Thursday, 2015-02-26

*** gordc has quit IRC00:10
*** david-lyle is now known as david-lyle_afk00:14
*** ViswaV has quit IRC00:51
*** sbfox1 has quit IRC00:56
*** fnaval has quit IRC01:04
*** ViswaV has joined #openstack-ceilometer01:23
*** rwsu is now known as rwsu-afk01:27
*** _cjones_ has quit IRC01:31
*** amalagon has joined #openstack-ceilometer01:38
*** ViswaV has quit IRC01:48
*** ddieterly has joined #openstack-ceilometer01:49
*** ddieterly has quit IRC02:02
*** ddieterl_ has joined #openstack-ceilometer02:02
*** ddieterl_ has quit IRC02:30
*** ddieterly has joined #openstack-ceilometer02:58
*** ddieterly has quit IRC03:02
openstackgerritZhiQiang Fan proposed openstack/ceilometer: drop deprecated novaclient.v1_1
openstackgerritZhiQiang Fan proposed openstack/ceilometer: fix config opts in objectstore.rgw
*** deepthi has joined #openstack-ceilometer03:51
*** nijaba has quit IRC04:06
*** nijaba has joined #openstack-ceilometer04:08
*** nijaba has quit IRC04:08
*** nijaba has joined #openstack-ceilometer04:08
*** sbfox has joined #openstack-ceilometer04:55
*** ddieterly has joined #openstack-ceilometer05:00
*** ddieterly has quit IRC05:04
*** fnaval has joined #openstack-ceilometer05:32
*** fnaval has quit IRC05:32
*** amalagon has quit IRC05:36
*** ParsectiX has quit IRC05:59
*** ParsectiX has joined #openstack-ceilometer06:00
*** ddieterly has joined #openstack-ceilometer06:01
*** ParsectiX has quit IRC06:04
*** ParsectiX has joined #openstack-ceilometer06:04
*** ddieterly has quit IRC06:05
*** ParsectiX has quit IRC06:05
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometer: Imported Translations from Transifex
*** ViswaV has joined #openstack-ceilometer06:11
*** ViswaV_ has joined #openstack-ceilometer06:15
*** ViswaV has quit IRC06:16
*** yatin has joined #openstack-ceilometer06:43
*** ddieterly has joined #openstack-ceilometer07:01
*** ddieterly has quit IRC07:06
*** ildikov has quit IRC07:21
*** ddieterly has joined #openstack-ceilometer08:02
*** ddieterly has quit IRC08:06
*** ildikov has joined #openstack-ceilometer08:09
*** ViswaV_ has quit IRC08:15
*** karimb has joined #openstack-ceilometer08:49
*** Ala has joined #openstack-ceilometer08:49
*** ccrouch has quit IRC08:56
*** ccrouch has joined #openstack-ceilometer09:11
openstackgerritMehdi Abaakouk proposed stackforge/gnocchi: ceph: don't rely on a particular agg. method
*** sbfox has quit IRC09:24
*** ddieterly has joined #openstack-ceilometer10:04
*** ddieterly has quit IRC10:08
*** EmilienM|afk is now known as EmilienM10:11
vrovacheveglynn-officeafk: jd__: I just want to remind about integration tests :)
*** IvanBerezovskiy_ has joined #openstack-ceilometer10:27
*** eglynn-officeafk is now known as eglynn-office10:35
eglynn-officevrovachev: thanks for the reminder! ... will get to it today10:36
vrovacheveglynn-office: Okay, thank you :)10:36
*** _nadya_ has joined #openstack-ceilometer10:37
*** _nadya_ has quit IRC10:37
*** eglynn has joined #openstack-ceilometer10:38
*** _nadya_ has joined #openstack-ceilometer10:38
eglynn-officevrovachev: just a couple of minor nitpicks, otherwise pretty much good to go IMO10:52
*** himangi has joined #openstack-ceilometer10:54
ildikoveglynn-office: hi10:57
eglynn-officeildikov: hey10:57
*** gzhai_mob has joined #openstack-ceilometer10:57
ildikoveglynn-office: maybe I'm missing some basics regarding to integration tests10:58
ildikoveglynn-office: those have to live in-tree too?10:58
ildikoveglynn-office: I mean functional test != integration test10:58
ildikoveglynn-office: or does that Nova folder cover both?10:59
eglynn-officeildikov: well it's arguable where the boundary lies between a func test and an integration test11:00
eglynn-officeildikov: IIUC in the nova case, there are tests to for example spin up a server11:00
eglynn-officeildikov: that would require glance IIUC11:00
eglynn-officeildikov: unless they mock that out in some11:01
*** gzhai_mob has quit IRC11:02
ildikoveglynn-office: ok, I see11:03
ildikoveglynn-office: but do we have our own boundary?11:03
ildikoveglynn-office: or is it a decision to keep everything in-tree?11:04
*** ddieterly has joined #openstack-ceilometer11:04
eglynn-officeildikov: in order for ceilometer to consume actual notifications and poll actual services, we need to run a DSVM no?11:05
ildikoveglynn-office: yes11:06
eglynn-officejust checked with the nova folks BTW and glance is mocked for those tests11:06
eglynn-officealso the olso.messaging bus is faked out11:06
eglynn-officeso only the nova services run11:06
ildikoveglynn-office: I wasn't sure about that part in the functional tests, we have some patches in NOva with gibi, but it's a mess in my head that which test was in which folder, but we had issues with RPC in that oslo.messaging mock...11:07
eglynnildikov: call starting, bbiab11:07
ildikoveglynn: ok11:08
openstackgerritVadim Rovachev proposed openstack/ceilometer-specs: Add specification for integration tests.
*** ddieterly has quit IRC11:09
*** nellysmitt has joined #openstack-ceilometer11:10
vrovacheveglynn: ildikov: Thank you so much for review, I answered and resolved comments.11:11
*** cdent has joined #openstack-ceilometer11:11
openstackgerritZhiQiang Fan proposed openstack/ceilometer: Add the function of deleting alarm history
vrovachevildikov: Hi, all tests for differents DBs it's functional tests, DSVM-style tests(not yet written) it's integration tests.11:16
ildikovvrovachev: I'm fine with DB tests and that those are functional tests11:17
ildikovvrovachev: as for integration tests I thought it will remain in Tempest conceptually11:17
ildikovvrovachev: therefore I got confused a bit11:18
cdentildikov: from the conversataions I’ve had with the tempest folk, long term they want to shrink tempest11:19
*** pcaruana has joined #openstack-ceilometer11:19
cdentespecially with regard to what they consider non-core projects11:19
ildikovcdent: shrink == remove or smth like?11:19
vrovachevildikov: QAs don't wants them to have such specific tests on tempest. This was discussed at the summit.11:19
cdentI don’t think it is black and white: It’s a progression: the projects take on greater responsibility for self-hosting their own test code, whatever the type.11:20
cdent(the type of tests)11:20
ildikovvrovachev: well, functional tests are the less specific I guess :)11:21
vrovachevildikov: If QAs were not against it, tests in temset would of course it's much easier :)11:21
*** cdent has quit IRC11:21
ildikovcdent: in case of integration with other services I guess this a very "grey" question...11:22
vrovachevildikov: so, functional tests were originally in-tree :) We want just move these to different folder11:22
ildikovvrovachev: as I said the functional test folder is fine, just like how eglynn-office mentioned Nova as an example11:23
vrovachevildikov: ah,  okay :)11:23
ildikovvrovachev: just the integration test folder looks odd a bit11:24
ildikovvrovachev: as the part is not only for testing our code11:25
vrovachevildikov: It's extreme measures, unfortunately.11:25
vrovachevildikov: In fact, many stackforge projects are practicing this approach11:26
ildikovvrovachev: sure, well, even if it will be this way I will not like it, but obviously that's not your fault :)11:27
vrovachevildikov: ah, thanks :)11:28
ildikovvrovachev: so as a sum of the above discussion, the thing is that we would like to have the integration tests in Tempest, but it's not inline with what the QA folks think, so we now plan to implement these in-tree to still have coverage11:32
ildikovvrovachev: did I understand right the "issue"?11:33
vrovachevildikov: allright11:33
ildikovvrovachev: that sounds a bit sad to me :(, but thanks for clarification :)11:34
ildikovvrovachev: just one more quick question11:35
vrovachevildikov: yes?11:35
ildikovvrovachev: would that be possible to move these integration tests to Tempest in the future, if that wouldn't be blocked conceptually?11:36
ildikovvrovachev: I mean if let's say a half year later someone would tell that we are free to put our tests there, then it would mean a complete rewrite or just some modifications?11:38
*** _nadya_ has quit IRC11:39
vrovachevildikov: i think. that it's blocked conceptually. Because of on DSVMs JOBs uses mysql or postgres and on these DB and with very more strong tempest load for these DB adding notification to DB is very long(~20min in end of tests)11:40
ildikovvrovachev: uh, this sounds bad in general :(11:41
vrovachevildikov: We wanted to create Job with Mongo, but that QAs did not take11:41
ildikovvrovachev: got it :(11:42
ildikovvrovachev: I just have the feeling with all the integartion tests in-tree that we kinda take the responsibility for the sender side of notifications and every other stuff that is affected11:45
ildikovvrovachev: but I guess I just have to change my viewpoint... :)11:45
*** deepthi has quit IRC11:46
eglynnvrovachev, ildikov: sorry, on a call, back now11:49
vrovachevildikov: I think, when we will finish to write integration tests, we will to see backward compatibility for all integrated projects. So, I think it's not so bad.11:49
eglynnso as cdent said above, the long term preference of QA is for things to migrate out of Tempest11:49
eglynnso that Tempest ends up as a slimed down set of pure API tests IIUC11:50
ildikoveglynn: hmm, so API testing in the functional tests and in Tempest too?11:50
ildikoveglynn: well, it will be easier to maintain that scope for sure11:51
eglynnildikov: some duplication initially with a view to removing eventually from Tempest11:51
eglynnso take notification driven tests11:51
eglynnthe QA don't like that we have them currently in tempest11:51
eglynnso we replicate equivalent tests in tree11:51
eglynnrun experimentally for a while11:52
eglynnthen eventually deprecate the equivalent tests in tempests11:52
ildikoveglynn: I got the process, I just had a memory snippet that Tempest will keep the integration test part, which did not mean only API tests for me11:52
vrovachevildikov:  For example, if some project makes changes in notification code(eg change in notification json), then we will to see it on fails job. And will can to create a bug for ceilo or for some project.11:53
ildikoveglynn: so I clearly understand what and how we would like to do, I just didn't really understand the why part11:53
*** cdent has joined #openstack-ceilometer11:54
eglynnildikov: well, part of it is that Tempest has just grown into a monster that the QA team are finding very difficult to manage11:54
ildikovvrovachev: as for notifications, I hope that the notification schema initiative will not die before the implementation phase to make life easier for everyone11:54
eglynnildikov: specifically when unexplained failure arise11:55
eglynnildikov: they can't see that approach of throwing the kitchen sink into tempest as being sustainable11:55
vrovachevildikov: Oh, that's for sure, I hope so too.11:55
eglynnildikov: esp. in the context of big tent11:55
ildikoveglynn: for a monster like OpenStack QA is hell anyway IMHO...11:56
cdentildikov: the last I heard from the notification schema folk is that they are currently codifying existing stuff. I'm not sure this is the best way forward, nor is their lack of visibility great.11:57
ildikoveglynn: I just did not have the information that inetgration tests are not really welcomed there, I know that our integration tests are even more difficult to get in11:57
*** _nadya_ has joined #openstack-ceilometer11:58
ildikovcdent: the progress is good, otherwise we will not get anywhere ever, but the lack of visibility is a problem I agree11:58
eglynnildikov: giving us the flexibility to write integration tests that wouldn't pass muster in tempest (e.g. because they need to do odd things like restarting services) is one of the big motivations from our PoV11:59
eglynnildikov: i.e. the motivation is not just ... we're doing this to keep the qa folks happy ;)11:59
ildikoveglynn: I have my opinion about that part... :)12:00
cdentI think the biggest win of in tree integration and functional tests is that I can run them (given suitable tooling) before I ever submit code.12:00
ildikoveglynn: but hopefully we will not restart those services till the end of our or OpenStack's life ;)12:01
ildikovcdent: you can run tempest tests too12:01
cdentyes, but it is a good deal more painful and less focused12:02
cdent(assuming we do it right on our side)12:02
eglynnildikov: end of openstack's life? ... this reich will last a thousand beers! ... sorry, years!12:04
ildikovwell, I'm still not convinced to like this, but I can accept that we do not have (m)any other choices12:04
eglynnildikov: more seriouslu, it allows us the flexibility to take "tactical" approaches in the meantime even if restarting services etc. won't be needed forevere12:04
*** ddieterly has joined #openstack-ceilometer12:05
cdentildikov: I may have missed you saying this already, but why do you prefer in tempest?12:05
ildikoveglynn: let's try to make integration as easy as it possible rather than have many tactics in the test suite12:06
ildikovcdent: I wrote above that I think I have to change my viewpoint12:07
vrovachevcdent: because of integration test must be with other integration tests :)12:07
ildikovcdent: integration is one level higher for than just simply how our code works12:08
_nadya_actually it's interesting how is tempest doing? is it under active development now? vrovachev, have you committed anything there in this cycle?12:09
*** ddieterly has quit IRC12:09
_nadya_hi folks :)12:09
vrovachevildikov: Yes, that is why we will to create DSVM style job12:09
ildikovcdent: like we would be the only responsibles to fully adapt to everyones continuously changing notification formats12:09
ildikovvrovachev: yeap, sure, that's all good12:10
vrovachev_nadya_: in tempest I have two not merged patches, on in-tree integration tests I have spec :)12:11
ildikovvrovachev: it's just I'm the "big picture girl" here, it looks annoying from the outside I guess, but well, that's me :) and I usually think that it's better to share concerns coming from this way of thinking as it gives a chance to consider even more pros and cons12:11
ildikovbut I don't think it's a problem if someone has another opinion12:13
vrovachevildikov: okay :)12:13
ildikovcdent, eglynn: as for other opinion, I had time to only quickly check the asyncio patch12:14
eglynnildikov: thanks!12:15
* eglynn forages for food12:15
*** eglynn is now known as eglynn-lunch12:15
ildikovvrovachev: so I'm not against your patch, just wanted to see the whole picture :)12:15
ildikovcdent, eglynn-lunch: I learnt that it's still not a 100% welcomed change and I still have my concern regarding to it will be too much for us as for changes12:16
*** rodrigods is now known as rodrigods_12:16
*** rodrigods_ is now known as rodrigods__12:16
*** rodrigods__ is now known as rodrigod`12:16
*** rodrigod` is now known as rodrigods12:17
cdentsorry ildikov was on the phone12:17
* cdent catches up12:17
vrovachevildikov: Okay, I understand this :)12:17
ildikovcdent, eglynn-lunch: so I have my vote against it currently, but if we will consider this anyways, then I support cdent's idea about having a small scope12:18
ildikovcdent, eglynn-lunch: just sharing now as I will not be able to attend the meeting today sadly :(12:18
cdentIt looks like the proposer has already started to code and started with the collector instead of the notification agent12:18
ildikovvrovachev: ok, cool, tnx :)12:18
_nadya_ildikov: what change are you discussing? could you please give a link?12:19
_nadya_ildikov: or spec12:19
cdentre integration tests: I don't think tempest should stop doing multi-party integration testing or really much of what it is already doing, just that each project should also do more of their own testing, with critical partners12:19
ildikov_nadya_: hi, sorry for the late response :)12:19
cdentsome people think that will lead to duplication, but I think it will lead to more tests, which is not quite the same thing as duplication12:19
ildikov_nadya_: you mean now?12:19
cdentin other words there are many layers of integration and we can distribute them between tempest and us12:19
*** himangi has quit IRC12:20
_nadya_ildikov: aha, thanks12:20
ildikovcdent: that is why I asked if there will be a chance to move some tests to tempest later if there is a chance for having this distribution for instance12:21
vrovachevcdent: Question for delete ceilo tests on tempest is secondary, but I think that ceilo tests on tempest  should not be removed12:24
cdentI think that there are plenty of opportunities for tests to go in _both_ directions12:25
cdentbut we don't really need to decide what and when, yet. First we need to just get started with providing the in-tree structure that divides up unit from not-unit tests12:25
ildikov_nadya_: we have a meeting topic for this, or at least I kept it in the agenda as I remembered that we will discuss this today12:27
_nadya_tests for Ceilo should be in all test-related projects: Tempest, Rally,... do we have more:)?12:27
_nadya_ildikov: ok, great!12:28
_nadya_guys, haven't you discussed on the last meeting? Or anywhere else? I'd like to know what cores think about this implementation in general...12:28
ildikov_nadya_: it's on my todo list to review it in details12:29
ildikov_nadya_: I like the base idea to move the pipeline config into DB, but I need to look deeper in the actual implementation12:29
_nadya_ildikov: cool! As I understood, me and Gordon are a little bit confused about complexity :)12:30
ildikov_nadya_: well, I already should have... :(12:30
*** akrivoka has joined #openstack-ceilometer12:31
ildikov_nadya_: yeap, I remember those discussions on gerrit12:31
ildikov_nadya_: the DB based dynamic config idea should be achievable, but unfortunately I'm not the expert of DB and DB model optimization for this... :(12:33
ildikov_nadya_: so reviewing it will take a shorter while, but we should have more eyes on it for sure, so thanks for bring it up12:33
ildikov_nadya_: I will not be able to attend the today's meeting, but maybe a few words about this topic on the meeting would be good for having a bit more attention12:36
_nadya_ildikov: yep, I cannot understand why it's better to store config in db and "to decide the interval in which you want to evaluate pipeline changes". We can use the same "interval" and just load the file again in memory. Without db at all... I still think that I may miss something important12:37
akrivokais there a way to use substrings in a query? e.g. somethink like: ceilometer sample-list -m hardware.ipmi.current -q 'resource_id.startswith("blahblah")'12:38
ildikov_nadya_: it gives some flexibility to the way how the data collection process is configured12:39
ildikov_nadya_: like using an API endpoint and later having more fine grained config like based on projects or even lower than that12:40
eglynn-lunchakrivoka: regex matching is now supported, but only for "complex queries"12:41
eglynn-lunchakrivoka: see
akrivokaeglynn-lunch: thanks!12:42
_nadya_ildikov: I will think about it, thanks for pointing on API again12:43
ildikov_nadya_: cool, np12:43
cdentlinuxhermit: I've had some feedback from sdague, he thinks we should be commenting the gabbi yaml files much more. I've thought this too but was waiting for someone to say it (to not preload people's thinking).12:48
* cdent needs lunch12:51
*** rodrigods is now known as rodrigod`12:51
*** rodrigod` is now known as rodrigods12:53
linuxhermitlike explaining each test? cdent13:00
cdentpretty much linuxhermit13:02
cdent"because those might all seem obvious now about why, but as I'm sifting through 3 year old tests in the nova tree, lots of things are non obvious :)"13:03
*** _nadya_ has quit IRC13:04
*** jkraj has joined #openstack-ceilometer13:05
*** yatin has quit IRC13:06
*** ddieterly has joined #openstack-ceilometer13:06
linuxhermitcdent okay althought that is case with all tests13:09
linuxhermitand comment refactor rarely happens13:09
cdentI'm not entirely sold on the idea that it is needed, just something to think about13:10
cdent(that quote is from sdague, not me, in case it wasn't clear)13:10
cdentIf, for example there's a sequence being represented it might make sense to say so13:11
*** ddieterly has quit IRC13:11
cdentand for an especially long sequence it might make sense for it to be its own file13:11
linuxhermitoh like a process of create alarm send data destory alarm etc13:11
linuxhermitor workflow processes13:12
* cdent nods13:17
cdentbasically if one of the original goals of gabbi is to make the http api learnable and visible, the tests need to be readable sort of like a story13:18
cdentI know that I was better at that in the first few experiments, but as I've moved along I've become a bit sloppy, so need to remind myself of the point...13:18
cdentthe latest resources stuff I've done for gnocchi is too hairy13:19
linuxhermitI'll go back and comment up my events one13:20
ildikoveglynn-lunch: I need to run soon, my "feelings" about asyncio are above, as I will not be able to join the meeting today13:20
linuxhermitit doesn't have any workflow type things13:20
linuxhermitbut a good narative never hurts13:20
eglynn-lunchildikov: ok, thanks for heads-up13:21
*** eglynn-lunch is now known as eglynn13:21
ildikoveglynn: np13:23
*** moravec has joined #openstack-ceilometer13:26
openstackgerritChris Dent proposed stackforge/gnocchi: Cover the basic resource API with gabbi tests
*** ildikov has quit IRC13:38
*** ildikov has joined #openstack-ceilometer13:51
*** ildikov is now known as ildikov_afk13:51
*** ddieterly has joined #openstack-ceilometer13:51
*** ddieterly has quit IRC13:58
*** ddieterly has joined #openstack-ceilometer13:58
openstackgerritNelly Kuznetsova proposed stackforge/gnocchi: Driver for KairosDB with Cassandra
openstackgerritJason Myers proposed openstack/ceilometer: Adding Gabbi Tests to Events API
*** _nadya_ has joined #openstack-ceilometer14:08
openstackgerritMerged openstack/ceilometer-specs: Add specification for integration tests.
openstackgerritJason Myers proposed openstack/ceilometer: Adding Gabbi Tests to Events API
*** exploreshaifali has joined #openstack-ceilometer14:11
*** thomasem has joined #openstack-ceilometer14:13
*** changbl has quit IRC14:17
*** gordc has joined #openstack-ceilometer14:17
openstackgerritMerged stackforge/gnocchi: Add a py27-cover tox target
openstackgerritMerged stackforge/gnocchi: Remove a redundant existence check in resource PATCH
openstackgerritDina Belova proposed stackforge/gnocchi: [WIP] Driver for KairosDB with Cassandra
openstackgerritgordon chung proposed openstack/ceilometer: create a developer section and refactor
*** j_king_ is now known as j_king14:43
*** ccrouch has quit IRC14:44
gordccdent: i saw big lebowski. great movie.14:54
cdentI feel better about you now14:54
gordcgoodman's character is the best though.14:55
cdentHave you seen Barton Fink?14:56
gordcnever heard of it.14:57
*** eglynn-office has quit IRC14:58
gordcoh. i've seen the poster. (or image)14:58
gordcanother coen brothers film eh? i'll add it to the list.14:58
cdentmuch different style, but goodman is awesome in that too14:59
*** exploreshaifali has quit IRC14:59
*** prad has joined #openstack-ceilometer15:00
gordcnice... will have to check that out.15:00
cdentIf this were a slightly less public irc channel we could quote big lebowski all day15:01
gordclol yeah, i was going to start referencing stuff. than i realised i'd probably be booted15:05
*** ccrouch has joined #openstack-ceilometer15:06
*** ccrouch has left #openstack-ceilometer15:07
*** ccrouch has joined #openstack-ceilometer15:12
*** robsparker has joined #openstack-ceilometer15:13
*** yeungp has joined #openstack-ceilometer15:14
*** nellysmitt has quit IRC15:19
openstackgerritJason Myers proposed openstack/ceilometer: Adding Gabbi Tests to Events API
*** thomasem has quit IRC15:51
*** thomasem has joined #openstack-ceilometer15:57
*** thomasem has quit IRC15:57
*** thomasem has joined #openstack-ceilometer16:00
*** ildikov_afk is now known as ildikov16:01
ildikoveglynn: hi16:01
ildikoveglynn: I've just arrived back, I checked the logs and I'm satisfied with the outcome, so will check that branch to see what's going on :)16:02
*** mikedillion has joined #openstack-ceilometer16:03
gordcopen question to all: should we order list of traits returned with events when querying get_events?
openstackLaunchpad bug 1423714 in Ceilometer "tests wrongly assume trait ordering" [Medium,In progress] - Assigned to gordon chung (chungg)16:03
gordcie. should we fix test or fix code. i'm indifferent.16:03
*** himangi has joined #openstack-ceilometer16:03
*** himangi has quit IRC16:08
*** himangi has joined #openstack-ceilometer16:08
*** changbl has joined #openstack-ceilometer16:09
*** mikedillion has quit IRC16:11
*** himangi has quit IRC16:12
*** himangi has joined #openstack-ceilometer16:13
eglynnildikov: cool enough16:14
*** exploreshaifali has joined #openstack-ceilometer16:22
ildikovgordc: is there any easonable order that we could set as default?16:22
gordcildikov: trait key name. that's what mongo/es/hbase default to. it might be because that's what the test case assumes and those dbs were all written after test cases16:24
gordctbh, i'd probably lean more to the no order side but just barely.16:25
ildikovgordc: well, then I guess that could be the default behaviour if that is returned in case of most drivers already16:28
ildikovgordc: at least there will not be any bug about this in the future that whatever backend returns traits in different order than others16:30
ildikovgordc: but it's just an opinion :)16:30
gordcildiko: fair enough.16:31
silehtcdent, I have started looking on gabby, and wonder if you have a plan to inject default headers via os.environ or fixtures16:32
silehtcdent, I'm trying to see how I can provide for a keystone token16:32
silehtcdent, in live tests context16:33
cdentthere is a defaults key that can be in the yaml file16:33
cdentwhich forms the base dictionary for all the tests in that single file16:33
* cdent looks for example16:33
cdentanywhere url is not used in one of the tests, the default /cow?alpha=1 will be used16:34
cdentyou can put any valid keys in defaults16:34
cdentso a request_header x-roles or similar to be used by the whole file would be possible16:34
cdentdoes that make sense?16:34
*** david-lyle_afk is now known as david-lyle16:36
silehtcdent, I try to see how I can replace this line by the gabbi tests suite you have written:
silehtcdent, I don't want to rewrite tests, and the content of the token will change between each gate run16:36
cdentat the moment all the tests I've written have been to the internal gnocchi wsgi-app with keystone disabled16:37
cdentto use a real keystone token one option would be to read it in from disk at test run time (the harness would generate the token and save it somewhere known then the yaml file would read the token)16:38
cdenthowever, at the moment reading from disk is only supported for body data, not for headers but that could change16:39
cdenthowever, I'm not sure how much we want gabbi carting around keystone tokens16:39
silehtcdent, WDYT about adding a template thing that looks: $ENVIRON['KEYSTONE_TOKEN'] ?16:39
cdentthat would probably be okay16:40
cdentbut to what extent do we want to think of keystone as an integral part of API testing?16:40
cdentI've generally been on the side of leaving that out of it, but there are certainly ways to do it.16:40
silehtcdent, yes to have some tests check the acl stuffs16:41
cdentyou don't need a token to check the acl's do you? you need to pass in the headers that the token gets changed to by the middleware?16:41
cdentx-user-id, x-projecd-id, x-roles16:42
silehtthat's true16:42
cdentso whereas we want to do AuthZ in gabbi tests, doing AuthN isn't really required?16:42
*** himangi has quit IRC16:42
silehtyour right, i will make some check and probably disable keystone too16:44
silehtcdent, thanks16:44
cdentyou're welcome, it's great to see more people looking into it16:44
cdentI'm sure you will find things that are incomplete, please let me know and I'll fix what I can16:44
cdentthe $ENVIRON thing is a good idea but I'd like to hold off on adding it until something really requires it16:45
*** fnaval has joined #openstack-ceilometer16:45
*** himangi has joined #openstack-ceilometer16:46
*** pcaruana has quit IRC16:50
*** rwsu-afk is now known as rwsu16:50
openstackgerritgordon chung proposed openstack/ceilometer: order traits returned within events
openstackgerritMehdi Abaakouk proposed stackforge/gnocchi: [TEST] launch gabbi live tests
openstackgerritMehdi Abaakouk proposed stackforge/gnocchi: Add the nova shiny test output
silehtcdent, I have run all tests on an live gnocchi (apache+mod_wsgi) and I got 3 failures16:59
cdentI was just going to look at that, give me a sec to read it16:59
silehtcdent, I will investigate on them tomorow and review the new tests you have written17:00
linuxhermitcdent gordc can I bug you to look at again only change is more comments and removal of a duplicated test17:03
*** yatin has joined #openstack-ceilometer17:04
gordclinuxhermit: cool with me.17:05
*** lsmola has quit IRC17:06
*** Ala has quit IRC17:08
*** thomasem_ has joined #openstack-ceilometer17:08
*** thomasem has quit IRC17:11
*** thomasem_ has quit IRC17:12
*** akrivoka has quit IRC17:13
*** eglynn-office has joined #openstack-ceilometer17:14
*** thomasem has joined #openstack-ceilometer17:16
*** nellysmitt has joined #openstack-ceilometer17:20
*** ViswaV has joined #openstack-ceilometer17:20
*** _cjones_ has joined #openstack-ceilometer17:21
*** nellysmitt has quit IRC17:25
*** sbfox has joined #openstack-ceilometer17:28
*** ViswaV_ has joined #openstack-ceilometer17:28
*** ViswaV has quit IRC17:28
*** IvanBerezovskiy_ has quit IRC17:29
*** thomasem has quit IRC17:37
*** fabiog has joined #openstack-ceilometer17:38
*** karimb has quit IRC17:39
*** nellysmitt has joined #openstack-ceilometer17:41
*** karimb has joined #openstack-ceilometer17:43
*** zul has quit IRC17:43
*** karimb has quit IRC17:43
*** thomasem has joined #openstack-ceilometer17:46
*** kbyrne has quit IRC18:03
*** fabiog has quit IRC18:03
*** thomasem has quit IRC18:04
*** kbyrne has joined #openstack-ceilometer18:05
*** thomasem has joined #openstack-ceilometer18:08
*** zul has joined #openstack-ceilometer18:09
*** cdent has quit IRC18:34
*** _nadya_ has quit IRC18:38
*** yatin has quit IRC18:53
*** eglynn-office is now known as eglynn-officeafk18:54
*** exploreshaifali has quit IRC18:55
*** Ctina has joined #openstack-ceilometer18:55
Ctinahey i was wondering if someone had a second to help me with some potentially dumb ceilometer questions?18:56
CtinaI'm having trouble getting the cpu meter to show up, i started an instance and i see the disk.root.size, instance.scheduled, instance:<flavor> meters along with memory and vcpu, just not cpu or cpu_util18:57
openstackgerritJason Myers proposed openstack/ceilometer: Adding Gabbi Tests to Events API
*** prad has quit IRC19:01
*** eglynn has quit IRC19:08
gordcCtina: the meters you mentioned tend to come from notification messages that nova emits19:11
Ctinagordc: so do you think it sounds like i'm missing a config in my nova.conf?19:12
gordccpu meter comes from ceilometer's compute agent... by default that polls every 10 minutes so it may be because it hasn't actually polled yet.19:12
CtinaI updated the pipeline.yaml to poll every 60 seconds and i'm still not seeing anything19:13
gordcCtina: if you have disk.root.size it means you should be ok in regards to nova.conf. you may want to look at the ceilometer-acompute logs to make sure that is functioning properly19:13
gordcit seems like you're missing logs from polling agent and not the notification agent.19:14
gordcmissing meters*19:14
CtinaSo i see the polling for cpu and cpu_util in the ceilometer/compute.log actually19:14
Ctina2015-02-26 16:24:35.235 31966 INFO ceilometer.agent [-] Polling pollster cpu_util19:15
CtinaI'm running Icehouse if that helps19:17
Ctinait looks like those logs i'm seeing are about 3 hours old too..hmm19:17
gordcCtina: i see.. it could be your service is stopped. probably try restarting it and check if the logs are throwing any errors.19:19
Ctinai restarted the ceilometer-compute service and my ceilometer/compute.log wasn't updated at all19:21
Ctinamy alarm-evaluator just started throwing unauthorized errors, would that cause issues with the ceilometer-compute service?19:23
gordcno teh compute service can run separately from the alarm.19:24
gordcthe service should log some configuration steps at start up so it's strange you see nothing in compute log19:25
*** sbfox has quit IRC19:26
*** himangi has quit IRC19:26
gordcare you using vmware by any chance?19:27
*** sbfox has joined #openstack-ceilometer19:27
*** sbfox has quit IRC19:28
Ctinanope, not using vmware19:29
*** exploreshaifali has joined #openstack-ceilometer19:30
*** thomasem has quit IRC19:31
gordcok. i was just looking through backlog and i noticed there was a vmware related issue that hasn't been backported to icehouse... looking at automated builds there doesn't appear to be any noticeable issues with compute agent in icehouse19:32
Ctinaboo yeah this isn't devstack or anything. The last message in my ceilometer/compute.log is caught sigterm, exiting but it hasn't updated since i restarted the service :-\19:34
gordcsorry, i'm not really sure what is happening without logs. there might be some stray process that you might need to kill that is blocking compute agent from restarting19:44
openstackgerritJim Rollenhagen proposed openstack/ceilometer: Do not default pecan_debug to CONF.debug
Ctinaokay, at least I have an avenue to head down now, get the compute service logging! thank you for your help!!19:46
gordcnp. good luck19:47
*** ddieterly has quit IRC19:47
*** ddieterly has joined #openstack-ceilometer19:47
*** jkraj has quit IRC19:55
*** exploreshaifali has quit IRC19:56
*** Ctina_ has joined #openstack-ceilometer20:01
*** himangi has joined #openstack-ceilometer20:03
*** Ctina has quit IRC20:05
*** Ctina_ has quit IRC20:06
*** thomasem has joined #openstack-ceilometer20:07
*** nellysmitt has quit IRC20:09
*** himangi has quit IRC20:15
*** rbak_ has joined #openstack-ceilometer20:16
*** rbak has quit IRC20:16
dhellmannjroll: as a config change, does ^^ need a DocImpact tag in the commit message?20:21
jrolldhellmann: mmm, that's a good point, I don't know20:22
jrolldhellmann: I can't imagine that's something a deployer would want to enable, but maybe20:22
dhellmannjroll: well, it's more of a release-note thing, I think, since the default behavior is changing. Maybe ask annegentle?20:23
jrolldhellmann: don't see her online though20:24
dhellmannjroll: let's err on the side of caution and add the tag20:25
jrollyeah, just didn't want to slow it down unnecessarily20:26
* jroll does it20:26
openstackgerritJim Rollenhagen proposed openstack/ceilometer: Do not default pecan_debug to CONF.debug
dhellmannjroll: I think it's async to the code review process, so it shouldn't be an issue20:26
jrollgordc: can you re-approve that please?20:26
jrolldhellmann: I meant it going back to the end of the gate20:27
*** himangi has joined #openstack-ceilometer20:27
dhellmannjroll, gordc : done20:30
openstackgerritCharles Crouch proposed openstack/ceilometer: Fix a few typos in the docs
*** himangi has quit IRC20:33
ccrouchgordc: +1 on your last update to
ccrouchI'm so glad "Push Agents" section was removed :-)20:34
ccrouchit was kinda blowing my mind :-)20:34
*** himangi has joined #openstack-ceilometer20:40
gordcsorry, was on a call. thanks dhellmann20:46
gordcccrouch: :) yeah. i was hoping to redirect some ops to the docs we had. then i realised they were pretty patchy. ildikov and i are working on the main openstack docs so things should be better as we work on it.20:48
ccrouchgreat i'm looking through them too trying to come up to speed with more of the details, so I'll try to do some reviews on your changes20:49
gordcccrouch: cool cool. most of my changes are just cleanup/refactoring. thanks for the reviews.20:55
*** changbl has quit IRC21:21
*** sbfox has joined #openstack-ceilometer21:26
*** claudiub has joined #openstack-ceilometer21:45
openstackgerritgordon chung proposed openstack/ceilometer: create a developer section and refactor
*** karimb has joined #openstack-ceilometer21:58
openstackgerritMerged openstack/ceilometer: Adding Gabbi Tests to Events API
*** gordc has quit IRC22:04
*** rwsu has quit IRC22:06
*** claudiub has quit IRC22:16
*** rwsu has joined #openstack-ceilometer22:30
*** rodrigods is now known as rodrigod`22:42
*** rodrigod` is now known as rodrigods22:43
*** rodrigods is now known as rodrigod`22:51
*** rodrigod` is now known as rodrigods22:51
*** thomasem has quit IRC22:51
*** yeungp has quit IRC23:15
*** yeungp has joined #openstack-ceilometer23:15
openstackgerritMerged openstack/ceilometer: Do not default pecan_debug to CONF.debug
*** EmilienM is now known as EmilienM|afk23:49
*** ddieterly has quit IRC23:53
*** himangi has quit IRC23:54

Generated by 2.14.0 by Marius Gedminas - find it at!