Wednesday, 2016-03-02

« Tuesday, 2016-03-01 Index Thursday, 2016-03-03 »
*** thorst_ has joined #openstack-telemetry00:30
*** Qiming has quit IRC00:34
*** thorst_ has quit IRC00:38
*** thorst_ has joined #openstack-telemetry00:38
*** thorst_ has quit IRC00:40
*** thorst_ has joined #openstack-telemetry00:41
*** Qiming has joined #openstack-telemetry01:23
*** cheneydc has joined #openstack-telemetry01:38
*** jaypipes has quit IRC01:54
*** thorst_ has quit IRC01:56
*** thorst_ has joined #openstack-telemetry01:57
*** thorst_ has quit IRC02:05
openstackgerritZi Lian Ji proposed openstack/python-aodhclient: Fix aodh client fails when command with the arg --time-constraint  https://review.openstack.org/27294602:50
*** achatterjee has joined #openstack-telemetry02:51
*** liamji has joined #openstack-telemetry03:02
*** thorst_ has joined #openstack-telemetry03:04
*** thorst_ has quit IRC03:10
*** david-lyle has joined #openstack-telemetry03:24
*** larainema has quit IRC03:27
*** larainema has joined #openstack-telemetry03:29
*** Ephur has quit IRC03:33
openstackgerritMerged openstack/python-aodhclient: add alarm history unit test  https://review.openstack.org/27339703:46
*** thorst_ has joined #openstack-telemetry04:08
openstackgerritgordon chung proposed openstack/gnocchi: remove unnecessary sort_index  https://review.openstack.org/28697504:11
*** thorst_ has quit IRC04:16
*** links has joined #openstack-telemetry04:51
*** yprokule has joined #openstack-telemetry04:57
*** thorst_ has joined #openstack-telemetry05:14
*** thorst_ has quit IRC05:21
*** pcaruana has joined #openstack-telemetry05:36
*** pcaruana has quit IRC05:49
*** rcernin has joined #openstack-telemetry05:51
*** dave-mccowan has quit IRC05:51
*** thorst_ has joined #openstack-telemetry06:19
*** thorst_ has quit IRC06:26
*** shardy has joined #openstack-telemetry07:00
*** thorst_ has joined #openstack-telemetry07:23
*** thorst_ has quit IRC07:31
*** vishwanathj is now known as vishwanathj_zzz07:38
*** belmoreira has joined #openstack-telemetry07:38
openstackgerritZi Lian Ji proposed openstack/ceilometer: Add the meter example file 'lbaas-v2-meter-definitions.yaml'  https://review.openstack.org/28643107:59
*** pcaruana has joined #openstack-telemetry08:02
*** fwdit has joined #openstack-telemetry08:08
*** thorst_ has joined #openstack-telemetry08:29
openstackgerritliusheng proposed openstack/ceilometer: Caching info in the instance discovery  https://review.openstack.org/26707808:31
*** thorst_ has quit IRC08:35
openstackgerritliusheng proposed openstack/ceilometer: Caching info in the instance discovery  https://review.openstack.org/26707808:40
*** permalac has joined #openstack-telemetry08:42
*** permalac has quit IRC08:42
openstackgerritliusheng proposed openstack/ceilometer: Caching info in the instance discovery  https://review.openstack.org/26707808:59
*** liamji has quit IRC09:00
*** permalac has joined #openstack-telemetry09:00
*** vishwanathj_zzz has quit IRC09:03
permalachello guys, I'm with kilo and about to set up ceilometer in HA, as I understand that only the central agent can be in HA does this mean the other agents must be stoped on all the non working nodes?09:11
*** thorst_ has joined #openstack-telemetry09:33
*** thorst_ has quit IRC09:40
*** Qiming has quit IRC09:45
*** sheeprine has quit IRC09:49
*** sheeprine has joined #openstack-telemetry09:50
*** _nadya_ has joined #openstack-telemetry09:58
*** cheneydc has quit IRC09:59
*** yassine__ has joined #openstack-telemetry10:28
_nadya_liusheng: hi!10:28
_nadya_liusheng: I see that you're doing some changes in metadata cache patch. Did you see that one https://review.openstack.org/#/c/284322/ ?10:30
*** ityaptin_laptop has joined #openstack-telemetry10:30
*** ityaptin_ has quit IRC10:33
*** dikonoor has joined #openstack-telemetry10:36
*** thorst_ has joined #openstack-telemetry10:38
*** dikonoor has quit IRC10:40
*** _nadya__ has joined #openstack-telemetry10:43
*** _nadya_ has quit IRC10:43
*** thorst_ has quit IRC10:45
*** Qiming has joined #openstack-telemetry10:49
*** cdent has joined #openstack-telemetry10:59
*** ityaptin_laptop has quit IRC11:03
*** ityaptin_laptop has joined #openstack-telemetry11:07
*** _nadya__ has quit IRC11:10
*** _nadya_ has joined #openstack-telemetry11:40
*** thorst_ has joined #openstack-telemetry11:43
*** _nadya_ has quit IRC11:45
*** thorst_ has quit IRC11:51
*** dave-mccowan has joined #openstack-telemetry11:53
*** _nadya_ has joined #openstack-telemetry11:55
*** kudemi has joined #openstack-telemetry11:57
*** cdent has quit IRC12:06
liusheng_nadya_: Hi, yes, I saw it, but I don' think that change can perform a big improvement, right ?12:22
liusheng_nadya_: thanks for reminding me :)12:22
*** Liuqing has joined #openstack-telemetry12:27
_nadya_liusheng: we can update the cache once in 10 minutes, but poll once in 5 seconds. It's not actually "polling improvement", I agree12:28
_nadya_liusheng: the problem we faced with that if there are, let's say, 100000 instances, nova will return this list using pagination12:29
_nadya_liusheng: thus, we have one more thing to handle12:30
liusheng_nadya_: maybe the instances in cache is out-of-dated for us if we update it per 10min12:31
liusheng_nadya_: you mean in the global cache approach ?12:31
_nadya_liusheng: yep, they can. To have everything up-to-date, we need to do updates from notification agent12:31
_nadya_liusheng: yep, global cache approach, right12:32
_nadya_liusheng: looks like we need a session at the summit about this. too many questions still12:32
liusheng_nadya_: we have inputted 'change-since' parameter, it unlikly to meet the nova API limitation for listing instances12:33
liusheng_nadya_: not sure :)12:33
_nadya_liusheng: actually, we should not get all instances at all. If we start Ceilometer from the very beginning, we will always use "change-since". We just cannot launch 10000+ instances between cache update (10 minutes?). But if we lose the cache (how can we lose it if we have the cache in HA?), we need to populate it12:37
_nadya_liusheng: very corner-case problem12:38
liusheng_nadya_: yes, I agree, even if we need to query 10000+ instances, can we address that case by code logic ?12:40
_nadya_liusheng: theoretically, yes. nova-experts have some doubts about performance. What is better: run several get_instance_by_host or one big query with pagination. We're going to test it at scale12:41
liushengliusheng: I am afraid that will make process more complex :(, what do you mean "several get_instance_by_host"12:44
liusheng_nadya_: maybe we can give a maximum value (< 10000) to query instances from Nova API, if the return of Nova API is more than 10000, we query by serveral pagination query12:47
*** thorst_ has joined #openstack-telemetry12:48
openstackgerritMichael Krotscheck proposed openstack/ceilometer: Moved CORS middleware configuration into oslo-config-generator  https://review.openstack.org/28678112:48
_nadya_liusheng: so, we want improve performance: instead of several "get_all_by host" (get_instance_by_host is incorrect name) from several computes we want ask get_all from one compute. But if get_all reurns 100000+ computes, performance may not be improved actually12:48
_nadya_liusheng: yep, probably, the only thing we may do is to support pagination12:50
_nadya_liusheng: *we should do12:50
liusheng_nadya_: if we have more than 100000 instances, the compote nodes will be more thant "several" :)12:53
_nadya_liusheng: yep, definitely :)12:54
liusheng_nadya_: I think it is easy to support nova API pagination query in ceilometer12:54
_nadya_liusheng: we need Gordon and ityaptin_laptop involved in this discussion once more. May do it today or tomorrow in irc meeting. Because, it seems that I don't see the actual blocker here.. "support pagination" doesn't seem a blocker12:56
*** Qiming has quit IRC12:59
*** links has quit IRC12:59
liusheng_nadya_: cool, at least, the global cache is just a optional way, we can support local cache as before12:59
*** _nadya_ has quit IRC13:01
*** nicodemus_ has joined #openstack-telemetry13:01
openstackgerritgordon chung proposed openstack/gnocchi: remove unnecessary sort_index  https://review.openstack.org/28697513:21
*** vgridnev has joined #openstack-telemetry13:22
*** vgridnev has quit IRC13:23
*** pradk has quit IRC13:32
*** Ephur has joined #openstack-telemetry13:47
*** dikonoor has joined #openstack-telemetry13:48
*** _nadya_ has joined #openstack-telemetry13:51
*** dikonoor has quit IRC13:51
*** fwdit has quit IRC13:56
*** liamji has joined #openstack-telemetry14:01
*** bapalm has quit IRC14:05
*** jwcroppe_ is now known as jwcroppe14:05
*** bapalm has joined #openstack-telemetry14:06
*** dikonoor has joined #openstack-telemetry14:08
*** yprokule has quit IRC14:08
*** dikonoo has joined #openstack-telemetry14:09
*** dikonoo has quit IRC14:09
*** dikonoor has quit IRC14:09
*** dikonoor has joined #openstack-telemetry14:10
*** dikonoo has joined #openstack-telemetry14:10
*** dikonoor has quit IRC14:10
*** dikonoo has quit IRC14:10
*** yprokule has joined #openstack-telemetry14:13
*** thumpba has joined #openstack-telemetry14:16
nicodemus_hello gordc14:17
nicodemus_oh, he's not around.14:19
*** ildikov has quit IRC14:23
*** twm2016 has joined #openstack-telemetry14:28
*** nicodemus_ has quit IRC14:31
*** nicodemus_ has joined #openstack-telemetry14:31
*** ljxiash has joined #openstack-telemetry14:33
*** ljxiash has quit IRC14:36
*** ljxiash has joined #openstack-telemetry14:36
*** Liuqing has quit IRC14:44
*** alejandrito has joined #openstack-telemetry14:49
*** rbak has joined #openstack-telemetry14:55
*** pradk has joined #openstack-telemetry14:55
openstackgerritIgor Degtiarov proposed openstack/ceilometer: timedelta plugin for meter definition process  https://review.openstack.org/28646914:56
*** ildikov has joined #openstack-telemetry14:56
*** ljxiash has quit IRC14:57
*** ljxiash has joined #openstack-telemetry15:03
twm2016Hi everyone, I am working on this bug https://bugs.launchpad.net/ceilometer/+bug/1528243 and I am in the middle of adding a record_events function in the gnocchi dispatcher. I was wondering how one would acquire the parameters for calling the _update_resouce function for an event instead of a metric.15:10
openstackLaunchpad bug 1528243 in Ceilometer "gnocchi dispatcher should leverage events" [Wishlist,Triaged] - Assigned to Trevor McCasland (twm2016)15:10
*** rickyrem has joined #openstack-telemetry15:15
*** jaypipes has joined #openstack-telemetry15:17
alejandritodoes anyone knows if gnocchi supports keystone domains ?15:20
*** kudemi has quit IRC15:25
*** vishwanathj_zzz has joined #openstack-telemetry15:26
*** vishwanathj_zzz is now known as vishwanathj15:27
*** vishwana_ has joined #openstack-telemetry15:29
*** yprokule has quit IRC15:30
*** gordc has joined #openstack-telemetry15:31
*** vishwana_ is now known as vishwanathj_15:31
*** vishwanathj has quit IRC15:31
*** yprokule has joined #openstack-telemetry15:42
*** liamji has quit IRC15:42
*** yprokule has quit IRC15:43
*** KrishR has joined #openstack-telemetry15:44
*** albertom has left #openstack-telemetry15:45
*** belmoreira has quit IRC15:50
*** krypto has joined #openstack-telemetry15:54
kryptostrange issue with ceilometer , "ceilometer alarm-list" shows gateway timeout ;api process is running;restarting process fixes this issue .Any idea?15:57
kryptonothing in logs ;have a haproxy frontend which is under utilised15:58
gordckrypto: nothing in logs == doesn't even know the api request happened??16:02
kryptowill "ceilometer alarm-list " logs anything in api.log?16:04
kryptogordc16:04
*** pcaruana has quit IRC16:05
gordcit should, yes.16:05
gordclet me try finding an example16:05
gordckrypto: something like http://logs.openstack.org/60/262660/6/check/gate-telemetry-dsvm-integration-ceilometer/708e8a7/logs/apache/aodh_access.txt.gz16:06
*** ljxiash has quit IRC16:07
nicodemus_hello gordc, quick question: does gnocchi support keystone domains? Or is that an extra metadata we should add?16:08
*** jaypipes has quit IRC16:08
*** _nadya_ has quit IRC16:09
*** BigWillie has joined #openstack-telemetry16:11
gordcnicodemus_: i'm not sure we store that information. (i assume you're asking about storing the data vs authentication)16:12
*** ityaptin_laptop has quit IRC16:14
*** alextricity25_ has joined #openstack-telemetry16:14
*** smurke_ has quit IRC16:15
*** sudorandom has quit IRC16:15
*** alextricity25 has quit IRC16:15
*** thorst_ has quit IRC16:15
nicodemus_I was wondering if I enable domains in keystone, if a resource would *magically* have also domain id (meaning, the ceilometer agents would collect and push domain id as well)16:15
*** sudorandom has joined #openstack-telemetry16:15
*** smurke has joined #openstack-telemetry16:15
nicodemus_or if that would be something needed to add to the gnocchi_resources.yaml as additional metadata I want to have on my resources16:16
*** thorst_ has joined #openstack-telemetry16:16
kryptogordc sorry for the wrong information it logs: "Max retries exceeded with url: /v2/alarms/26d567a1-ea26-40b2-a0a7-4a592085039b/state (Caused by <class 'httplib.BadStatusLine'>: '')16:16
krypto2016-03-02 16:11:36.273 23501 TRACE ceilometer.alarm.evaluator16:16
krypto"16:16
kryptohttp://paste.ubuntu.com/15267787/16:17
gordcnicodemus_: you'd have to map it. it's pretty rigid right now in terms of adding new attrs/resource types.16:19
gordci think you'll need to wait for 2.1.x which will have sileht's patch to make things more dynamic16:20
*** yassine__ has quit IRC16:20
gordchttps://review.openstack.org/#/q/status:open+project:openstack/gnocchi+branch:master+topic:bp/resource-type-rest-api16:20
nicodemus_gordc: good to know. Thanks!16:20
*** jaypipes has joined #openstack-telemetry16:22
gordckrypto: are you using aodh or ceilometer(liberty)?16:24
gordchm i guess ceilometer.16:24
kryptogordc ceilometer16:25
gordci don't know. it seems like the server is rejecting calls from evaluator... so it's probably blocking your client reqs too.16:25
gordckrypto: what version of client are you using?16:26
kryptogordc ceilometer --version16:27
krypto1.0.1216:27
kryptocan this be an issue with HAProxy itself?16:27
gordcpossibly? i've never seen that error to be honest16:28
gordckrypto: are you on juno?16:28
kryptoyes16:29
gordcoh... yeah, i was wondering why your client was so old.16:31
gordchmm.. well you do need some form of web server so it will need to work with haproxy16:32
gordci'm not sure. i'd need to google. but i have a few other items to track currently.16:32
*** datravis has joined #openstack-telemetry16:33
kryptoother services are fine with haproxy but ceilometer sometimes go down;if i restart ceilometer it works again [ahh thats a hint so issue must be with ceilometer api process itself]16:33
*** datravis has left #openstack-telemetry16:34
*** eglynn has joined #openstack-telemetry16:34
*** david-lyle has quit IRC16:36
*** jwcroppe has quit IRC16:37
gordckrypto: possibly. it does seem that some rate limiting is kicking in and it just resets when you restart api which is why it works at start and then stops after a while.16:38
*** david-lyle has joined #openstack-telemetry16:38
*** ityaptin_laptop has joined #openstack-telemetry16:38
kryptogordc: good point thanks i will check for that16:40
alejandritogordc, one big issue that we are having with gnocchi 2.0  (the one nicodemus_ hacked with the now timestamp ) is a critical one that all different aggregates show the sames values16:44
*** BigWillie has quit IRC16:45
*** ildikov has quit IRC16:46
gordcalejandrito: i'm installing ubuntu on a vm now to test16:48
gordcalejandrito: have you tried with file backend? (trying to pinpoint whether it's a ceph thing, ubuntu thing, config thing, or something else)16:48
alejandritogordc, i can test it ... but want to know in what environment you are testing and works16:50
alejandritogordc, OS, backend for example16:50
gordcfedora, file|swift.16:50
alejandritogordc, we can try file and nicodemus_ will let you know16:51
gordcthe gate uses ubuntu and file16:51
gordci'm not sure what sileht uses. debian and ceph probably.16:51
alejandritooka, switching from ubuntu/ceph to ubuntu/file on gnocchi 2.016:52
nicodemus_gordc: I'm switching to file and check the aggregates with our hack: http://paste.openstack.org/show/489008/16:52
*** belmoreira has joined #openstack-telemetry16:53
gordcnicodemus_: are you asking me to validate the hack is valid?16:54
silehtgordc, for Gnocchi/Ceph testing ubuntu trusty and ceph packages from ceph repo16:54
gordcit will probably mess up all your data if you just reset the time16:54
*** jwcroppe has joined #openstack-telemetry16:55
nicodemus_gordc: no, we did that just for metricd on 2.0 to start processing metrics16:55
nicodemus_if that is causing all aggregates to have the same value, it shouldn't change with file16:55
silehtgordc, should we release a 2.0.1 with your upgrade fix, or did we wait some other fixes ?16:56
*** krypto has quit IRC16:56
gordcsileht: *gasp* you should be careful... red hat won't be happy ;)16:56
*** twm2016 has quit IRC16:56
silehtgordc, I have ton of VM with different OS16:56
gordcsileht: good answer :P16:56
gordcsileht: shall we merge https://review.openstack.org/#/c/286749/16:57
silehtgordc, of course , I have done it16:57
gordci have https://review.openstack.org/#/c/286246/ but it doesn't really affect anything... just throws errors when you stop service16:57
gordcwe can release 2.0.1. we can always release again16:59
*** _nadya_ has joined #openstack-telemetry17:00
gordcnicodemus_: i see. does it actually catch exception? i thought when i told you to add 'try/except Exception' it didn't catch error?17:02
nicodemus_gordc: nope, it doesn't17:03
gordcnicodemus_: :(... i wonder what that value is that's cause mayhem.17:04
gordcnothing in your logs seems strange: https://gist.github.com/nvlan/5e3e7f184fde2480508c17:04
*** ildikov has joined #openstack-telemetry17:12
*** krypto has joined #openstack-telemetry17:18
*** krypto has quit IRC17:22
*** krypto has joined #openstack-telemetry17:22
*** jaypipes has quit IRC17:22
*** liusheng has quit IRC17:25
*** liusheng has joined #openstack-telemetry17:27
*** cdent has joined #openstack-telemetry17:30
*** rbak has quit IRC17:31
*** rbak has joined #openstack-telemetry17:33
*** boris-42 has joined #openstack-telemetry17:35
*** jaypipes has joined #openstack-telemetry17:36
*** rickyrem has quit IRC17:37
gordcnicodemus_: what's the archive policy that you use?17:40
nicodemus_gordc: http://paste.openstack.org/show/489015/17:41
gordcok. let me try that too17:43
*** jaypipes has quit IRC17:47
*** _nadya_ has quit IRC17:53
*** rbak_ has joined #openstack-telemetry17:57
*** rbak has quit IRC17:59
*** KrishR has quit IRC17:59
*** belmoreira has quit IRC18:03
*** safchain has joined #openstack-telemetry18:05
*** shardy has quit IRC18:06
*** cdent has left #openstack-telemetry18:06
*** rbak__ has joined #openstack-telemetry18:07
*** slaweq has joined #openstack-telemetry18:07
slaweqhello18:07
slaweqI have question about ceilometer "backend"18:07
slaweqI want to store metrics data in some opentsdb database (I will push such data with my own mechanism to db)18:08
slaweqand I want to use ceilometer API to access to this data18:08
slaweqis it possible somehow?18:08
slaweqmaybe there is some kind of mechanism to write own "plugin" or some "backend driver" to ceilometer?18:09
slaweqif yes, can You maybe give me some links to documentation about that?18:09
slaweqthx18:09
*** rbak_ has quit IRC18:09
gordcslaweq: i'm going to lunch but quick answer is it depends on how opentsdb stores data and how much work you want to put it to map that model to ceilometer-api18:11
gordcyou'll need to extend https://github.com/openstack/ceilometer/blob/master/ceilometer/storage/base.py18:12
*** david-lyle has quit IRC18:12
gordci probably wouldn't waste my time on it unless you have a very good reason for using ceilomter-api models.18:12
slaweqgordc: thx18:12
gordcit's extremely verbose so doesn't really lend itself to timeserie models18:12
gordcbbl18:12
*** david-lyle has joined #openstack-telemetry18:13
slaweqthx for link I will take a look on it, I have data about my cluster stored in such db and I want to use watcher18:13
slaweqwatcher is asking ceilometer about data18:13
slaweqso I need ceilometer API as some kind of "middleware" there :)18:13
slaweqthx a lot for help18:13
slaweqand bon appetit :)18:14
stevelleslaweq: gnocchi is designed for this storage, influxdb driver might be related to your effort18:15
stevelleif there is another tsdb that you would like to use, a gnocchi driver might be easier18:16
slaweqstevelle: ok, thx I will take a look on gnocchi also18:16
slaweqbut isn't gnocchi something what push metrics from ceilometer collector to opentsdb?18:16
stevelleceilometer publishes metrics to gnocchi, gnocchi provides an API for read/write of metrics backed by various storage engines18:17
stevellewatcher can query gnocchi more efficiently18:18
*** jwcroppe has quit IRC18:19
slaweqok, thx a lot for tips18:20
slaweqI will check it18:20
*** safchain has quit IRC18:25
*** alejandrito has quit IRC18:43
openstackgerritMerged openstack/ceilometermiddleware: Updated from global requirements  https://review.openstack.org/28501518:43
nicodemus_gordc: I just finished testing gnocchi 2.0 with file backend and our datetime.now() hack for metricd to be able to process measures18:44
nicodemus_the metrics I queried return the same values for several aggregation methods18:45
gordcnicodemus_: i don't understand, i thought you said your hack doesn't catch anything.18:52
gordcslaweq: what stevelle said :)18:53
nicodemus_gordc: my bad, the "print" statement didn't print anything18:55
nicodemus_http://paste.openstack.org/show/489028/18:57
nicodemus_the try-catch is catching the exception18:58
gordcnicodemus_: you can actually catch it? i was under impression it didnt' work18:58
gordcnicodemus_: can you see what the value is? that is throwing error..18:58
gordcyour hack is more than likely just corrupting the data so i'd stop doing that.18:59
*** thorst_ has quit IRC19:00
*** thorst has joined #openstack-telemetry19:01
*** thorst has quit IRC19:01
*** thorst has joined #openstack-telemetry19:01
nicodemus_gordc: http://paste.openstack.org/show/489029/ is that date correct?19:02
gordc... nope. is one of your system's time messed up?19:05
gordci don't think it should throw an error though. http://pandas.pydata.org/pandas-docs/stable/gotchas.html#minimum-and-maximum-timestamps19:05
gordcnicodemus_: are you printing 'ts'? https://github.com/openstack/gnocchi/blob/master/gnocchi/carbonara.py#L14319:06
*** rickyrem has joined #openstack-telemetry19:11
gordchow come you have policies to store 5, 15, 30 and 60 years worth of data?lol19:17
*** KrishR has joined #openstack-telemetry19:22
nicodemus_gordc: yeah... those timespans were left there from the first archive policy we started using19:23
gordckk. building with those rules now. maybe it's an edge case that is rounding the dates to some crazy old/new number.19:24
nicodemus_strangely, with this code: http://paste.openstack.org/show/489037/ I only see "OVERFLOW ERROR" with a 1956 date19:25
nicodemus_I'll try again just to be sure19:26
gordccan you check the type as well in line 719:26
gordcand i guess the value of 'freq'19:27
gordci can see it.19:38
gordcit has something to do with the policies you have19:38
nicodemus_gordc: http://paste.openstack.org/show/489042/19:42
nicodemus_every time it's that 1956 date19:42
nicodemus_hmmm... maybe I shouldn't try to store 60 years of data19:43
gordcyeah, it's the 4.478976e+20 freq value19:46
gordcwhen it tries to grab it's bucket of data, it's rounding back to some ridiculously old time.19:46
nicodemus_should I try dropping every definition above 1 year for example?19:47
gordcsure... 5 years might be ok (re: no overflow)19:49
*** jwcroppe has joined #openstack-telemetry19:49
*** jwcroppe has quit IRC19:50
gordci don't know why you have 1956 tsthough. maybe it's a rounded value...19:50
gordci don't know why you have 1956 ts though. maybe it's a rounded value...19:50
*** jwcroppe has joined #openstack-telemetry19:50
*** belmoreira has joined #openstack-telemetry19:50
*** belmoreira has quit IRC19:51
*** tongli has joined #openstack-telemetry19:53
nicodemus_gordc: great. I'll give it a try andlet you know19:55
*** thorst has quit IRC20:01
*** thorst has joined #openstack-telemetry20:01
*** thorst has quit IRC20:05
*** thorst has joined #openstack-telemetry20:08
nicodemus_gordc: it appears to be working with this policy: http://paste.openstack.org/show/489043/ , I'd like to test it again with clean data20:15
gordcnicodemus_: kk, i'm not sure exact fix but i lowered the priority since my laziness says i don't want to worry about  60 year data scenario just yet :)20:17
gordcalthough the 1956 value does make a bit more sense now that i think about it.20:18
*** permalac has quit IRC20:30
*** krypto has quit IRC20:34
*** krypto has joined #openstack-telemetry20:34
*** rcernin has quit IRC20:36
*** rickyrem has quit IRC20:44
*** thumpba has quit IRC20:44
*** david-lyle has quit IRC20:45
*** david-lyle has joined #openstack-telemetry20:47
*** rickyrem has joined #openstack-telemetry20:57
*** krypto has quit IRC21:00
*** krypto has joined #openstack-telemetry21:01
*** julim has quit IRC21:11
*** Ephur has quit IRC21:24
openstackgerritVitaly Gridnev proposed openstack/ceilometer: [sahara] add events definitions regarding new notifications  https://review.openstack.org/28122621:39
nicodemus_gordc: how many measures should I have for the different aggregate methods to return different values?21:46
*** r-mibu has quit IRC21:46
*** r-mibu has joined #openstack-telemetry21:47
*** eglynn has quit IRC21:49
gordcdepends on which metric and aggregate method... and frequency ceilometer is gathering data.21:50
gordcprobably easiest way to see something is to just make the polling intervals more frequent21:51
nicodemus_if using gnocchi client I execute "gnocchi measures show <UUID> --granularity 300 --aggregation X", being UUID a metric that has 10 different values, and X different aggregation methods21:52
nicodemus_the values displayed should change?21:52
nicodemus_I didn't try issuing a GET directly to the gnocchi API yet21:53
nicodemus_as per the docs21:53
nicodemus_for instance: http://paste.openstack.org/show/489059/21:56
*** thorst has quit IRC22:01
*** thorst has joined #openstack-telemetry22:02
gordcnicodemus_: hm... yes, i imagine the value should change. what happens if you use aggregation 'sum' or 'count'?22:05
gordcerr sorry. ignore sum, just saw you have it22:05
nicodemus_same values for count as well22:05
gordcweird. i'll take a look a look at tests22:06
gordcfor client22:06
nicodemus_I see that there are a lot of objects in the ceph pool now for each metric, when with 1.3 we had about 1022:06
*** thorst has quit IRC22:06
nicodemus_now there are gnocchi_UUID_number_aggregationmethod_timespan.022:08
gordcnicodemus_: yeah, before we stored all the datapoints for all the aggregates under one file. so if we wanted to update one, we had to pull in this giant blob22:08
nicodemus_gordc: that makes sense22:08
gordcnow it's all split up... and in smaller chunks so we're sending/handling less data22:09
nicodemus_would you like me to submit a bug report?22:09
gordcnicodemus_: sure, let's target against client for now. https://bugs.launchpad.net/python-gnocchiclient/+bugs?orderby=-id&start=022:10
gordcmight want to include client version22:10
nicodemus_gordc: will do. Be advised, if I curl directly to gnocchi-api the same unchanged values are returned22:12
nicodemus_issuing a GET to v1/metric/fc8b7686-dae6-46a8-863d-fb89f79365c1/measures?aggregation=min22:12
gordcnicodemus_: put it against gnocchi. don't really have time to look at it right now  (sorry).22:13
gordci would include archive policy and frequency of pollin in ceiloemter22:13
nicodemus_gordc: will do, np22:13
nicodemus_gordc: thanks!22:14
gordcand the name of metric22:14
*** thorst has joined #openstack-telemetry22:23
tongli@gordc, gordon, are you around?22:26
*** krypto has quit IRC22:27
*** thorst has quit IRC22:27
tongliinstalled Ceilometer Liberty. For some reason, notification agent won't place the meters on the metering queue.22:29
tongliCan someone tell me what configuration option is needed?22:29
*** nicodemus_ has quit IRC22:34
openstackgerritMerged openstack/ceilometer: Remove unused pngmath Sphinx extension  https://review.openstack.org/28620222:49
*** yarkot_ has joined #openstack-telemetry22:53
*** yarkot_ has quit IRC22:57
*** yarkot_ has joined #openstack-telemetry22:57
*** yarkot_ has quit IRC22:59
*** Qiming has joined #openstack-telemetry23:05
*** gordc has quit IRC23:09
*** tongli has quit IRC23:14
*** rickyrem has quit IRC23:28
*** ljxiash has joined #openstack-telemetry23:33
*** KrishR has quit IRC23:51
*** pradk has quit IRC23:53
« Tuesday, 2016-03-01 Index Thursday, 2016-03-03 »

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