Friday, 2015-09-11

*** promulo__ has joined #openstack-ceilometer00:13
*** promulo has quit IRC00:17
*** ftco has quit IRC00:37
*** ftco has joined #openstack-ceilometer00:40
*** kun_huang has joined #openstack-ceilometer01:17
liusheng_hi, ildikov, pradk, sileht, around here ?01:23
liusheng_could you please take a look at
liusheng_rally peaple told me that the event.list() API request alwayls timeout in gate job, may it is because the large events amount, this change can extremly reduce the unneccessary events.01:27
*** rjaiswal has quit IRC01:30
boris-42liusheng_: uuu01:40
boris-42liusheng_: nice thinks01:40
boris-42liusheng_: if you would like we can check that01:40
*** tiny-hands has joined #openstack-ceilometer01:49
liusheng_boris-42, thanks, you mean check with rally ?01:50
boris-42liusheng_: we can create depends-on patch in rally01:51
boris-42liusheng_: and see the results01:51
boris-42liusheng_: this patch actually helps to fix issue*01:52
liusheng_boris-42, cool, thanks, so this cross-project depends can check the issue, does this patch try to find the reason of the api timeout bug?01:54
liusheng_boris-42, kun_huang has talked with me about that issue01:54
boris-42liusheng_: exactly that01:54
boris-42liusheng_: and it actually fixes the issue01:54
boris-42liusheng_: so reverting that patch helps a lot01:54
*** nadya has joined #openstack-ceilometer01:55
kun_huangliusheng_: boris-42 the test is running01:55
boris-42kun_huang: ^ so in about 1 hour we will know does it help or not01:55
liusheng_boris-42, maybe I was wrong01:55
liusheng_boris-42, if this revert can fix that, my patch do different things with this gordon's patch :(01:56
boris-42liusheng_: so that's sad01:57
boris-42kun_huang: I believe we should remove from now Ceilometer stuff from Rally gates01:57
boris-42kun_huang: and put revert of andreykurilin patch01:57
boris-42kun_huang: and wait until things get better01:57
kun_huangboris-42: hehe okay01:58
boris-42kun_huang: I would like to cut releese as soon as possible01:58
boris-42kun_huang: and we need to merge some patches01:59
liusheng_boris-42, let us wait the result,  since the patch will reduce most events and the api timout seems like the amount of return is too large01:59
*** nadya has quit IRC01:59
*** ankita_w_ has quit IRC02:00
*** ankita_wagh has joined #openstack-ceilometer02:00
boris-42liusheng_: ok02:00
*** ankita_wagh has quit IRC02:05
*** ankita_wagh has joined #openstack-ceilometer02:28
*** jaypipes has quit IRC02:49
openstackgerritMerged openstack/ceilometer: devstack: add new option to support event-alarm
*** tiny-hands has quit IRC02:53
*** tiny-hands has joined #openstack-ceilometer03:03
*** tiny-hands has quit IRC03:05
*** changbl has quit IRC03:07
*** rohit_ has joined #openstack-ceilometer03:13
*** deepthi has joined #openstack-ceilometer03:21
*** ryanpetrello has quit IRC03:29
*** ryanpetrello has joined #openstack-ceilometer03:34
*** links has joined #openstack-ceilometer03:41
*** nadya has joined #openstack-ceilometer03:55
*** ankita_wagh has quit IRC03:56
*** ankita_wagh has joined #openstack-ceilometer03:57
*** nadya has quit IRC04:00
*** yprokule has joined #openstack-ceilometer04:29
*** rohit_ has quit IRC05:20
*** nadya has joined #openstack-ceilometer05:35
*** nadya has quit IRC05:40
*** nadya has joined #openstack-ceilometer05:48
*** nadya has quit IRC05:56
*** shardy has joined #openstack-ceilometer06:25
*** ildikov has quit IRC06:37
*** idegtiarov_ has quit IRC06:46
*** ankita_wagh has quit IRC06:59
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Remove useless base class
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Refactor keystone handling in discovery manager
*** eglynn has joined #openstack-ceilometer07:05
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Refactor keystone handling in discovery manager
*** ankita_wagh has joined #openstack-ceilometer07:14
*** ankita_wagh has quit IRC07:15
*** ankita_wagh has joined #openstack-ceilometer07:15
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Refactor keystone handling in discovery manager
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Refactor keystone handling in discovery manager
*** ildikov has joined #openstack-ceilometer07:21
*** lsmola has joined #openstack-ceilometer07:28
*** lsmola has quit IRC07:29
*** eglynn has quit IRC07:29
openstackgerritWen Zhi Yu proposed openstack/ceilometer: Remove duplicated pipeline support meter check when publish samples. Fix bug1494156.
*** cdent has joined #openstack-ceilometer07:45
*** lsmola has joined #openstack-ceilometer07:49
*** gzhai has joined #openstack-ceilometer08:00
*** gzhai has quit IRC08:02
*** gzhai has joined #openstack-ceilometer08:04
*** gzhai has quit IRC08:05
*** prnk28 has joined #openstack-ceilometer08:06
*** gzhai has joined #openstack-ceilometer08:06
*** yassine__ has joined #openstack-ceilometer08:07
*** ildikov has quit IRC08:08
*** safchain has joined #openstack-ceilometer08:08
*** Ala has joined #openstack-ceilometer08:09
prnk28Hi! I'm looking around for a mentor for the ceilometer project. Could anyone put me in touch with one?08:10
*** kun_huang has left #openstack-ceilometer08:12
*** eglynn has joined #openstack-ceilometer08:15
*** ankita_wagh has quit IRC08:16
*** nadya has joined #openstack-ceilometer08:22
*** gzhai has quit IRC08:25
*** ildikov has joined #openstack-ceilometer08:27
*** prnk28 has quit IRC08:28
*** bmotz has quit IRC08:31
*** IBerezovskiy has joined #openstack-ceilometer08:39
*** r-mibu has quit IRC08:39
*** bmotz has joined #openstack-ceilometer08:43
*** r-mibu has joined #openstack-ceilometer08:45
*** cdent has quit IRC08:52
*** cdent has joined #openstack-ceilometer08:56
*** shardy_ has joined #openstack-ceilometer09:07
*** shardy has quit IRC09:08
*** shardy_ has quit IRC09:12
*** shardy has joined #openstack-ceilometer09:13
*** nadya_ has joined #openstack-ceilometer09:20
*** nadya has quit IRC09:20
openstackgerritWen Zhi Yu proposed openstack/ceilometer: Remove duplicated pipeline support meter check
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Remove useless base class
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Refactor keystone handling in discovery manager
*** prnk28 has joined #openstack-ceilometer10:06
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Refactor keystone handling in discovery manager
*** f13o has joined #openstack-ceilometer10:22
*** idegtiarov has joined #openstack-ceilometer10:47
cdentjd__: did you mean to say "has no support for resource ID that are not UUID" or maybe "now" instead?10:50
jd__now I guess11:05
jd__cdent: corrected myself, thanks11:06
*** gordc has joined #openstack-ceilometer11:42
*** thorst has joined #openstack-ceilometer11:43
*** ildikov has quit IRC12:01
*** tiny-hands has joined #openstack-ceilometer12:02
*** ankita_wagh has joined #openstack-ceilometer12:13
*** deepthi has quit IRC12:14
*** ildikov has joined #openstack-ceilometer12:17
*** ankita_wagh has quit IRC12:17
*** edmondsw has joined #openstack-ceilometer12:31
*** ildikov has quit IRC12:36
*** changbl has joined #openstack-ceilometer12:52
*** ildikov has joined #openstack-ceilometer12:52
*** nadya_ has quit IRC12:52
*** nadya has joined #openstack-ceilometer12:57
*** lintonv has joined #openstack-ceilometer12:58
lintonvDoes ceilometer publish events using rabbitmq?12:59
lintonvI see multiple events in the output of 'ceilometer event-list'12:59
lintonvbut when I try to connect to rabbitmq using yagi or stacktack, I don't see those events13:00
gordclintonv: ceilometer publishes events to queue (whatever you have configured in oslo.messaging)13:03
gordcbut by default, the collector is grabbing those events and storing it13:04
lintonvoslo.messaging - is that in the ceilometer conf file?13:05
gordcif you want another consumer of ceilometer events, you can either: publish to a different topic, change collector to listen to another topic, or just disable collector13:05
lintonvThank you. How can I publish to a different topic? Is that in the ceilometer conf file?13:06
gordclintonv: are you using Juno, Kilo, or master?13:07
gordcyou'll need to modify event_pipeline.yaml (usually /etc/ceilometer/event_pipeline.yaml)13:08
gordcyou'll notice a publishers attribute... you shoudl change that to 'notifier://?topic=<topic name>'13:09
lintonvok. Thank you gordc. I will try that.13:09
lintonv--- sources:     - name: event_source       events:           - "*"       sinks:           - event_sink sinks:     - name: event_sink       transformers:       triggers:       publishers:           - direct://13:09
lintonvthis is all i have in there now13:09
gordclintonv: np. just a heads up, if you do this, ceilometer event-list won't work as it won't be stored/collected by ceilometer collector13:10
gordclintonv: right. just change direct:// to notifier://?topic=<any topic name>13:10
lintonvand the topic name is?13:10
gordci'm not sure how stacktach works exactly but you would need to listen to that queue13:10
lintonvgot it. Thanks again!13:11
gordclintonv: what are you trying to acheive anyways? the events logic in ceilometer overlaps stacktach already as it was originally implemented by stacktach team13:11
*** f13o has quit IRC13:12
lintonvI am trying to get notifications like instance-create, instance-delete from outside of openstack.13:12
*** nadya has quit IRC13:12
gordcif you want stacktach as it is, im not sure you need ceilometer.13:13
lintonvstacktack seems to look at the queue and notify via a web interface or api or via email, when an event occurs13:13
gordclintonv: i see. you can technically do that with alarms and events in aodh/ceilometer but you'll need to be using Aodh event alarm service.13:14
gordcnot the email part.13:14
*** logan2 has quit IRC13:15
lintonvOk. I am not tied to anything. If ceilometer does the job, that would be great. I just want a reliable way to be notified when network change and instance changes occur13:15
*** jaypipes has joined #openstack-ceilometer13:16
lintonv@gordc - is aodh/ceilometer different? as in a different repo from the native openstack?13:18
gordclintonv: well it's very new functionality in Aodh (the alarming compoenent of openstack)... the basic idea is that you create a rule to match an event_type and when detected, it should emit a http request.13:20
lintonvOk, so it's not in kilo. But is being worked on?13:20
gordclintonv: right. so we recently moved teh existing alarming functionality from Ceilometer to it's own repo: basically contains the same functionality but also supports the ability to alarm on events now as well.13:21
*** jaypipes is now known as leakypipes13:22
gordcsince it is it's own service now, i'd imagine Aodh could be used against Kilo as well.13:22
openstackgerritJulien Danjou proposed openstack/gnocchi: doc: remove details option
lintonvThat is excellent. So, it does not have to live in the openstack cluster that is emitting the events? it can live outside?13:23
gordclintonv: that said, if you have an existing stacktach service configured, you might want to just continue using that.13:23
lintonvok. This is great information. I have a few things to investigate.13:27
lintonv@gordc. I will let you know how the results look13:28
gordclintonv: cool cool13:29
*** jroll is now known as jtroll13:33
*** jtroll is now known as Guest1373013:34
*** Guest13730 is now known as jroll13:34
*** links has quit IRC13:38
*** logan2 has joined #openstack-ceilometer13:43
*** ddieterly has joined #openstack-ceilometer13:51
*** ddieterly has quit IRC13:52
*** ddieterly has joined #openstack-ceilometer13:52
*** cdent has quit IRC13:55
*** ddieterly has quit IRC13:57
*** ddieterly has joined #openstack-ceilometer14:01
*** ddieterl_ has joined #openstack-ceilometer14:08
*** ddieterl_ has quit IRC14:09
*** ddieterly has quit IRC14:11
openstackgerritJulien Danjou proposed openstack/ceilometer: doc: update devstack usage
*** ildikov has quit IRC14:22
*** yprokule has quit IRC14:24
*** nadya has joined #openstack-ceilometer14:27
*** tongli has joined #openstack-ceilometer14:35
*** liusheng_ has quit IRC14:40
*** cdent has joined #openstack-ceilometer14:41
cdentgordc: make the unicorns stop14:47
*** r-daneel has joined #openstack-ceilometer14:52
*** gordc has quit IRC14:53
*** gordc has joined #openstack-ceilometer15:12
gordccdent: what i'd miss? they're doing some network stuff right now so i pretty much dropped right when you messaged.15:14
cdentnothing, there's just all these dragons and unicorns for the past 24 hours and they won't stop15:14
gordccdent: in a dark environment or among rainbows?15:17
gordccdent: vastly different experiences.15:17
cdentit keeps changing15:18
gordcthat's a bad trip15:18
gordci did notice that they discovered the mess that is juno ceilometerclient versioning15:18
openstackgerritJulien Danjou proposed openstack/ceilometer: doc: update devstack usage
openstackgerritgordon chung proposed openstack/ceilometer: update architecture documentation
*** prnk28 has quit IRC15:32
*** AlexeyElagin has quit IRC15:33
*** AlexeyElagin has joined #openstack-ceilometer15:35
*** Ala has quit IRC15:46
gordcpradk: does the fact that we removed the ceilometer-central/compute-agent logic affect packaging?15:51
pradkgordc, ceilo packaging?15:51
gordcpradk: right.15:51
gordcto start services you need to define namespace and what not... i don't know if that's relevant from packaging pov.15:52
pradkgordc, if its just python code then it shouldnt, but it involves removing service scripts/config files then yes15:52
pradki think we already have the namespace thingy handled in kilo i thought when starting the polling agent15:53
gordcah, good to know.15:53
gordcwas going to ask if it helps to release an rc1 with the central/compute agent removed to help with packaging.15:54
gordcif you aren't using it already then we all good.15:54
pradkgordc, so did we remove /usr/bin/ceilometer-agent-central and compute? or its still there?15:55
gordcit should be effectively gone as of
pradkgordc, ok the polling agent pkging has support for namespaces already.15:58
gordccool cool15:59
pradkbut we still carry around the central/comput agent service files.. those will need to be removed15:59
*** ankita_wagh has joined #openstack-ceilometer16:03
cdentpradk: I was talking to someone in the rdo channel yesterday and I think they were interested in maintaining different service files16:03
*** safchain has quit IRC16:03
cdenthe (number80 maybe) was going to change the running command but not the service name16:04
pradkcdent, we already have separate service files .. you mean keep the service files and let each using polling service with the corresponding name space16:05
cdentI think that's what he was saying. I don't have a strong opinion, however keeping the service names the same makes upgrades (for people who have scripts and what not) easier16:06
*** devlaps has joined #openstack-ceilometer16:06
gordcyeah... i was asking that as well. because of openstack-manual16:06
gordcalthough i might just make change anyways and note that you can still use old way16:07
pradkyea makes sense from upgrade standpoint, so long as the docs say the same16:07
*** rohit_ has joined #openstack-ceilometer16:07
pradkgordc, that would be better, just note the old names are just aliases16:08
gordcpradk: kk16:09
gordcstill downloading openstack-manuals repo... blazing 65kb/s16:09
cdentHang on, I'm confused16:09
cdentare we talking about commands or service names?16:09
cdent(or both?)16:10
rohit_mriedem: Please review and approve when you get a chance, thanks16:10
gordccdent: i'm talking about this:
gordcwhat's the diff between command and service names?16:11
pradkcdent, central-agent service command is an alias to polling-agent --namespace central16:11
pradkso systemctl start central-agent will kick off the above command16:11
pradkits a service script from packaging pov16:12
cdentright, but that's different from what gordc appears to be talking about16:12
cdentthus my question16:12
gordccdent: i was asking whether it breaks packaging that we removed central/compute agent entrypoints... which apparently they don't use anyways.16:14
gordccdent: regarding service names, from what i understand we'll keep all three, ceilometer-polling, ceilometer-central, and ceilometer-compute (the latter two for backward compat)16:15
pradkentry point wont hurt packaging, but service scripts do which use /usr/bin/ceilo-central-agent command5 and i'm suggesting we just change that to use polling-agent --namespace16:16
pradkyea and keep the service names as is16:16
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Fix for AttributeError in discovery when keystone unavailable
cdentgordc: while you're around, have you heard anything more on that rally failure?16:26
gordccdent: nope... i've no idea what event they are querying on.16:26
gordci'm fairly confident it's because of restriction rather than performance but i'm going to test that out today16:27
*** ankita_wagh has quit IRC16:30
*** ankita_wagh has joined #openstack-ceilometer16:31
*** bmotz has quit IRC16:33
gordczigo: i noticed there's no ceilometer-polling package in debian. just to let you know, the central-agent and compute-agent commands are removed in liberty ^^16:41
*** IBerezovskiy has quit IRC16:41
zigogordc: There's a ceilometer-polling package.16:42
zigogordc: Ok, so I should remove the ceilometer-agent-compute and ceilometer-agent-central packages ?16:42
*** lsmola has quit IRC16:42
gordczigo: oh? i'm probably using the search wrong then.16:43
gordcgood to know16:43
gordczigo: speaking with pradk and cdent it's probably best to keep central/compute packages just for backward compatibility16:43
gordcbut they'll need to be changed to point to ceilometer-polling --namespace central16:44
zigogordc: Did you just kept a /usr/bin/ceilometer-polling that replaces the agents?16:45
gordczigo: yeah, basically a year ago we merge the agents into one... but we kept the original entrypoints around to maintain compability16:46
gordcwe removed taht here:
zigogordc: What's the point? Multiple entry point is ... hum ... comfortable !16:47
zigoThanks for letting me know anyway.16:47
gordcsimiplification. :)16:47
gordczigo: is this the right place to search for packges?
zigoJust this is enough:
gordczigo: ah ok, got it. original search was filtering too much16:49
gordczigo: good to know16:49
*** jasonamyers has joined #openstack-ceilometer16:53
zigogordc: So, I have to do "ceilometer-polling --namespace central" for the central agent, and what for the compute agent?16:53
*** jasonamyers has quit IRC16:53
gordcceilometer-polling --polling-namespaces compute16:55
gordcsorry wrong option name16:55
gordcand ceilometer-polilng --polling-namespaces ipmi for agent-ipmi16:56
*** changbl has quit IRC16:58
*** ankita_wagh has quit IRC17:03
*** exploreshaifali has joined #openstack-ceilometer17:05
*** ankita_wagh has joined #openstack-ceilometer17:10
*** ankita_wagh has quit IRC17:11
*** ankita_wagh has joined #openstack-ceilometer17:11
rohit_looking for reviews for
*** ankita_wagh has quit IRC17:20
*** leakypipes has quit IRC17:28
*** ankita_wagh has joined #openstack-ceilometer17:32
*** ankita_wagh has quit IRC17:32
*** ankita_wagh has joined #openstack-ceilometer17:33
*** cdent has quit IRC17:39
*** changbl has joined #openstack-ceilometer17:40
lintonvgordc: I still see no events through stacktack's webui17:45
lintonvshould the event_pipeline file have notifier:?topic=<rabbitmq queue>17:45
lintonvthat should push the events to rabbitmq?17:46
*** ankita_w_ has joined #openstack-ceilometer17:47
*** ankita_wagh has quit IRC17:51
*** ankita_w_ has quit IRC18:00
*** ankita_wagh has joined #openstack-ceilometer18:00
*** thumpba has joined #openstack-ceilometer18:08
gordclintonv: i'm not sure what queue stacktach is configured to consume from... if you're going to stacktach route, i'd probably suggest turning all of ceilometer off.18:18
gordcsorry i don't have much information on the latest stacktach. after they stopped integration efforts, a lot of the development went internal for a bit.18:19
lintonvso the events get pushed to rabbitmq, even without ceilometer18:19
gordclintonv: right. nova and the other services will emit notifications. ceilometer(or stacktach) will consume them and convert them into normalised events.18:19
gordcthe stacktach information is based on how it functioned a year ago. they might have changed their model (i'm not entirely sure)18:20
lintonvI see. Can aodh and ceilometer work together? Are they supposed to? I think they are one and maybe I should use aodh18:21
gordcyeah, they are designed to work together.18:22
gordcright now Aodh is only designed to notify by http request... but it's relatively simple to build other notification mechanism... you'd need to create a spec/bug for that those... or let me know what you're requirements are and we can look at adding support in.18:23
lintonvOk. I will pull down aodh and configure to listen to the rabbitmq queue. I am going to install aodh outside of an openstack cluster. I will leave ceilometer as it is on the cluster. Does that make sense?18:24
gordcthe only real requirements are that the aodh-listener service needs to have access to same message queue as ceilometer... of course you'll need to register aodh service and such.18:26
gordcwe're still working on that documentation right now.18:26
*** yassine__ has quit IRC18:30
*** lintonv_ has joined #openstack-ceilometer18:33
*** lintonv has quit IRC18:33
lintonv_gordc: I will open a bug/request with what I am trying to do here. Just saw that the community can comment or try to fix it. I will try to fix it as well, when I have narrowed down exactly what needs to be done18:34
*** nadya has quit IRC18:34
gordclintonv_: yep, anyone can propose and fix stuff and if you're volunteering all the better :)18:35
gordclintonv_: let us know if you have questions.18:35
rohit_gordc: quick question on, do we need to fix the client? i think the apis are all fixed now except event-list which returns empty list when -q limit=118:35
openstackLaunchpad bug 1494245 in python-ceilometerclient "--limit can collide with query string on some commands" [Medium,New] - Assigned to Rohit Jaiswal (rohit-jaiswal-3)18:35
lintonv_gordc: the main challenge is understanding the system better and seeing what it can do now. I can fix after that :)18:36
gordclintonv_: cool cool18:36
gordcrohit_: i'm actually not sure if we need a fix on client side (i haven't checked though)18:37
gordcrohit_: i'd think all the behaviour is dictated by the api.18:37
gordcrohit_: does it return empty list because it's querying for limit trait or because it doesn't know what to do with limit param?18:38
*** dyasny has joined #openstack-ceilometer18:41
dyasnygordc, ping18:41
rohit_gordc: not sure, let me check18:43
gordcdyasny: hey, whatsup?18:43
dyasnygordc, all good, you? How's the new place treating you?18:44
gordcdyasny: it's a place.lol18:44
*** ahilan has quit IRC18:44
gordci assume montreal is all well18:44
dyasnygordc, :) yeah, nice and french, as always18:45
dyasnygordc, vladik tells me part of your lasers-pew-pew thing was an ELK stack18:45
gordcdyasny: it's th E and K part18:47
gordcand a lot of lasers.18:47
gordcwe don't use logstash18:47
dyasnygordc, what did you use? fluentd?18:47
gordcwe have our own processing service... we had existing code and we don't really do anything complex so we skipped logstash18:49
gordci suppose you could always use logstash to actually capture system logs or what not18:49
rohit_gordc: its the former, it constructs a traitfilter with key as limit18:50
gordcrohit_: yeah, that's what i guessed... which means the empty list makes sense.18:51
gordcrohit_: i don't think we should try to assume they intended to set limit rather than query for limit trait.18:52
dyasnygordc, I need to gather logs, but I'm not sure I need the actual logs, and not some sort of ETL18:53
rohit_gordc: oh ok, then i will leave it as-is18:53
gordcrohit_: sounds good18:53
rohit_gordc: thanks, Btw, can you please review
dyasnygordc, the task is to get operational metrics, e.g. how many VMs got started between X and Y date-time, and how many started and how many failed18:53
dyasnygordc, avg VM lifetime maybe... you know, typical BI stuff18:54
*** ankita_w_ has joined #openstack-ceilometer18:54
*** ankita_wagh has quit IRC18:54
gordcdyasny: oh. yeah i don't think we have support for that yet. not from statistical pov. you'd have to do the maths outside ceilometer18:55
gordcyou might be able to calculate it via kibana (if you use elasticsearch)18:56
gordci never tried though.18:56
dyasnygordc, so what CAN ceilometer do? I always thought it was something like ETL+DWH for openstack18:57
gordcit's definitely not a data warehouse.18:59
gordcfrom events pov, all we have right now is the ability to capture events that happens on the services and normalise it. from there you can query on specific attributes.19:00
gordcwe don't support querying like you need. but again, if you using elasticsearch you can probably craft the appropriate queries in kibana19:01
*** changbl has quit IRC19:03
dyasnygordc, I see, thanks19:09
*** exploreshaifali has quit IRC19:10
*** ankita_w_ has quit IRC19:25
*** eglynn has quit IRC19:30
rohit_gordc:, does it need something from oslo_messaging?19:39
openstackLaunchpad bug 1484695 in Ceilometer "pollster/api publish to info queue" [Medium,Triaged] - Assigned to Rohit Jaiswal (rohit-jaiswal-3)19:39
gordcrohit_: nope... or not necessarily.19:43
gordcrohit_: you should be able to add a sample endpoint to notificatino agent similar the info endpoint it already has19:44
rohit_gordc: oslo_messaging support publishing at sample level"
rohit_yep thats what i commented in the bug19:44
gordcrohit_: yeah, i believe jd's comment was more granular/flexible control19:45
rohit_gordc: ok, cool19:45
gordcrohit_: it has a fairly high chance of overlapping with liusheng patch's
gordcactually... now taht i think about it. i'm not sure what topic the polling/api publishes to19:48
gordcif it's the notifications.sample topic it might actually overlap with the queue taht notification agent -> collector publishes19:48
rohit_gordc: i thought thats the metering.sample topic19:50
rohit_for the collector19:50
gordcrohit_: ah yeah you're right19:50
gordci guess we just need to make sure they don't overlap in anyway19:50
rohit_i think liushengs patch is to disable telemetry samples to be consumed as events, we will be changing the topic it gets published with so that it can still be consumed as samples by noti agent19:52
gordcrohit_: right. but the notification agent doesn't listen to *.sample topic when building events so it will effectively do the same thing.19:53
gordcso yeah, i'm pretty sure it makes his option useless... so we either need to drop his patch or you need to add support in event endpoint to listen to sample topic (if you were to code this).19:56
rohit_yes, got it, so just changing the topic where telemetry.* events get published will make his patch redundant19:56
gordctbh, i'd imagine we'd be using the *.sample topic for events eventually.19:56
gordcie. we can create health check events from polling.19:57
gordcbut that's all hypothetical.19:57
rohit_ok, that sounds like a good use case19:57
rohit_gordc: thanks,  i will keep Luisheng's patch in minid19:58
gordci have a slight preference to do the topic change now and maybe delay liusheng's patch.19:58
rohit_gordc: yeah, i shud have it ready by early next week19:59
gordcrohit_: ok. i'm going to hold off on liusheng's patch for now... we can discuss which one we want later (or both)19:59
rohit_or we can use gerrit Depends-on19:59
gordcyeah. i'm thinking one only but we can mark depends on later.20:01
rohit_ok, sounds good20:02
*** lintonv_ has quit IRC20:11
*** dyasny has quit IRC20:19
*** ankita_wagh has joined #openstack-ceilometer20:24
*** tiny-hands has quit IRC21:03
*** thorst has quit IRC21:07
*** thumpba has quit IRC21:07
*** pradk has quit IRC21:10
*** pradk has joined #openstack-ceilometer21:10
*** thorst has joined #openstack-ceilometer21:41
*** gordc has quit IRC21:44
*** shardy has quit IRC21:47
*** edmondsw has quit IRC22:05
*** rohit_ has quit IRC22:20
-openstackstatus- NOTICE: 30 minute warning, Gerrit will be offline from 23:00 to 23:30 UTC while some projects are renamed
-openstackstatus- NOTICE: Gerrit is offline from 23:00 to 23:30 UTC while some projects are renamed.
*** ChanServ changes topic to "Gerrit is offline from 23:00 to 23:30 UTC while some projects are renamed."23:01
*** rex_lee has quit IRC23:03
*** ondergetekende has quit IRC23:04
*** ondergetekende has joined #openstack-ceilometer23:05
*** rex_lee has joined #openstack-ceilometer23:09
*** tongli has quit IRC23:14
*** ChanServ changes topic to "OpenStack Telemetry (Ceilometer) |"23:38
*** openstackgerrit has quit IRC23:46
*** openstackgerrit has joined #openstack-ceilometer23:46
*** ankita_wagh has quit IRC23:49
*** kbyrne has quit IRC23:58

Generated by 2.14.0 by Marius Gedminas - find it at!