| *** yassine has quit IRC | 00:02 | |
| *** jmckind has quit IRC | 00:15 | |
| *** dperaza has quit IRC | 00:17 | |
| *** ryanpetrello has quit IRC | 00:33 | |
| *** _ruhe is now known as ruhe | 00:43 | |
| *** ryanpetrello has joined #openstack-ceilometer | 00:45 | |
| *** gordc has quit IRC | 00:46 | |
| *** ryanpetrello has quit IRC | 01:00 | |
| *** xianghui has joined #openstack-ceilometer | 01:07 | |
| *** ruhe is now known as _ruhe | 01:48 | |
| openstackgerrit | litong01 proposed a change to openstack/python-ceilometerclient: add notification alarm https://review.openstack.org/69474 | 01:51 |
|---|---|---|
| *** ryanpetrello has joined #openstack-ceilometer | 01:55 | |
| openstackgerrit | litong01 proposed a change to openstack/ceilometer: add notification alarm https://review.openstack.org/69473 | 02:14 |
| openstackgerrit | litong01 proposed a change to openstack/ceilometer: add notification alarm https://review.openstack.org/69473 | 02:18 |
| *** xianghui has quit IRC | 02:19 | |
| *** xmltok has quit IRC | 02:22 | |
| *** flwang has joined #openstack-ceilometer | 02:27 | |
| *** _ruhe is now known as ruhe | 02:28 | |
| *** ryanpetrello has quit IRC | 02:31 | |
| *** xianghui has joined #openstack-ceilometer | 02:32 | |
| *** ryanpetrello has joined #openstack-ceilometer | 02:32 | |
| *** gordc has joined #openstack-ceilometer | 02:39 | |
| *** ryanpetrello has quit IRC | 02:54 | |
| *** gordc has quit IRC | 03:23 | |
| *** sayali has quit IRC | 03:25 | |
| *** ok_delta has joined #openstack-ceilometer | 03:50 | |
| *** ok_delta__ has joined #openstack-ceilometer | 03:50 | |
| openstackgerrit | Yuuichi Fujioka proposed a change to openstack/ceilometer: Implements monitoring-network https://review.openstack.org/60473 | 04:22 |
| *** dperaza has joined #openstack-ceilometer | 04:53 | |
| *** sayali has joined #openstack-ceilometer | 05:05 | |
| *** sayali_ has joined #openstack-ceilometer | 05:05 | |
| *** sayali_ has quit IRC | 05:06 | |
| *** ryanpetrello has joined #openstack-ceilometer | 05:12 | |
| *** dperaza has quit IRC | 05:16 | |
| *** ok_delta__ has quit IRC | 05:18 | |
| *** ok_delta has quit IRC | 05:18 | |
| *** SergeyLukjanov_ is now known as SergeyLukjanov | 05:28 | |
| *** krast has quit IRC | 05:35 | |
| openstackgerrit | Yuuichi Fujioka proposed a change to openstack/ceilometer: Implements monitoring-network-from-opendaylight https://review.openstack.org/63890 | 05:38 |
| *** sayali has quit IRC | 05:42 | |
| *** xianghui has quit IRC | 05:57 | |
| *** AMike has quit IRC | 05:59 | |
| openstackgerrit | Jenkins proposed a change to openstack/ceilometer: Imported Translations from Transifex https://review.openstack.org/62808 | 06:03 |
| *** xianghui has joined #openstack-ceilometer | 06:04 | |
| *** ildikov_ has quit IRC | 06:10 | |
| *** AMike has joined #openstack-ceilometer | 06:20 | |
| *** AMike has joined #openstack-ceilometer | 06:21 | |
| *** ryanpetrello has quit IRC | 06:23 | |
| *** sayali has joined #openstack-ceilometer | 06:26 | |
| *** ildikov_ has joined #openstack-ceilometer | 06:46 | |
| *** sayali has quit IRC | 06:52 | |
| *** SergeyLukjanov is now known as SergeyLukjanov_ | 06:58 | |
| *** sayali has joined #openstack-ceilometer | 06:59 | |
| *** nsaje has joined #openstack-ceilometer | 07:44 | |
| *** nsaje has quit IRC | 07:45 | |
| *** nsaje_ is now known as nsaje | 07:45 | |
| *** rwsu has quit IRC | 07:46 | |
| *** _nadya_ has joined #openstack-ceilometer | 07:48 | |
| *** xianghui has quit IRC | 07:49 | |
| *** rwsu has joined #openstack-ceilometer | 07:50 | |
| *** xianghui has joined #openstack-ceilometer | 08:00 | |
| *** _nadya_ has quit IRC | 08:03 | |
| *** sayali_ has joined #openstack-ceilometer | 08:12 | |
| *** _nadya_ has joined #openstack-ceilometer | 08:12 | |
| *** sayali has quit IRC | 08:15 | |
| *** _nadya_ has quit IRC | 08:21 | |
| *** yassine has joined #openstack-ceilometer | 08:50 | |
| *** urulama has joined #openstack-ceilometer | 08:54 | |
| *** SergeyLukjanov_ is now known as SergeyLukjanov | 08:56 | |
| *** eglynn has joined #openstack-ceilometer | 09:01 | |
| *** ildikov_ has quit IRC | 09:05 | |
| eglynn | urulama, nsaje: good morning gentlemen! | 09:06 |
| urulama | eglynn: morning :D | 09:06 |
| eglynn | urulama, nsaje: I see you guys are interested in time-constrained alarms | 09:06 |
| eglynn | i.e. https://blueprints.launchpad.net/ceilometer/+spec/time-constrained-alarms | 09:07 |
| nsaje | eglynn: good morning eoghan! | 09:07 |
| urulama | eglynn: yes | 09:07 |
| urulama | if it ok with you | 09:07 |
| eglynn | that's great, have at it! | 09:07 |
| eglynn | I'll review your etherpad | 09:07 |
| eglynn | i.e. https://etherpad.openstack.org/p/alarm_with_time_constraints ... this morning and leave any feedback there | 09:07 |
| urulama | great | 09:08 |
| urulama | we're doing some hpc stuff and these constraints will come handy ... | 09:08 |
| *** ildikov_ has joined #openstack-ceilometer | 09:09 | |
| eglynn | cool ... you can just assign the BP to yourself in launchpad | 09:09 |
| eglynn | ... and update the Implementation field to Started to relfect the fact that you're actively working on it | 09:09 |
| eglynn | urulama, nsaje: thank you sirs! | 09:09 |
| urulama | eglynn: will be our pleasure (i hope) :D | 09:10 |
| *** SergeyLukjanov is now known as SergeyLukjanov_ | 09:25 | |
| sileht | eglynn, I haven't fully read this blueprint (alarm_with_time_constraints) but I think have made a similar proposition at the summit https://blueprints.launchpad.net/ceilometer/+spec/alarm-over-time | 09:26 |
| sileht | https://wiki.openstack.org/wiki/Ceilometer/blueprints/alarm-over-time | 09:26 |
| sileht | The approch differ in adding a new field 'time_contraints' or adding a new alarm type | 09:30 |
| eglynn | sileht: yeah, I think we ended up with two BPs for pretty much the same thing | 09:30 |
| eglynn | sileht: ... the one I filed came out of this session at summit https://etherpad.openstack.org/p/icehouse-summit-ceilometer-future-of-alarming | 09:30 |
| eglynn | sileht: ... but seems like there was a pre-existing BP already filed | 09:31 |
| sileht | eglynn, In all case, I'm ok with both approch, and I have no time to code this now, so don't worry | 09:33 |
| eglynn | sileht: cool enough! | 09:33 |
| nsaje | sileht, eglynn : They look different to me though, sileht's triggers an alarm based on time, while eglynn's still triggers it based on threshold/meta - the time is just a constraint, not the actual trigger | 09:34 |
| * eglynn looks again | 09:34 | |
| sileht | nsaje, sure but my plan is to use combination alarm to constraint the threshold alarm with the time alarm | 09:34 |
| nsaje | sileht: ah, ok | 09:35 |
| sileht | but having the time constraint in the base alarm json permit a more efficient evaluation I think | 09:36 |
| sileht | eglynn, nsaje I have marked my BP supperseded by the Approved one | 09:36 |
| eglynn | sileht: thank you sir! | 09:36 |
| *** SergeyLukjanov_ is now known as SergeyLukjanov | 09:41 | |
| eglynn | jd__: good morning sir! | 09:45 |
| jd__ | hi eglynn | 09:45 |
| eglynn | jd__: so I see https://blueprints.launchpad.net/ceilometer/+spec/central-agent-improvement was bumped to juno | 09:45 |
| eglynn | (due to unreadiness of Tooz) | 09:46 |
| eglynn | seems https://blueprints.launchpad.net/ceilometer/+spec/rebase-alarm-partition-coordination will also have to move out to juno | 09:46 |
| * jd__ puppy eyes | 09:46 | |
| eglynn | (given that it's got the same dependency on a generic oslo co-ordination service) | 09:46 |
| jd__ | you're right | 09:46 |
| jd__ | changed | 09:47 |
| eglynn | jd__: thanks! | 09:47 |
| llu | jd__: ping | 09:51 |
| jd__ | llu: ? | 09:51 |
| llu | jd__: just saw your proposal about the resource loader thing. Are you suggesting that we add another item resource_loader into the pipeline, besides the 'resources'? | 09:52 |
| jd__ | llu: yes | 09:52 |
| jd__ | well feel free to have a counter proposal, it's just my wake-up's idea | 09:53 |
| llu | i'm just wondering if there is a way to combine resources resources loader into a single pipeline item | 09:54 |
| llu | resources and resource loader | 09:54 |
| jd__ | we can consider that if there's :// it's an url otherwise a loader I guess | 09:59 |
| llu | jd__: well, what if we need to pass some additional params to the loader? | 10:01 |
| llu | how about we choose a special type of url for loaders, for those URL using 'loader' as scheme, we treat it like a loader | 10:03 |
| llu | e.g. loader://nova_vms, loader://ironic_api?param1=value1¶m2=value2 | 10:03 |
| llu | for URL with other schemes, we just treat it like a final target endpoint | 10:04 |
| * jd__ thinks | 10:07 | |
| jd__ | llu: why not | 10:07 |
| llu | jd__: sorry for my poor english understanding. is it a yes or no to the 'loader://'? (why not? or why not something else?) | 10:10 |
| jd__ | ah sorry, that's a yes | 10:12 |
| jd__ | llu: http://idioms.thefreedictionary.com/why+not :) | 10:12 |
| llu | ok. thanks for your input. need to go to middle school english class :( | 10:13 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ceilometer: WIP: Add test for check sync models and migrations https://review.openstack.org/69674 | 10:15 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ceilometer: WIP: Add test for check sync models and migrations https://review.openstack.org/69674 | 10:17 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: samples: fix test case status code check https://review.openstack.org/69648 | 10:26 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: storage: bases of a Cassandra driver https://review.openstack.org/62779 | 10:26 |
| openstackgerrit | A change was merged to openstack/ceilometer: Remove unused db engine variable in api https://review.openstack.org/69340 | 10:28 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: storage: bases of a Cassandra driver https://review.openstack.org/62779 | 10:30 |
| openstackgerrit | Eugeniya Kudryashova proposed a change to openstack/python-ceilometerclient: Use HTTPClient from common Oslo code https://review.openstack.org/68939 | 10:50 |
| eglynn | urulama, nsaje: FYI I've put some initial feedback on https://etherpad.openstack.org/p/alarm_with_time_constraints | 10:58 |
| urulama | eglynn: tnx, will look at it soon. | 11:01 |
| urulama | eglynn: btw, i guess i don't have the rights to change the BP assignee ... | 11:02 |
| eglynn | urulama: np, now assigned to uros-jovanovic | 11:05 |
| openstackgerrit | A change was merged to openstack/ceilometer: Update dev docs to include notification-agent https://review.openstack.org/67252 | 11:20 |
| *** zul has quit IRC | 11:20 | |
| *** zul has joined #openstack-ceilometer | 11:24 | |
| *** ityaptin has joined #openstack-ceilometer | 11:32 | |
| *** xianghui has quit IRC | 12:31 | |
| * eglynn is about to pull the trigger on python-ceilometerclient 1.0.9 | 12:41 | |
| eglynn | ... with the following tag description http://fpaste.org/72701/90999234/raw/ | 12:41 |
| eglynn | ... if you want anything else included, speak now or forever hold your peace! ;) | 12:42 |
| eglynn | ... going, going ... | 12:45 |
| eglynn | ... gone! | 12:46 |
| *** yassine has quit IRC | 12:47 | |
| *** yassine has joined #openstack-ceilometer | 12:48 | |
| *** yassine has quit IRC | 12:48 | |
| *** yassine has joined #openstack-ceilometer | 12:48 | |
| *** yassine has quit IRC | 12:48 | |
| *** yassine has joined #openstack-ceilometer | 12:49 | |
| eglynn | 1.0.9 tarball: http://tarballs.openstack.org/python-ceilometerclient/python-ceilometerclient-1.0.9.tar.gz | 12:49 |
| eglynn | also up on pypi as always: https://pypi.python.org/pypi/python-ceilometerclient/1.0.9 | 12:49 |
| *** prad has joined #openstack-ceilometer | 13:01 | |
| *** yassine has joined #openstack-ceilometer | 13:05 | |
| nsaje | eglynn: would you please take a look at https://blueprints.launchpad.net/ceilometer/+spec/alarm-notification-details ? | 13:07 |
| eglynn | nsaje: will do | 13:07 |
| *** jdob has joined #openstack-ceilometer | 13:08 | |
| openstackgerrit | Pradeep Kilambi proposed a change to openstack/ceilometer: Fix measurement docs to correctly represent Existance meters https://review.openstack.org/69675 | 13:10 |
| openstackgerrit | Pradeep Kilambi proposed a change to openstack/ceilometer: Fix docs on what an instance meter represents https://review.openstack.org/66746 | 13:11 |
| eglynn | nsaje: yes, I think including reason_data in addition to the existing reason string in the alarm notification would be a good idea | 13:19 |
| eglynn | ... so basically number samples outside threshold, most recent datapoint, that kind thing right? | 13:19 |
| eglynn | *kind of thing | 13:19 |
| nsaje | exactly | 13:19 |
| nsaje | I have a use case currently where I have to parse these things from the reason string :) | 13:20 |
| eglynn | cool, knock yourself out! | 13:20 |
| eglynn | nsaje: ... obviously though the reason_data format would be different for combination alarms as opposed to threshold-oriented alarms | 13:20 |
| nsaje | eglynn: indeed, it is in the BP | 13:21 |
| eglynn | nsaje: cool, so it is! | 13:22 |
| eglynn | nsaje: ... I guess the list of alarms that fired is only strictly required in the OR case? | 13:23 |
| eglynn | nsaje: ... though yeah, you're right, probably best to include in all cases for consistency | 13:24 |
| nsaje | eglynn: the reason string currently contains all the alarms, not just the ones that fired | 13:25 |
| nsaje | eglynn: and it says "at least one of {alarms} fired" | 13:25 |
| *** sayali_ has quit IRC | 13:25 | |
| *** kwhitney has joined #openstack-ceilometer | 13:25 | |
| eglynn | nsaje: OK, so maybe an opportunity to improve that | 13:26 |
| nsaje | eglynn: but I think it would be good if only the one fired was listed | 13:26 |
| nsaje | eglynn: agreed! | 13:26 |
| eglynn | nsaje: ... as the only new info is which alarms actually fired in the OR case | 13:26 |
| eglynn | nsaje: cool | 13:26 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ceilometer: WIP: Add test for check sync models and migrations https://review.openstack.org/69674 | 13:31 |
| *** vrovachev has joined #openstack-ceilometer | 13:50 | |
| *** eglynn is now known as eglynn-lunch | 14:07 | |
| openstackgerrit | Ilya Tyaptin proposed a change to openstack/ceilometer: Skip unit tests with mongo or db2 when environment variables aren't set https://review.openstack.org/69644 | 14:10 |
| *** ryanpetrello has joined #openstack-ceilometer | 14:21 | |
| *** jmckind has joined #openstack-ceilometer | 14:36 | |
| *** tongli has joined #openstack-ceilometer | 14:37 | |
| *** eglynn-lunch is now known as eglynn | 14:43 | |
| *** SergeyLukjanov is now known as SergeyLukjanov_a | 14:46 | |
| *** SergeyLukjanov_a is now known as SergeyLukjanov_ | 14:47 | |
| gibi | eglynn: hi! | 14:48 |
| eglynn | gibi: hey | 14:48 |
| gibi | eglynn: first of all thanks for the positive comment on the complex query patch. | 14:48 |
| eglynn | gibi: np! | 14:48 |
| * gibi trying to gather his thoughts about distinct... | 14:49 | |
| eglynn | gibi: ... you had some hold-overs from yesterday's discussion on multiple groupbys in a single query? | 14:49 |
| eglynn | k | 14:49 |
| gibi | eglynn: jepp | 14:49 |
| gibi | eglynn: actually we have some thoughts on the distinct itself | 14:49 |
| eglynn | k, shoot | 14:50 |
| ildikov_ | eglynn: this was your API proposal: GET /v2/meters/instance/statistics?aggregate-by=distinct&aggregate-on=resource_id&period=p | 14:51 |
| eglynn | ildikov_: roughly speaking yes | 14:51 |
| ildikov_ | eglynn: and the mapping on db level: func.count(distinct(models.Meter.resource_id)).label('distinct') | 14:51 |
| eglynn | (modulo some tidy up on the query param names, e.g. s/aggregate-by/aggregateby/ | 14:52 |
| eglynn | yes, for the sqlalchemy driver | 14:52 |
| ildikov_ | eglynn: I just copied it from the log, it is not the main point :) | 14:52 |
| *** SergeyLukjanov_ is now known as SergeyLukjanov | 14:53 | |
| eglynn | different storage driver, different mapping ... extra logic in the reduce function for mongodb for ex. | 14:53 |
| *** SergeyLukjanov is now known as SergeyLukjanov_a | 14:53 | |
| ildikov_ | the point here is, that this API and db mapping is applicable only for the data that is needed for the dashboards | 14:53 |
| *** ruruj has joined #openstack-ceilometer | 14:54 | |
| eglynn | are we sure no other client of ceilo would want to disctinct values? | 14:54 |
| ruruj | hi | 14:54 |
| ildikov_ | eglynn: I mean, this mapping is now limited to the count() aggregate and it is important there that we filter to resource_id and project_id | 14:54 |
| ildikov_ | eglynn: it does not seem to be generic | 14:54 |
| gibi | eglynn: as ildikov_ said this distinct query is quite special we havent found any other query where the distinct helps | 14:54 |
| ruruj | regarding autoscaling...how can I evalutate only the ceilometer's data of those instances belonging to the stack instead of the data from all instances? | 14:54 |
| eglynn | ildikov_: so the idea is that aggregate-on being parameterized makes it generic | 14:55 |
| eglynn | ildikov_: one sec | 14:55 |
| *** SergeyLukjanov_a is now known as SergeyLukjanov_ | 14:55 | |
| ildikov_ | eglynn: do you have another query example for distinct? | 14:55 |
| gibi | eglynn: if you make distinct on resource_id then we can only apply count on the result | 14:56 |
| gibi | eglynn: at least max, min avg, does not have real meaning on resource_id | 14:56 |
| gibi | eglynn: if you apply distinct on counter_volume then again avg, max and min does not give to valuable result | 14:57 |
| gibi | s/to/too/ | 14:57 |
| eglynn | gibi: no the intention is that aggregate-on is a parameter *only* for the aggregate-by function | 14:58 |
| eglynn | gibi: not for all functions | 14:58 |
| eglynn | gibi: another example would be quantile computation | 14:58 |
| eglynn | say GET /v2/meters/cpu_util/statistics?aggregate-by=quantile&aggregate-on=0.99&period=p | 14:58 |
| eglynn | that would compute the 99th percentile for cpu_util in each period P | 14:59 |
| *** SergeyLukjanov_ is now known as SergeyLukjanov | 14:59 | |
| eglynn | but not max(0.99) or sum(0.99) which obviously has no meaning | 14:59 |
| eglynn | the idea is individually *parameterized* aggregate functions | 15:00 |
| *** sayali has joined #openstack-ceilometer | 15:00 | |
| gibi | eglynn: I guess quantile needs two parameter 0.99 and a counter_volume | 15:00 |
| * gibi is not good at math :) | 15:01 | |
| * ildikov_ neither :S | 15:01 | |
| eglynn | my thought was that for numeric aggregates (max, sum, avg, quantile, stddev), the counter_volume is implicit | 15:01 |
| eglynn | as it only ever makes sense the calculate say the standard deviation of the volumes | 15:01 |
| eglynn | not the timestamps or the meter_name or the resource_metadata | 15:02 |
| eglynn | ruruj: that's done on the basis of resource metadata identifying the instance as members of an autoscaling group | 15:02 |
| gibi | eglynn: OK. Then in the the distinct example query the distinct function would get also two parameter the value of the aggregate_on field and the counter_volume. However it does not make too much sense as here we only need the aggregate_on | 15:03 |
| eglynn | gibi: no, the counter-volme is not relevant to distinct | 15:03 |
| ildikov_ | eglynn: it seems that distinct is odd here | 15:04 |
| gibi | eglynn: yes, distinct is quite different from quantile | 15:04 |
| eglynn | gibi: the idea is that *some* aggregate functions require *some* parameters to guide their actions | 15:04 |
| eglynn | (not that *all* aggregate functions need to be parameterized) | 15:04 |
| ildikov_ | eglynn: I mean that mapping on the db level is one good solution for Horizon, but I do not see how it would be usabale in general | 15:04 |
| ildikov_ | eglynn: as for the other aggregates you mentioned, I totally agree, those look good | 15:05 |
| eglynn | why not usable in general? | 15:05 |
| *** boris-42 has quit IRC | 15:05 | |
| eglynn | do you think that horizon is the only usecase for distinct counting? | 15:06 |
| gibi | eglynn: distinct is not a numeric aggregate function, it can be applied on any type of fields. In the other hand when distinct is applyed then there is only count() meaningfull on the result | 15:06 |
| gibi | eglynn: this makes distinct is so specific | 15:06 |
| gibi | s/is// | 15:06 |
| ildikov_ | eglynn: I do not see too many use cases right now for distinct counting, but I can be wrong here | 15:06 |
| eglynn | TBH distinct wasn't my main interest in adding parameterizable aggregates | 15:07 |
| ruruj | eglynn, can I explicit differentiate the instances from the horizon web gui when launch a new stack? | 15:07 |
| eglynn | ruruj: I can't remember if horizon displays resource metadata | 15:08 |
| eglynn | ruruj: you may have to use the nova CLI | 15:08 |
| eglynn | ruruj: ... isn't the stack name also embedded in the generated server names? | 15:08 |
| gibi | eglynn: we dont have problems with your idea of parameterized aggregates. I like that proposal. But distinct does not fitt well in it. | 15:08 |
| ildikov_ | eglynn: here comes my idea :) | 15:09 |
| ruruj | eglynn, yes it is .. :| | 15:09 |
| ildikov_ | eglynn: we were talking about meters yesterday for supporting the needs of Horizon | 15:09 |
| ildikov_ | eglynn: I had a wild thought yesterday, that we could create a new meter for this need actually | 15:10 |
| ildikov_ | eglynn: we could use the notifications about instances (create, delete) or have a pollster maybe | 15:10 |
| eglynn | ildikov_: hmmm, seems like meter proliferation | 15:11 |
| eglynn | ildikov_: I mean we already have the necessary data in the metering store | 15:11 |
| ruruj | but when I launch a new stack the total cpu utilization is calculated among all instances :| | 15:11 |
| openstackgerrit | A change was merged to openstack/ceilometer: Use swift master https://review.openstack.org/68150 | 15:12 |
| *** boris-42 has joined #openstack-ceilometer | 15:12 | |
| eglynn | ruruj: have you access to the ceilometer CLI? | 15:12 |
| ruruj | eglynn, yes | 15:12 |
| *** vrovachev has left #openstack-ceilometer | 15:12 | |
| eglynn | ruruj: can you run ceilometer alarm-list and ceilometer alarm-show -a $ALARM_ID for the high CPU alarm for your stack? | 15:13 |
| eglynn | folks, I'm gonna have to drop for 10-15 mins for an internal huddle | 15:13 |
| ildikov_ | eglynn: I was thinking about leaving the db out of the game, as it is an extra load to regenerate this in every iteration for Horizon | 15:13 |
| eglynn | ildikov_: can you hold that thought for a few mins? | 15:14 |
| *** eglynn is now known as eglynn-call | 15:14 | |
| ildikov_ | eglynn-call: sure :) | 15:15 |
| ruruj | eglynn-call, alarm-list output: http://pastebin.com/eUgUhkW7 and alarm-show : http://pastebin.com/8AATbSZZ | 15:18 |
| ityaptin | sileht, hi! I want to implement hbase ttl from bp https://blueprints.launchpad.net/ceilometer/+spec/db-ttl which is assigned to you. | 15:19 |
| sileht | ityaptin, this BP is already implemented and only cover mongodb and sqlalchemy | 15:20 |
| sileht | ityaptin, you can create a new one for hbase | 15:20 |
| sileht | ityaptin, I'm pretty sure jd__ will approved it | 15:20 |
| sileht | jd__, ? | 15:20 |
| jd__ | sure | 15:21 |
| ityaptin | sileht, jd__, it's great) | 15:21 |
| ityaptin | jd__, bp for this - https://blueprints.launchpad.net/ceilometer/+spec/hbase-db-ttl | 15:26 |
| jd__ | ityaptin: you're good | 15:26 |
| *** _nadya_ has joined #openstack-ceilometer | 15:27 | |
| ityaptin | jd__, thanks) | 15:27 |
| ruruj | eglynn-call, and if it may be useful, this is my template http://pastebin.com/UissYAEk | 15:28 |
| *** zul has quit IRC | 15:30 | |
| *** zul has joined #openstack-ceilometer | 15:30 | |
| *** gordc has joined #openstack-ceilometer | 15:35 | |
| eglynn-call | ruruj: you're missing something like "matching_metadata: {'metadata.user_metadata.server_group': 'ServerGroup'}" in the CPUAlarmHigh and CPUAlarmLow | 15:37 |
| *** sayali has quit IRC | 15:37 | |
| *** sayali has joined #openstack-ceilometer | 15:37 | |
| *** eglynn-call is now known as eglynn | 15:37 | |
| ruruj | eglynn, uh...thank you!! I'll try to fix! | 15:38 |
| *** sayali has quit IRC | 15:38 | |
| eglynn | ruruj: also you'll need ... | 15:38 |
| eglynn | repeat_actions: True | 15:38 |
| *** sayali has joined #openstack-ceilometer | 15:39 | |
| eglynn | otherwise the lack of continuous notification will cuase the heat cooldown state machine to never progress to actual scaling | 15:39 |
| eglynn | gibi, ildikov_: sorry 'bout that, back now | 15:39 |
| ruruj | eglynn, repeat_actions...uhm..thanks | 15:40 |
| ildikov_ | eglynn: np! | 15:40 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ceilometer: POC: Sync models to actual PostgreSQL db state https://review.openstack.org/69883 | 15:40 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ceilometer: POC: Add test for check sync models and migrations https://review.openstack.org/69674 | 15:40 |
| eglynn | ildikov_: so the alternative proposal is a completely new meter, instances_per_tenant? | 15:40 |
| ildikov_ | eglynn: kind of yes | 15:41 |
| eglynn | what happens if the horizon folks come back tomorrow and ask for instances_per_user instead? ;) | 15:41 |
| gibi | eglynn: or instance_count:active, instance_count:stopped ... etc | 15:41 |
| eglynn | i.e. it seems like a new meter is quite static | 15:42 |
| eglynn | (WRT changing requirements) | 15:42 |
| eglynn | whereas a query could be reformulated | 15:42 |
| eglynn | in this case groupby=project_id could be replaced by groupby=user_id | 15:43 |
| ildikov_ | eglynn: you have a point here, it seems to be a usable information, so I thought to reduce the number of periodic queries on the db | 15:45 |
| eglynn | ildikov_: wouldn't we have to go to the DB anyway to retrieve the special-purpose meter? | 15:46 |
| ildikov_ | eglynn: the distinct itself to be shown on the API also does not seem to be a completely valid solution in my opinion, so I was thinking about hiding it behind a meter | 15:46 |
| eglynn | (though, yeah, it would be a much cheaper periodic DB access to grab the counter value on a specific meter) | 15:46 |
| openstackgerrit | Ryan Petrello proposed a change to openstack/ceilometer: Correct a misuse of RestController in the Event API. https://review.openstack.org/69885 | 15:47 |
| ildikov_ | eglynn: in my idea is to generate the value according to the incoming data (from notifications or pollster) and then the query would target a simple meter without the need of an aggregation | 15:48 |
| openstackgerrit | Ryan Petrello proposed a change to openstack/ceilometer: Correct a misuse of RestController in the Event API. https://review.openstack.org/69885 | 15:48 |
| eglynn | ildikov_: the notifications are received per-instance, as opposed to per-tenant | 15:50 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ceilometer: POC: Sync models to actual PostgreSQL db state https://review.openstack.org/69883 | 15:50 |
| openstackgerrit | Victor Sergeyev proposed a change to openstack/ceilometer: POC: Add test for check sync models and migrations https://review.openstack.org/69674 | 15:50 |
| eglynn | ildikov_: (so I'm not sure you could count the number of instances per-tenant in a notification handler) | 15:50 |
| eglynn | ildikov_: so you'd be taking about a pollster that calls into the nova servers API with all_tenants=True and then calculates the total for each tenant? | 15:51 |
| gibi | eglynn: yes, it should be a pollster | 15:52 |
| ildikov_ | eglynn: yes, a pollster sounds reasonable here | 15:52 |
| eglynn | gibi, ildikov_: a pollster shifting through a potentially large number of servers to calculate the per-tenant totals? | 15:53 |
| ildikov_ | eglynn: sorry, yesterday the light was suddenly turned on in my head, so I do not have a deeply worked out solution, just the base idea, which I thought to validate here if it could work or not | 15:53 |
| gibi | eglynn: actually yes, we should get all Server state from the nocva | 15:54 |
| gibi | s/nocva/nova | 15:54 |
| eglynn | gibi: well we could get these data from nova, but having a pollster doing the "slicing and dicing" on a potentially large dataset seems a bit wrong | 15:55 |
| eglynn | (... by "slicing and dicing" I mean the totalling up per-tenant or per-user or whatever) | 15:57 |
| gibi | eglynn: I accept that it is a big no-no. :) | 15:58 |
| ildikov_ | eglynn: yes it seems to be a wrong direction, I haven't calculated with the fact that we do not have the latest value of that meter on the pollster level | 15:58 |
| ildikov_ | eglynn: s/pollster/notification/ | 15:58 |
| eglynn | ildikov_: a-ha, ok, I see what you mean ... yeah without that being directly available from nova, there's a bit of a burden on the pollster | 15:59 |
| eglynn | sorry folks, I'm running up against the shot-clock here | 16:00 |
| eglynn | as always ;) | 16:00 |
| gibi | eglynn: thanks for your help | 16:00 |
| *** _nadya_ has quit IRC | 16:00 | |
| gibi | eglynn: as always :) | 16:00 |
| eglynn | so to sum up ... | 16:01 |
| eglynn | 1. maybe the special-purpose meter isn't a great idea from the PoV of the analytic burden placed on the pollster | 16:01 |
| ildikov_ | eglynn: another option is to create a kind of aggregated meter to hide the distinct maybe, but continue it another time then :) | 16:01 |
| eglynn | 2. but ildikov_ & gibi still not loving the idea of distinct as a paremeterizable aggregate function | 16:01 |
| gibi | eglynn: yes, agree with your sum up | 16:02 |
| eglynn | gibi, ildikov_: cool, let's ruminate on the options a bit and chat again later in the week | 16:02 |
| gibi | eglynn: OK. Thank you again. | 16:02 |
| ildikov_ | eglynn: that sounds good | 16:03 |
| ildikov_ | eglynn: many-many thanks :) | 16:03 |
| eglynn | ildikov_: np! | 16:03 |
| *** yassine has quit IRC | 16:12 | |
| *** yassine has joined #openstack-ceilometer | 16:13 | |
| *** sayali_ has joined #openstack-ceilometer | 16:22 | |
| *** sayali has quit IRC | 16:25 | |
| openstackgerrit | Eugeniya Kudryashova proposed a change to openstack/python-ceilometerclient: Use HTTPClient from common Oslo code https://review.openstack.org/68939 | 16:26 |
| *** eglynn is now known as eglynn-afk | 16:27 | |
| *** SergeyLukjanov is now known as SergeyLukjanov_ | 16:34 | |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: Update oslo https://review.openstack.org/69903 | 16:43 |
| openstackgerrit | Jenkins proposed a change to openstack/ceilometer: Updated from global requirements https://review.openstack.org/69905 | 16:57 |
| *** SergeyLukjanov_ is now known as SergeyLukjanov | 17:04 | |
| *** ildikov_ has quit IRC | 17:14 | |
| *** xmltok has joined #openstack-ceilometer | 17:25 | |
| *** SergeyLukjanov is now known as SergeyLukjanov_a | 17:29 | |
| *** SergeyLukjanov_a is now known as SergeyLukjanov_ | 17:30 | |
| *** jaypipes_ has joined #openstack-ceilometer | 17:37 | |
| *** jaypipes_ has quit IRC | 17:37 | |
| *** SergeyLukjanov_ is now known as SergeyLukjanov | 17:38 | |
| *** ruruj has quit IRC | 17:51 | |
| *** ildikov has joined #openstack-ceilometer | 18:13 | |
| *** doug-fish has joined #openstack-ceilometer | 18:42 | |
| doug-fish | Hi Ceilometer people, I'm trying to advise translators on the meaning of "Duration of network" which is in http://docs.openstack.org/developer/ceilometer/measurements.html#network-neutron | 18:45 |
| doug-fish | Does it mean the amount of time from when the network was started until now? | 18:46 |
| gordc | doug-fish: you might not want to use 'Duration of xyz' | 18:54 |
| gordc | we're fixing that note to something more accurate here: https://review.openstack.org/#/c/69675/3 | 18:54 |
| openstackgerrit | Ryan Petrello proposed a change to openstack/ceilometer: Correct a misuse of RestController in the Event API. https://review.openstack.org/69885 | 18:54 |
| doug-fish | gordc: thanks! I'll look at the update. | 18:55 |
| gordc | doug-fish: np. not sure what term we'll finalise on but it definitely won't be 'Duration' | 18:56 |
| openstackgerrit | Ryan Petrello proposed a change to openstack/ceilometer: Correct a misuse of RestController in the Event API. https://review.openstack.org/69885 | 18:57 |
| *** yassine has quit IRC | 19:02 | |
| *** yassine has joined #openstack-ceilometer | 19:03 | |
| *** yassine has quit IRC | 19:04 | |
| *** yassine has joined #openstack-ceilometer | 19:04 | |
| *** sayali_ has quit IRC | 19:07 | |
| *** yassine has quit IRC | 19:09 | |
| dhellmann | is anyone else seeing intermittent errors running the unit tests? | 19:32 |
| gordc | dhellmann: off master? | 19:38 |
| dhellmann | gordc: yeah | 19:39 |
| dhellmann | gordc: I fixed the testtools dependency and downgraded tox, so these are just ceilometer's tests failing | 19:39 |
| dhellmann | tests.test_bin.BinSendCounterTestCase.test_send_counter_run | 19:39 |
| dhellmann | tests.test_utils.TestUtils.test_dict_to_kv | 19:39 |
| dhellmann | they vary | 19:39 |
| gordc | hmm. i'll give it a try. i just realised when i was cleaning up python packages i deleted tox so need to fix that before i can try. | 19:40 |
| *** _nadya_ has joined #openstack-ceilometer | 19:41 | |
| dhellmann | gordc: I wonder if these are somehow related to testing library changes | 19:48 |
| *** DanD_ has joined #openstack-ceilometer | 19:48 | |
| *** jmckind has quit IRC | 19:49 | |
| dhellmann | gordc: ah, I can reproduce a failure in test_bin.py reliably | 19:52 |
| gordc | dhellmann: i actually get quite a few failures... not sure if it's related to all the system changes i've been making though. | 19:55 |
| dhellmann | gordc: yeah, I just rebuilt my dev vm so I had similar questions | 19:56 |
| gordc | dhellmann: do you get errors relating to alarms? | 19:57 |
| dhellmann | gordc: yeah, I was seeing some of those, too | 19:57 |
| dhellmann | less frequently than some of the others | 19:58 |
| dhellmann | tests.test_bin.BinSendCounterTestCase.test_send_counter_run fails every time, with "ImportError: No module named ceilometer.openstack.common" from the script being run out of bin | 19:58 |
| gordc | i guess it's a bad thing we both were playing around with our systems. feels like blind leading the blind.lol | 19:58 |
| dhellmann | heh | 20:00 |
| gordc | dhellmann: i don't have the test_bin error (maybe i just don't see it) | 20:01 |
| gordc | dhellmann: i did get less errors this second pass i ran... maybe related to new tox? i just noticed it got bumped to 1.7 (which i downloaded) | 20:01 |
| dhellmann | gordc: and now I'm not seeing the alarm errors | 20:01 |
| dhellmann | fantastic, this passes: tox -e py27 -- ceilometer/tests/test_bin.py | 20:03 |
| dhellmann | but without the arg to run just the one test, that test fails | 20:03 |
| gordc | hmm. i got more alarm errors the third time running tests. | 20:06 |
| dhellmann | I can't make testr show me any info, so I've resorted to writing a temporary file from within the test | 20:08 |
| tongli | @dhellmann, I did exactly the same. can anyone tell me how to show some varilables when run_tests.sh? | 20:11 |
| dhellmann | tongli: try setting OS_STDOUT_CAPTURE=1 | 20:13 |
| dhellmann | that didn't work in my case, but maybe it will for you | 20:14 |
| tongli | @dhellmann, thanks, let me try that. | 20:14 |
| *** jdob has quit IRC | 20:15 | |
| tongli | @dhellmann, that did not work for me. | 20:21 |
| dhellmann | tongli: sorry :-( | 20:22 |
| tongli | @dhellmann, yeah, I did what you did, write to a file. so annoying. | 20:22 |
| dhellmann | tongli: I've been able to make it work sometimes | 20:23 |
| tongli | @dhellmann, without writing to a file? | 20:23 |
| openstackgerrit | litong01 proposed a change to openstack/python-ceilometerclient: add notification alarm https://review.openstack.org/69474 | 20:25 |
| openstackgerrit | Doug Hellmann proposed a change to openstack/ceilometer: Force import path to be right in test_bin.py https://review.openstack.org/69963 | 20:28 |
| dhellmann | gordc: ^^ | 20:28 |
| gordc | dhellmann: i've resorted to reverting back to a snapshot i took 4 months ago. maybe starting from there will help me. | 20:29 |
| gordc | will give a try once i get it back up to date | 20:30 |
| *** _nadya_ has quit IRC | 20:30 | |
| dhellmann | gordc: ok | 20:31 |
| dhellmann | gordc: I ran the tests several times without any failures after making that change | 20:31 |
| dhellmann | I have no idea why that might have made the alarm tests fail | 20:31 |
| tongli | @dhellmann, Doug, I run run_tests.sh without problem. I have latest code. | 20:34 |
| dhellmann | tongli: :-/ | 20:34 |
| tongli | @dhellmann, I did not have your 69963 patch. | 20:35 |
| tongli | ./run_tests.sh run ok, tox -e py27 ok as well. | 20:35 |
| *** jmckind has joined #openstack-ceilometer | 20:42 | |
| openstackgerrit | Ryan Petrello proposed a change to openstack/ceilometer: Correct a misuse of RestController in the Event API https://review.openstack.org/69885 | 20:46 |
| ryanpetrello | submitting a patch that passes Python tests, but realizing jenkins failed you because the commit message has a period at the end: http://i.imgur.com/IiMp5.gif | 20:48 |
| *** _nadya_ has joined #openstack-ceilometer | 20:50 | |
| *** terriyu has joined #openstack-ceilometer | 20:52 | |
| *** _nadya_ has quit IRC | 20:52 | |
| *** nadya has joined #openstack-ceilometer | 20:58 | |
| *** nadya is now known as Guest83076 | 20:58 | |
| *** Guest83076 is now known as _nadya_ | 20:59 | |
| *** eglynn-afk is now known as eglynn | 21:04 | |
| *** _nadya__ has joined #openstack-ceilometer | 21:05 | |
| *** _nadya_ has quit IRC | 21:07 | |
| chmouel | ryanpetrello: i may use that on osreactions :) | 21:13 |
| ryanpetrello | do it :) | 21:13 |
| chmouel | ryanpetrello: do you have a twitter handle so i can give you credit for? | 21:13 |
| ryanpetrello | @ryanpetrello :) | 21:13 |
| chmouel | nice and easy cool | 21:13 |
| *** _nadya__ has quit IRC | 21:13 | |
| chmouel | ryanpetrello: on another note a lot of projects have removed that don't allow a period in title rule, i think ceilometer should do too | 21:15 |
| jd__ | honestly we don't encounter that often :) | 21:27 |
| *** lsmola has joined #openstack-ceilometer | 21:41 | |
| *** urulama has quit IRC | 21:47 | |
| *** Alexei_987 has quit IRC | 22:00 | |
| *** tongli has quit IRC | 22:00 | |
| *** SergeyLukjanov is now known as SergeyLukjanov_ | 22:01 | |
| *** lsmola has quit IRC | 22:05 | |
| ryanpetrello | jd__: yea, only dummies like me :) | 22:08 |
| *** gordc has quit IRC | 22:18 | |
| *** jmckind has quit IRC | 22:21 | |
| *** ildikov has quit IRC | 22:34 | |
| *** ryanpetrello has quit IRC | 23:17 | |
| *** prad__ has joined #openstack-ceilometer | 23:25 | |
| *** prad has quit IRC | 23:27 | |
| *** prad__ is now known as prad | 23:27 | |
| *** terriyu has quit IRC | 23:35 | |
| *** prad has quit IRC | 23:43 | |
| *** doug-fish has quit IRC | 23:53 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
