU load percentage is always for one cpu, I guess10:07
DinaBelovalocal thing10:07
DinaBelovapossibly you're right - your variant will look more understandable  Iguess10:07
cdentI think they both make sense, as long as you choose one and make it clear which is being used.10:07
nsajeDinaBelova: pardon me for asking, do you take offence at 'in soviet russia' jokes? :)10:08
cdentIf you're the sort of person who is used to staring at `top` then %ages over 100 make sense10:08
DinaBelovacdent, btw, the problem is not only with the fact ceilo procudes much data to store on controllers - the thing is that with SATA disks we're close to the device saturaiton sometimes10:08
* cdent laughs at nsaje 10:08
DinaBelovacdent, hehe, yes :)10:08
DinaBelovawe're all top users here:S10:08
*** idegtiarov has quit IRC10:08
DinaBelovaif we had SSD controllers, I guess they'll be much more expensive, but we won't have device saturation stuff :)10:09
cdentI think there's room for a some kind of relialbe write-caching-queue thing in the ceilo stack.10:10
DinaBelovacdent, I don't know if that topic was discussed really...10:10
DinaBelovacdent, but that's important thing in real life for sure10:11
*** idegtiarov has joined #openstack-ceilometer10:11
DinaBelovansaje, about 'in soviet russia' jokes we're laughing really :)10:11
DinaBelovansaje, that's kind of funny opinion about current Russians in the eyes of all the world10:12
nsajeDinaBelova: ok, in soviet russia, CPU loads you (sorry, lame joke, but couldn't resist :-) )10:13
DinaBelovanot speaking about the current situation at all and all the Putin and Russian Bear jokes around :)10:13
DinaBelovansaje, hahaha!!! :D lol10:13
cdentthat's so bad its good10:14
DinaBelovansaje - :)10:14
DinaBelovacdent, ;)10:15
cdentthat is, by far, the best openstack t-shirt I've ever seen10:15
DinaBelovacdent, at least it's funny :)10:15
nsajeah, yeah, I remember that t-shirt! I was laughing hard when I saw it in Atlanta10:16
DinaBelovacdent, nsaje - when Mirantis folks were trying to imagine how to make our t-shirts nice, funny ones, it ere two opinions :) the 'soviet russia' thing won :) although it were thoughts that'll be kind of aggressive for elder people10:17
DinaBelovaand all that kind of things10:17
DinaBelovasad and funny really10:17
cdentCan you guys help me figure out why the ceilometer middleware for swift runs in a local devstack but not in the gate? devstack/lib/swift suggests that if both s-proxy and ceilo are enabled, it should.10:17
nsajeyeah, it's always a fine line with things like these10:18
DinaBelovacdent, a-ha, probably that's devstack-gate thing10:18
cdentI'm clearly missing some piece of the pie, I couldn't find anything in devstack-gate10:18
DinaBelovaor tempest gives some flag for that to devstack-gate10:18
nsajecdent: phew, devstack is black magic to me :)10:18
cdentat least a black hole, you look in and never come back10:19
DinaBelovansaje, ;)10:19
DinaBelovacdent, let me try to find it out10:19
nsajeDinaBelova: if only :-)10:19
cdentThanks DinaBelova, and if in the process you could let me know the process you followed to find out, that would be great too. It seems the secrete to success round here is knowing who to ask...10:20
DinaBelovacdent, well my fiancé is infra-root, sometimes that helps :D I have lots of lections :D10:21
nsajeDinaBelova: there's some good connections right there! I'll know who to bug about devstack in the future!10:24
DinaBelovansaje, that10:24
DinaBelovais the reason I'm more or less good at infra stuff10:25
DinaBelovacdent, hehe, I see something really strange. As far as I remember, we need to define SWIFT_HASH, SWIFT_REPLICAS and SWIFT_DATA_DIR to work in devstack. And what's I see in devstack-gate: - SWIFT_DATA_DIR is defined only for the grenade10:33
DinaBelovaSWIFT_HASH and SWIFT REPLICA are defined without any if's10:34
cdentif you look at logs, though, the proxy is running:
DinaBelovaso possibly devstack job is not running Swift as it was not defined in the generag conf section?10:35
DinaBelovathat's for the grenade job10:35
DinaBelovayou were asking for other one yesterday10:35
DinaBelovais smth wrong with the grenade job? No swift for it?10:35
cdentI'm perhaps confused on my terminology10:36
cdentBasically what I want is for this: to be possible10:36
DinaBelovaa-ha, and the grenade job is failing10:36
cdentand at the moment the grenade-related jobs fail because the swift-proxy is not running the ceilo middleware (because the proxy config doesn't have it, despite lib/swift ...)10:37
cdentso presumably something is controlling the swift-proxy config, but I can't figure out what10:37
DinaBelovaokay, I got the problem10:38
DinaBelovahm, possibly the main problem is that something default is used?10:39
DinaBelovacdent, hehe, no, that's not the reason10:40
DinaBelovain the devstack
*** nsaje has quit IRC10:40
* cdent nods10:41
DinaBelovaso if it finds the service ceilometer, it should add that to the proxy conf10:41
cdentthat's why it works locally10:41
DinaBelovawell, devstack gate uses almost the same - devstack10:41
DinaBelovapossibly in some other way I guess10:41
*** nsaje has joined #openstack-ceilometer10:42
DinaBelovacdent - the strange moment10:50
DinaBelovaI see no ceilo service enabled10:50
DinaBelovafor the grenade local.conf10:50
cdentyeah, if you want to run grenade locally you have to cook that10:50
cdentbut for the gate it is enabled10:51
DinaBelovaa-ha, thnaks10:51
DinaBelovathat's crazy10:51
DinaBelovaif devstack will find ceilo it'll add it to the proxy xonf10:51
DinaBelovain theory I guess :s10:51
cdentyes, it's utterly bewildering10:51
cdentI'm glad it's not just me being confused ;)10:52
DinaBelovano-no, that's definately really strange10:52
DinaBelovacdent - does grenade use the freshest devstack?10:55
DinaBelova~month or smth like that ago swift+ceilo was turned off in devstack10:55
DinaBelovaI turned it on10:55
cdenthmm, I don't know10:56
DinaBelovaafter ceilo log levels were fixed enough not to flood the syslog10:56
DinaBelovathat's crazy idea, but who knows10:56
cdenthmmm, looking at grenaderc looks like perhaps it is using an old devstack10:58
DinaBelovawe should find out what is the BASE_DEVSTACK_BRANCH :)10:59
* cdent sighs10:59
cdenti think that's in grenaderc10:59
DinaBelovastable ones10:59
*** alexpilotti has joined #openstack-ceilometer10:59
cdentstable icehouse10:59
cdentso it won't have your changes10:59
cdentHow then, can people add scenario tests?11:00
DinaBelovacdent, dunno :(11:01
DinaBelovalet me find my commit to be sure I'm right and we don't have it11:01
DinaBelova6 weeks ago11:01
DinaBelovayeah, definitely no my change in grenade devstack11:01
DinaBelovacdent, we may backport needed changes to devstack and ceilo11:02
DinaBelovato the stable icehouse11:03
DinaBelovaI guess that's the main thing to be done here11:03
DinaBelovadunno if Sean will be happy with it :)11:03
cdentlooks like the disabling happened 19 march11:03
DinaBelovacdent, yeag
DinaBelovaso we need this backported to devstack and this - to the ceilo icehouse branches11:05
DinaBelovaafter that you'll be unblocked11:05
DinaBelovaI may try to propose changes and communicate with Sean to accept them :)11:06
cdentI'm unfamiliar with the backporting process, how does that work?11:06
DinaBelovaall of our openstacky guys love nice girls :)11:06
DinaBelovawell, we'll just add the needed commits to the icehouse branches11:06
DinaBelovaof needed projects11:06
DinaBelovajust the same - propose changes, etc11:07
DinaBelovaand core-reviewers review, etc.11:07
cdentah, okay, so its just a matter of making sure the parent is right11:07
DinaBelovachanges might be -2'ed if core-reviewers will think the change are not acceptable11:07
cdentno special magic11:07
DinaBelovacdent, yeah - all these changes are not dependent on others11:08
DinaBelovanothig special :)11:08
DinaBelovathe main thing is to persuade that is useful :)11:08
DinaBelovalet me try this girlish magic :D11:08
DinaBelovacdent, np :)11:09
DinaBelovaat least we know the reason :)11:09
cdentI've updated the original patchset with some notes on what's going on.11:11
*** cdent has left #openstack-ceilometer11:11
*** cdent has joined #openstack-ceilometer11:11
cdentSee, told you I can't type, I close the wrong window.11:11
DinaBelovadon't worry :)11:12
DinaBelovasometimes that happens :)11:12
*** nosnos has quit IRC11:17
DinaBelovacdent -
*** nosnos has joined #openstack-ceilometer11:17
DinaBelovadevstack backport is coming11:17
DinaBelovacdent - and here is the devstack thing
*** nosnos has quit IRC11:21
DinaBelovacdent, I'll catch Sean this evening11:22
DinaBelovalet's see what will happen :)11:22
*** deepthi has quit IRC11:23
*** _nadya_ has quit IRC11:24
cdentThanks DinaBelova11:32
DinaBelovacdent, np :)11:33
*** deepthi has joined #openstack-ceilometer11:36
asalkeldhi DinaBelova, got your email11:37
DinaBelovaasalkeld, hello :)11:37
asalkeldgreat work on that performance analysis11:38
DinaBelovaasalkeld, thanks so much11:38
DinaBelovawe had kind of rush about it :)11:38
DinaBelovawe had the 20nodes lab just for some days, and, as usual, everything went wrong11:38
asalkeldDinaBelova, any comparision with the other db types?11:38
DinaBelovaasalkeld, I can give you results of other kind of profiling (with measurements of how many messages per second can we process, etc)11:39
openstackgerritNejc Saje proposed a change to openstack/ceilometer: Central agent work-load partitioning
DinaBelovathe thing with last docco is that we used MOS11:39
DinaBelovaas that was kind of requirement11:39
DinaBelovaand MOS currently supports only mongo11:40
DinaBelovaasalkeld -
DinaBelovathat is also Icehouse results11:40
DinaBelovabut on smaller labs, with backends profiling11:40
DinaBelovamongo, mysql, hbase11:41
DinaBelovathis dc is elder, and we had not that shine lab that time :)11:41
*** _nadya_ has joined #openstack-ceilometer11:42
asalkeldso many db's makes it more difficult to optimise11:42
DinaBelovahehe, indeed :)11:42
DinaBelovaityaptin is working now on automatisation of the performance testing (like after-milestone-cut Jenkins job) and nice wiki page to present the results11:43
DinaBelovaasalkeld, so hopefully we'll have something so nice and really helpful really soon :)11:43
DinaBelovathe main problem here is that it makes sense to run job like that only on multi-node devstack installation, and that's kind of little challenge11:44
asalkelddoes stacktach poll?11:46
asalkeldcan't remember11:46
asalkeldwhat's the "real life" period for polling11:46
DinaBelovaasalkeld, unfortunately have no idea about the stacktach11:47
*** ildikov has joined #openstack-ceilometer11:50
DinaBelovaasalkeld, as I lurked now in the - stacktach is about events (notifications coming from the services), not the polling thing11:53
DinaBelovaasalkeld - am I terribly wrong?11:53
asalkeldDinaBelova, i think that's right11:53
asalkeldi just don't want us to have an unfar comprision11:54
*** deepthi has quit IRC11:54
DinaBelovaasalkeld, we tryed notification as well (like 1000VMs created a t the same time), CPU load was ridiculous11:54
DinaBelovanothing compared with polling thing11:54
asalkeldi'd set poll == 1 day11:54
asalkeldor nothing11:54
asalkeldthis is our problem, we are doing monitoring and metering11:55
asalkeldand those other projects are going to win as they can specialize11:55
DinaBelovaasalkeld, yeah, that's kind of big problem11:55
asalkeldyour performance test is really a monitoring secenario11:56
asalkeldnot metering11:56
asalkeldceilometer is doing both11:56
asalkeldDinaBelova, when you said about the notifications "cpu was ridiculous" - did you mean very high or low load11:57
nsajewe're quite weak on the events side11:57
DinaBelovavery-very low11:57
asalkeldso it could handle it easliy11:58
DinaBelovaasalkeld, yes, indeed11:58
asalkeldpersonally i am not at all against the seperate api idea11:59
DinaBelovaasalkeld - that's about the gnocchi thing?11:59
asalkeldDinaBelova, no having an api that stacktach and the other monitoring proj (forgot the name) could implement12:00
DinaBelovaasalkeld, aha, got your idea... although as far as I remember Eoghan and Ildiko were really against any monitoring things... ildikov at least for sure :)12:01
DinaBelovaasalkeld, yeah, there is this project12:03
*** eoutin has quit IRC12:03
asalkeldthat's what i meant more12:03
DinaBelova^^  - I meant inside the ceilo12:03
asalkeldo, totally12:03
asalkeldone option is to push the monitoring stuff off to monasca12:04
DinaBelovaasalkeld, yes, otherwise we have things working not great at any pov12:04
asalkeldand just do the metering part12:04
asalkeldthen at least we don't have to do polling12:04
DinaBelovaasalkeld, well, while the mid-cycle thing it were lots of talks about monasca as well, although, without clear result12:04
asalkeldand the performance wi be greate12:04
DinaBelovaasalkeld, well, that12:05
DinaBelova... is the solution12:05
DinaBelovasorry, typos :)12:05
asalkeldme too12:05
DinaBelovahehe, just saw :)12:05
*** zul has joined #openstack-ceilometer12:08
*** yasukun has quit IRC12:12
*** eoutin has joined #openstack-ceilometer12:18
openstackgerritOpenStack Proposal Bot proposed a change to openstack/ceilometer: Updated from global requirements
*** _nadya_ has quit IRC12:28
openstackgerritOpenStack Proposal Bot proposed a change to openstack/pycadf: Updated from global requirements
openstackgerritOpenStack Proposal Bot proposed a change to openstack/python-ceilometerclient: Updated from global requirements
*** yatin has quit IRC12:35
*** zul has quit IRC12:40
*** gordc has joined #openstack-ceilometer12:57
*** _nadya_ has joined #openstack-ceilometer13:03
openstackgerritJulien Danjou proposed a change to openstack/ceilometer: Switch to oslo.utils
gordcanyone have luck figuring out that pecanwsme doc issue?13:04
ildikovgordc: I couldn't reproduce it locally on precise13:05
gordcyeah i cant see it locally on fedora2013:05
ildikovgordc: I have afresh trusty, I've just installed tox and started the docs build13:05
*** asalkeld has quit IRC13:06
*** _nadya_ has quit IRC13:06
gordcildikov: cool cool. if it doesn't work ask clarkb  if he's tried rebooting the system? :)13:06
ildikovgordc: it seems my env is not ready for tox yet...13:06
ildikovgordc: I have another internal task now, I will let you know later if I have any results13:07
gordcildikov: no problem. no rush13:07
ildikovgordc: but dhellmann wrote on the dev ML, that he couldn't reproduce it on trusty, so I don't have to much hope in reproducing it locally... :(13:07
gordcyeah. i don't think my coworker could reproduce on trusty either.13:08
*** joesavak has quit IRC13:12
*** aviau has quit IRC13:14
*** zul has joined #openstack-ceilometer13:15
*** aviau has joined #openstack-ceilometer13:26
gordcsileht: are there any issues with oslodb patch if we pass in **dict(conf.database.keys())?
*** thomasem has joined #openstack-ceilometer13:28
silehtgordc, :D if it works good idea13:28
gordcsileht: i'll give it a try... if it works i'll ask alexei to update... or did you want to update your patch? since you posted it first.13:29
silehtgordc, I have given my patch to alexei, it's up to him13:30
gordccool cool. i'll see if he's still interested (assuming this works)13:30
gordcnsaje: regarding central agent patch, just wondering what proper way to set it up is... do i just enable partitioning in conf and start two central agent services?13:33
openstackgerritJulien Danjou proposed a change to stackforge/gnocchi: swift: avoid doing a HEAD to check container presence
*** dmsimard_away is now known as dmsimard13:35
*** jasondotstar has joined #openstack-ceilometer13:37
*** julim has joined #openstack-ceilometer13:37
*** thomasem has quit IRC13:38
*** thomasem has joined #openstack-ceilometer13:39
DinaBelovagordc, ildikov - I've posted link while you were not online :) - performance benchmarking thing I've mentioned yesterday13:48
openstackgerritJulien Danjou proposed a change to stackforge/gnocchi: sqlalchemy: use new DBReferenceError from oslo.db
*** russellb is now known as rustlebee13:54
*** _nadya_ has joined #openstack-ceilometer13:55
*** k4n0 has quit IRC13:55
gordcDinaBelova: just curious. were you guys running a single instance of collector and notification agents?13:56
DinaBelovagordc - 6 collectors and notification agents13:57
DinaBelovatwo for every controller13:57
gordcDinaBelova: whoops just noticed :)13:57
DinaBelovait was somewhere in this doc13:57
gordcthose are instances and not workers right?13:57
DinaBelovahm, sorry?13:57
DinaBelovacan't get your question, sorry13:57
openstackgerritJulien Danjou proposed a change to stackforge/gnocchi: Remove pandas version limitation
ekarlsowhy gnocchi vs influxdb ?13:59
gordcDinaBelova: i guess they're essentially the same but you can set the number of workers as an option.13:59
gordcjust wondering how you were defining the 2 collectors...curious if there's a difference14:00
DinaBelovaah, they all work together actually - they just grab messages from the queue and process them14:00
DinaBelovathat's why load more than 700 messages per second is possible14:00
DinaBelovaekarlso, why "vs"? they are "together" :)14:01
ekarlsoDinaBelova: realy :P14:01
DinaBelovaekarlso, one moment14:01
DinaBelovaekarlso -
DinaBelovahere it is :)14:02
DinaBelovadriver to use influxdb14:02
ekarlsowhat is gnocchi though vs ceilometer ?14:02
ekarlsoin terms of events or so14:02
DinaBelovagnocchi is time-series wrapper (around time-series DB, for instance) to then be merged inside ceilo as v3 API14:03
DinaBelovait'll store samples14:03
DinaBelovathat actually is time-series info14:03
ildikovDinaBelova: thanks for the docs and all the effort for profiling in general, I will check it later, my week was about internal stuff, I need to catch up a bit now :)14:03
DinaBelovafor instance, it's not about alarms, also implemented in ceilo14:03
DinaBelovaildikov, yeah, good luck with your internal work!14:04
ildikovDinaBelova: thanks, it's always needed :)14:06
gordcekarlso: ceilometer is a collection of multiple services (polling, notifications, collection.)... as DinaBelova mentioned it's a wrapper to a tsdb...14:06
gordcekarlso: so in a ceilometer sense, it would be another db option... with possible expansion to replace collection service14:07
*** rbak has joined #openstack-ceilometer14:11
gordcDinaBelova: regarding question how you launched multiple collectors... i was just curious because iiuc, the workers option uses eventlets while just starting two collector services separately doesn't.14:19
gordcor maybe that's a wrong assumption.14:19
*** eoutin has quit IRC14:21
openstackgerritgordon chung proposed a change to openstack/ceilometer: normalise resource data
openstackgerritgordon chung proposed a change to openstack/ceilometer: switch to sqlalchemy core
*** Yathi has joined #openstack-ceilometer14:25
*** idegtiarov has quit IRC14:26
*** sdake_ has joined #openstack-ceilometer14:26
*** prad has joined #openstack-ceilometer14:35
ildikovgordc: no surprise, I couldn't reproduce the issue nor with tox neither with running build_sphinx natively... :(14:51
ildikovgordc: I didn't have the patience to check whether it occurs only with one image on the gate or on all14:54
gordcildikov: so i guess next step is to figure out where the gate machines reside and we smash them?14:54
gordci took a look at the machines they were failing on...i couldn't find a pattern14:55
ildikovgordc: I'm kinda out of ideas14:57
ildikovgordc: I mean without checking it on the exact env used on the gate14:57
gordcildikov: no problem. i'll keep debuggig. i'm hoping it magically resolves itself14:58
ildikovgordc: I just wanted to say the same :)14:58
*** sdake has quit IRC14:59
*** sdake has joined #openstack-ceilometer14:59
ildikovgordc: I need to run now, but will try to keep an eye on it15:01
gordcildikov: ok. ttyl15:04
*** tongli has joined #openstack-ceilometer15:07
ildikovgordc: I still see the keystone issue in my devstack env, so I added a comment to that bug15:08
ildikovgordc: but well, TBH that issue is just as weird as this docco one...15:09
ildikovgordc: anyway, ttyl :)15:10
gordcgreat. love weird bugs... /sarcasm15:12
*** tongli has quit IRC15:14
*** cdent_ has joined #openstack-ceilometer15:15
*** cdent has quit IRC15:17
*** cdent_ is now known as cdent15:17
*** sdake has quit IRC15:21
*** shakamunyi has joined #openstack-ceilometer15:24
*** sbfox has joined #openstack-ceilometer15:24
openstackgerritgordon chung proposed a change to openstack/ceilometer: update requirements
openstackgerritgordon chung proposed a change to openstack/ceilometer: update requirements
*** IvanBerezovskiy has left #openstack-ceilometer15:36
*** _nadya_ has quit IRC15:40
*** shakamunyi has quit IRC15:41
openstackgerritgordon chung proposed a change to openstack/ceilometer: update requirements
*** atan8 has joined #openstack-ceilometer15:45
*** Yathi has quit IRC15:55
*** ildikov has quit IRC15:57
*** alexpilotti has quit IRC16:00
*** jaypipes is now known as leakypipes16:02
*** atan8 has quit IRC16:06
*** promulo has joined #openstack-ceilometer16:06
*** Dieterbe has quit IRC16:20
*** Dieterbe has joined #openstack-ceilometer16:21
*** Dieterbe has joined #openstack-ceilometer16:21
*** fnaval has joined #openstack-ceilometer16:23
*** cdent has quit IRC16:29
*** packet has joined #openstack-ceilometer16:30
rbakDoes anyone know a way to get the total number of neutron nets or cinder volumes from ceilometer data?16:34
*** openstackgerrit has quit IRC16:34
*** claudiub has quit IRC16:38
*** alexpilotti has joined #openstack-ceilometer16:44
*** ankur has joined #openstack-ceilometer16:49
ankurhi ..anyone around?16:50
ankurI am facing issue while configuring ceilometer service endpoint16:51
ankurdefault port 8775 is already busy in my system16:51
ankurand while configuring other ports, I am getting error of connection refused16:52
ankurAny idea which port I should use while creating keystone service point for ceilometer?16:54
ankurI am using  keystone endpoint-create --region RegionOne --service_id b8eaf53aa6a240f796a2a05a3f596d93 --publicurl "" --adminurl "" --internalurl "
*** ddutta has joined #openstack-ceilometer17:07
*** claudiub has joined #openstack-ceilometer17:18
ankurddutta: hi..any idea of my issue?17:19
dduttaankur: which issue17:19
*** Longgeek has quit IRC17:19
ankuractually during configuration of ceilometer I am facing issue of port to assign it17:20
ankurI am pasting the command loggs'17:21
*** Longgeek has joined #openstack-ceilometer17:21
*** packet has quit IRC17:22
*** atan8 has joined #openstack-ceilometer17:22
ankurnecadmin@cindernode:~$ keystone service-create --name=ceilometer --type=metering --description="Ceilometer Service" +-------------+----------------------------------+ |   Property  |              Value               | +-------------+----------------------------------+ | description |        Ceilometer Service        | |   enabled   |               True               | |      id     | d98ad03614a743fabda13aece2e0646d | |17:22
dduttahmmm I need to check this ... can I Catch you in 30mins ....17:23
ankurPlease refer -->
ankurddutta: k17:24
mnaserankur: ceilometer --debug meter-list17:24
mnaseralso that url seems wrong17:25
mnaserceilometer doesnt use port 5000 afaik17:25
mnaserit uses port 877717:25
*** ddutta has quit IRC17:25
ankurmnaser: you are right ..port no.should be wrong but I don't know how to find correct one..8775 port is already occupied in my system17:26
mnaserits 8777, not 8775.17:26
ankurk ..I try with this17:26
*** openstackgerrit has joined #openstack-ceilometer17:29
*** ddutta has joined #openstack-ceilometer17:32
*** ankur has quit IRC17:33
*** zul has quit IRC17:33
*** harlowja has joined #openstack-ceilometer17:36
*** ddutta has quit IRC17:36
*** ankur has joined #openstack-ceilometer17:37
ankurmnaser: There?17:37
ankurPlease refer above link17:37
mnaseris your ceilometer api not running17:38
ankurhow can I confirm it?17:38
mnaserwell because its not responding17:38
mnaserdid you change your port settings17:38
ankuryah I have change it to 877717:38
mnasernot sure why you changed it17:40
mnaserits that by default17:40
mnaseri wouldnt mess with the configs unless you know what you're doing17:40
gordcankur check your ceilometer logs usually under /var/log/17:41
*** sbfox has quit IRC17:41
ankurNo I didn'tchange it...I am configuring ceilometer first time on devstack17:41
gordcif you enable_service ceilometer-<services> that should configure everything for you on devstack17:42
ankurmnaser: I am not sure why I am facing all these issues...because previously, I was able to configure ceilometer in one go17:43
mnaseri havent really ever used devstack + ceilometer17:43
*** sbfox has joined #openstack-ceilometer17:44
ankurmnaser: k ..but any idea for this issue?17:45
mnaserafaik... what gordc is the way to go17:45
mnaserid just undo the whole openstack and run it again from scratch with enable_service17:45
*** leakypipes has quit IRC17:46
ankurgordc: I can't run again because I am working on rented system and I have faced several issues which I tried to reconfigure devstack again . So, I am bit scared that I can loose my existing system17:49
*** packet has joined #openstack-ceilometer17:50
gordcankur: you don't need to rerun stack... if you're on devstack jump into the screen session and look at ceilometer-api17:52
ankurgordc: can I run ./ again without run ./ and just change in localrc as "enable_service ceilometer"?17:52
gordcankur: that i'm not usre... there's but i don't know what that does tbh.17:52
gordci can give you the api command that devstack executes17:53
ankurgordc: k..please provide17:53
*** packet has quit IRC17:55
*** ankur has quit IRC18:00
*** tongli has joined #openstack-ceilometer18:05
*** leakypipes has joined #openstack-ceilometer18:08
*** claudiub has quit IRC18:21
*** _nadya_ has joined #openstack-ceilometer18:21
*** tongli has quit IRC18:38
*** sbfox has quit IRC18:43
*** _nadya_ has quit IRC18:43
*** zul has joined #openstack-ceilometer18:46
*** sbfox has joined #openstack-ceilometer18:49
*** ildikov has joined #openstack-ceilometer18:51
*** prad has quit IRC18:55
*** alexpilotti has quit IRC18:56
*** rbak has quit IRC19:05
*** fnaval has quit IRC19:12
*** maksimov has joined #openstack-ceilometer19:20
*** sdake_ has quit IRC19:36
*** sdake has joined #openstack-ceilometer19:36
*** sdake has joined #openstack-ceilometer19:36
*** renatoarmani has joined #openstack-ceilometer19:37
*** thomasem_ has joined #openstack-ceilometer20:01
*** thomasem has quit IRC20:01
*** thomasem_ has quit IRC20:05
*** rwsu has quit IRC20:13
*** sdake_ has joined #openstack-ceilometer20:57
*** Longgeek has quit IRC20:57
*** zul has quit IRC21:01
nealph_ankur: as gordc noted, you can stop/start ceilometer-api by going into screen and stopping/restarting21:02
nealph_Derp. Just saw the note about the localrc update. Rejoin-stack only reconnects to an existing screen hope there.21:04
nealph_ankur: but note that keystone commands are fairly descriptive. To address the root cause you might consider creating another keystone endpoint for ceilometer: keystone service-list, keyston endpoint-create with the ceilometer service ID.21:07
openstackgerritgordon chung proposed a change to openstack/ceilometer: re-add hashseed to avoid gate error
*** atan8 has quit IRC21:10
*** renatoarmani has quit IRC21:14
*** sdake_ has quit IRC21:31
openstackgerritgordon chung proposed a change to openstack/ceilometer: re-add hashseed to avoid gate error
*** _nadya_ has joined #openstack-ceilometer21:44
*** jasondotstar has quit IRC21:48
*** _nadya_ has quit IRC21:49
*** gordc has quit IRC21:55
*** gordc has joined #openstack-ceilometer22:22
*** dmsimard is now known as dmsimard_away22:23
*** gordc has quit IRC22:38
*** thomasem has joined #openstack-ceilometer22:59
*** thomasem has quit IRC23:06
*** thomasem has joined #openstack-ceilometer23:07
*** eglynn-office has quit IRC23:07
*** eglynn-office has joined #openstack-ceilometer23:09
*** thomasem has quit IRC23:11
openstackgerritA change was merged to openstack/pycadf: add federated credential
openstackgerritA change was merged to openstack/python-ceilometerclient: Removed undefined method in file
*** atan8 has joined #openstack-ceilometer23:24
*** atan8_ has joined #openstack-ceilometer23:25
openstackgerritEric Pendergrass proposed a change to openstack/python-ceilometerclient: Verify alarm found before modifying
*** atan8 has quit IRC23:28
openstackgerritOpenStack Proposal Bot proposed a change to openstack/pycadf: Updated from global requirements
openstackgerritOpenStack Proposal Bot proposed a change to openstack/python-ceilometerclient: Updated from global requirements
openstackgerritPrabhakar Kudva proposed a change to openstack/ceilometer: All changes to predictive failure Added unit tests Match specs submitted Abandoned previous commits Work in Progress
*** sbfox has quit IRC23:48
*** asalkeld has joined #openstack-ceilometer23:57

Generated by 2.14.0 by Marius Gedminas - find it at!