Thursday, 2015-08-13

*** ankita_wagh has quit IRC00:15
*** vishwanathj has quit IRC00:16
*** jwcroppe has quit IRC00:52
*** jwcroppe has joined #openstack-ceilometer00:52
*** jwcroppe has quit IRC00:57
*** boris-42 has quit IRC00:58
*** mstyne has quit IRC00:59
*** tasdomas has quit IRC00:59
*** boris-42 has joined #openstack-ceilometer01:00
*** mstyne has joined #openstack-ceilometer01:02
*** tasdomas has joined #openstack-ceilometer01:04
*** liusheng has joined #openstack-ceilometer01:09
openstackgerritliusheng proposed openstack/aodh: Use the Serializer from oslo.messaging
*** slogan621 has joined #openstack-ceilometer01:11
*** drjones has quit IRC01:11
*** rjaiswal has joined #openstack-ceilometer01:16
*** thorst has joined #openstack-ceilometer01:34
*** jwcroppe has joined #openstack-ceilometer01:34
*** thorst has quit IRC01:55
*** rohit has joined #openstack-ceilometer02:06
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometermiddleware: Updated from global requirements
openstackgerritliusheng proposed openstack/aodh: Introduce Guru Meditation Reports into Aodh
openstackgerritliusheng proposed openstack/aodh: Introduce Guru Meditation Reports into Aodh
openstackgerritgordon chung proposed openstack/ceilometer: do not translate debug logs
*** jwcroppe has quit IRC02:50
*** jwcroppe has joined #openstack-ceilometer02:51
*** jwcroppe has quit IRC02:55
*** ankita_wagh has joined #openstack-ceilometer02:55
*** rjaiswal has quit IRC03:20
*** ankita_wagh has quit IRC03:37
*** rbak has joined #openstack-ceilometer03:48
*** rbak has quit IRC03:49
*** rohit has quit IRC04:10
*** fawadkhaliq has joined #openstack-ceilometer04:32
*** Ephur has quit IRC04:36
*** fawadkhaliq has quit IRC04:39
*** fawadkhaliq has joined #openstack-ceilometer04:39
*** fawadkhaliq has quit IRC04:57
*** yprokule has joined #openstack-ceilometer04:57
*** fawadkhaliq has joined #openstack-ceilometer05:06
*** hhuang has joined #openstack-ceilometer05:36
*** hhuang has quit IRC05:38
*** hhuang has joined #openstack-ceilometer05:39
*** ildikov has quit IRC05:42
*** nadya has quit IRC05:44
*** nadya has joined #openstack-ceilometer05:58
*** nadya has quit IRC06:12
*** ildikov has joined #openstack-ceilometer06:31
openstackgerritliusheng proposed openstack/aodh: Avoid translating debug log
openstackgerritliusheng proposed openstack/aodh: Avoid translating debug log
*** afazekas has joined #openstack-ceilometer06:40
*** nadya has joined #openstack-ceilometer06:44
*** cdent has joined #openstack-ceilometer07:22
*** therve has quit IRC07:25
*** therve has joined #openstack-ceilometer07:25
*** changbl has quit IRC07:41
*** openstackgerrit has quit IRC07:41
*** tobias- has quit IRC07:41
*** csd has quit IRC07:41
*** changbl has joined #openstack-ceilometer07:42
*** openstackgerrit has joined #openstack-ceilometer07:42
*** tobias- has joined #openstack-ceilometer07:42
*** csd has joined #openstack-ceilometer07:42
*** csd has quit IRC07:45
*** openstackgerrit has quit IRC07:45
*** changbl has quit IRC07:46
*** tobias- has quit IRC07:46
*** tobias- has joined #openstack-ceilometer07:48
*** csd has joined #openstack-ceilometer07:52
*** openstackgerrit has joined #openstack-ceilometer08:00
*** changbl has joined #openstack-ceilometer08:00
*** yassine has joined #openstack-ceilometer08:04
openstackgerritMerged openstack/gnocchi: sqlalchemy: use DBReferenceError to generate the correct exception
openstackgerritMerged openstack/gnocchi: indexer: remove wrong FK catch
*** Kennan has quit IRC08:14
*** Kennan has joined #openstack-ceilometer08:21
openstackgerritliusheng proposed openstack/aodh: Introduce Guru Meditation Reports into Aodh
openstackgerritNadya Shakhat proposed openstack/ceilometer: Make it possible to run postgresql functional job
*** ankita_wagh has joined #openstack-ceilometer08:42
*** rodrigods has quit IRC09:13
*** rodrigods has joined #openstack-ceilometer09:13
*** jwcroppe has joined #openstack-ceilometer09:28
openstackgerritMerged openstack/ceilometer: remove alembic requirement
*** ankita_wagh has quit IRC09:41
*** exploreshaifali has joined #openstack-ceilometer09:46
openstackgerritMerged openstack/ceilometer: support multiple-meter payloads
openstackgerritMerged openstack/ceilometer: Convert instance, bandwidth and SwiftMiddleware meters
openstackgerritMerged openstack/ceilometer: start rpc deprecation
cdentjd__: gnocchi indexer structure question for you if you're about09:53
jd__cdent: shoot09:53
cdentI'm rerunning my tests from that document I pointed you at, with just one metricd process and I noticed something that seems funny in the metric table:09:54
cdentafter a few minutes of the 5163 rocs in the metric table only 442 have a resource_id that is not NULL09:54
cdentis that expected? What are all these disassociated metric entries?09:54
cdent(BTW, though a bit slower, it's fine with just one metricd process)09:55
openstackgerritNadya Shakhat proposed openstack/ceilometer: Make it possible to run postgresql functional job
*** boris-42 has quit IRC10:00
cdentjd__: may have been a disk space failure, will rerun10:00
jd__cdent: oh cool :)10:03
jd__cdent: are you using MySQL?10:03
openstackgerritChris Dent proposed openstack/aodh: Use generic keystone uri in devstack config
cdenthowever even if it s diskspace faiure it still seems a strange error to have happen10:04
cdentyeah, only because it reflects the defaults10:04
jd__well so it's not that strange10:04
jd__that's another good reason we recommend PostgreSQL10:04
cdentI agree the postgres is better, but even so10:04
cdentnot the sort of behavior we want to countenance really10:05
jd__that's not something that could happen with PG on a disk full10:05
jd__I mean we can't do anything about that in Gnocchi if it's really a problem with the RDBMS10:05
jd__unless we have a bug of course10:05
cdentI'll run it again and see10:06
cdentokay jd, fresh database, disk space available, within seconds of booting ten instances:10:22
cdentmetric count is 250310:23
cdentnull metric count is 218110:23
cdentnull meanning "no associated resource_id"10:23
cdentjd__: ^10:23
cdentwill now switch to postgres10:23
jd__ok, I need to check how the dispatcher work10:23
cdentI tried to figure that out earlier in the week and went a bit crazy10:24
jd__cdent: if you stop and wait for the collector to finish,you still have orphaned metrics?10:24
cdentlemme see10:24
jd__glancing at the dispatcher code I'm not sure it works correctly if resource_id is None – which should not happen but we never know10:28
jd__but that could be a theory of why you have orphaned metrics10:28
cdentthe collector and metricd are all caught up now and the counts are: 2833 total, 2391 no resource_id10:28
cdent13 rows in the resource table10:29
openstackgerritChris Dent proposed openstack/ceilometer: Remove version from os_auth_url in service_credentials
cdentjd__: this makes the aodh functional jobs voting, if you feel like we're ready for such things:
jd__cdent: what are the names of all those metrics? 13 resources and 2800 metrics?10:38
jd__I think the dispatcher is on a spree10:38
cdentno names either10:38
jd__you have gnocchi-api logs that shows all those POST?10:38
cdentshould do, looking10:39
cdentI've got 7749 POST /metric/v1/resource/instance with a 400 response10:42
cdentthe collector has this:
cdent(lots of that)10:43
cdentso that clarifies things a bit10:43
cdenton the dispatcher side we're still sending metrics for resources with ids we can't store10:44
cdentI'm guessing?10:44
cdentthis is something about the dispatcher logic that I think is backwards: it makes a resource after it first does some meters10:45
cdentthis strikes me as backwards10:45
cdentif we're never going to store the resource we should drop the measures on the floor10:45
* cdent coffees up10:45
jd__cdent: that smells that indeed10:46
jd__but we're supposed to store all the resource10:47
jd__that'd make a sticker10:47
EmilienM"just don't use mongodb"10:47
EmilienMthat could also be a sticker10:47
jd__so yeah it's a bug in the dispatcher as I expected10:47
cdentare we supposed to store all the resources? what do we do about id translation?10:50
cdentunfortunately no mongodb involved here10:51
cdentbut yeah, it is dire10:51
EmilienMgood to know10:51
cdentliusheng: in the bug you just reported is there a step missing: "wait for a polling cycle to pass"?10:53
cdentor something like that? I can't quite figure out what you're saying10:53
*** exploreshaifali has quit IRC11:00
jd__cdent: we are supposed to store all the resources but they should have an UUID as an id11:03
jd__cdent: I think some such as the NIC of instance have no UUID but something weird11:03
cdentthere are _tons_ of notifcations, polling info, etc that have resource_id set to something pretty far off from uuid11:03
cdentheat stacks, for example11:03
openstackgerritMerged openstack/aodh: Use the Serializer from oslo.messaging
*** eglynn has quit IRC11:15
*** dikonoor has joined #openstack-ceilometer11:30
*** dikonoo has joined #openstack-ceilometer11:30
*** dikonoor has quit IRC11:33
*** thorst has joined #openstack-ceilometer11:35
*** annasort has quit IRC11:42
*** ajc_ has joined #openstack-ceilometer11:43
*** ajc_ has quit IRC11:43
dikonooidegtiarov: Hi11:48
openstackgerritgordon chung proposed openstack/ceilometer: add test to validate jsonpath
openstackgerritMerged openstack/ceilometer: do not translate debug logs
dikonooidegtiarov: This is about your comment in Do you know if the current behaviour (that lets only admin acess events) documented somewhere ? IF yes, could you point me to it11:49
dikonooidegtiarov : If the current behavior is not documented, should the change ine behavior be documented. I had discussed this with gordc  and he was of the opinion that it isn't required as the existing behavior isn't documented11:50
cdentjd__: where did that earlier conversation go? do we need to make a bug or something?12:00
jd__cdent: I'd make a bug against Ceilometer because it's a problem with the dispatcher doing wrong thing if the resource_id is rejected12:01
jd__cdent: if you're willing to test I can try to make a patch I guess12:01
openstackgerritJulien Danjou proposed openstack/ceilometer: polling: remove deprecated agents
cdentjd__: okay, will make bug, trying to capture the sitch12:02
openstackgerritDivya K Konoor proposed openstack/ceilometer: Add user_id,project_id traits to audit events
openstackgerritgordon chung proposed openstack/ceilometer: add mandatory limit value to complex query list
*** gordc has joined #openstack-ceilometer12:13
*** liusheng has quit IRC12:15
*** liusheng has joined #openstack-ceilometer12:15
liushengcdent: Hi, no, I have set the polling interval to 20s12:17
liushengcdent: it cannot works12:17
liushengcdent: sorry, I was AFK just now12:21
*** liusheng_ has joined #openstack-ceilometer12:29
*** liusheng has quit IRC12:30
*** annasort has joined #openstack-ceilometer12:30
*** annasort_ has joined #openstack-ceilometer12:31
cdentgordc: you already approved this once before, but I updated the config to remove a depends-on, can you go ahead and reapprove12:33
cdentthat will get us one step closer on the grenade plugin front:
*** annasort has quit IRC12:35
*** annasort_ is now known as annasort12:35
gordccdent: tests12:35
gordcnot tests but testing* irc12:36
gordcnm. gerrit works now12:36
cdentgordc you never making any sense to me any more, I think we need counselling12:36
gordcwas checking if my internet crapped out or just gerrit12:36
cdentwhen the opportunity presents itself to blame gerrit, blame gerrit12:37
gordci'm not willing to work on it. i say no counseling!12:37
gordcnow bring me a beer!12:37
cdentthis relationship is so over12:37
gordc50-50 rule exists in uk?12:38
gordcyou can take half my suit. i'll take half your home by the beach.12:38
dikonoogordc: hi12:40
gordccdent: will this break your grenade?
dikonoogordc: this is about
gordcdikonoo: morning12:40
gordcdikonoo: sure. i have 10 mins before i drive to work... think it can be done?12:40
cdentgordc why would it?12:41
dikonoogordc: yeah.. first on the documentation comments I have been getting..Of course you didn't comment on that ..but does it have a doc impact or not12:41
cdentoh i see, that commit message is bad12:41
gordccdent: it drops the old compute/central agent12:41
dikonoogordc: Also, do you know where I can find the documentation if I have to change it12:41
cdentyes, it will, i'll comment12:41
dikonoogordc: second, on
gordcdikonoo: um... i'm not sure we have a place which explicitly says admin is required... but you should add DocImpact to your commit message and someone will follow up on it12:42
dikonoogordc: ok..12:43
dikonoogordc: can you point me to a sample12:43
gordcjust add "DocImpact"12:44
dikonoodikonoor: Ah..ok..Thanks..12:45
gordcit will auto create a bug in openstack-manuals12:45
dikonoogordc: on the test cases, I don't think / understands RBAC ..For eg. I tried changing the existing header at to non-admin hoping that will break the existing testcases12:47
dikonoogordc: The existing /events allow only admins..So when the X-Role in the header is not admin, the testcase should break..but it always passes12:47
gordcdikonoo: uh... well that sucks.12:49
dikonoogordc: The configuration of the underlying pecan.test make_app initialization sets the policy_file to /etc/ceilometer/policy.json . I tried overiding it with new file, adding new rules..different permutation & combinations but it always passes..the tests never break12:49
gordci can try looking at it later today.12:49
gordcdikonoo: sorry, i need to step out right now. i'll be back in a bit. feel free to email me or mailing list if you need to drop12:50
dikonoogordc: yeah..that will be great.. Thanks..Please leave me a note somewhere in the review if you don;t see me around12:51
*** gordc has quit IRC12:51
*** fawadkhaliq has quit IRC12:52
cdentjd__: I updated my lame vaguebug to be more associated with the thing we talked about this morning (orphaned measures and bad resource ids):
openstackLaunchpad bug 1483634 in Ceilometer "gnocchi dispatcher still sending far too many resource updates" [High,Triaged] - Assigned to Chris Dent (chdent)12:54
cdentaodh functional test just went gaiting12:55
cdentlets see if goes boom!12:55
jd__cdent: ok cool I'll try to cook a patch :)12:59
*** openstackgerrit has quit IRC13:01
*** openstackgerrit has joined #openstack-ceilometer13:01
*** jasonamyers has joined #openstack-ceilometer13:03
*** liusheng_ has quit IRC13:04
*** edmondsw has joined #openstack-ceilometer13:05
*** cmyster has joined #openstack-ceilometer13:06
*** cmyster has joined #openstack-ceilometer13:06
*** fawadkhaliq has joined #openstack-ceilometer13:13
*** julim has quit IRC13:19
*** cmyster has quit IRC13:26
*** rjaiswal has joined #openstack-ceilometer13:30
*** jwcroppe has quit IRC13:37
*** jwcroppe has joined #openstack-ceilometer13:37
*** prad_ has joined #openstack-ceilometer13:37
*** gordc has joined #openstack-ceilometer13:41
*** jwcroppe has quit IRC13:42
nadyacdent: hi! what do you think about it?
nadyacdent: I've realized that set DEVSTACK_GATE_POSTGRES in post_test_hook is too late13:43
nadyacdent: it shouldn't break our functional-mongo job I believe13:44
cdentnadya: looking13:45
*** julim has joined #openstack-ceilometer13:46
cdentI think that makes sense. the other option is to wait a few days for the devstack plugin to be active13:47
openstackgerritDivya K Konoor proposed openstack/ceilometer: Events RBAC needs scoped token
cdentbut that will probably still be a while13:48
*** jwcroppe has joined #openstack-ceilometer13:50
*** exploreshaifali has joined #openstack-ceilometer13:50
cdent(this infra stuff is so confusing)13:50
cdentnadya: have you been able to figure out what the GATE_POSTGRES thing actually does?13:52
cdent(I haven't)13:52
nadyacdent: as I understand it means smth like "install postgres when devstack is being deployed"13:53
cdentthat's what I've assumed too, but I've never been clear on where that's actually happening13:53
gordcdikonoo: i'm going to try tests out now13:56
*** rbak has joined #openstack-ceilometer13:57
*** thorst has quit IRC14:03
*** afazekas has quit IRC14:03
*** annasort has quit IRC14:03
*** thorst has joined #openstack-ceilometer14:11
openstackgerritDivya K Konoor proposed openstack/ceilometer: Add user_id,project_id traits to audit events
*** rbak has quit IRC14:13
*** devlaps has quit IRC14:24
*** rbak has joined #openstack-ceilometer14:25
*** exploreshaifali has quit IRC14:25
*** ocaner has joined #openstack-ceilometer14:28
openstackgerritDivya K Konoor proposed openstack/ceilometer: Events RBAC needs scoped token
openstackgerritDivya K Konoor proposed openstack/ceilometer: Control Events RBAC from policy.json
*** dikonoo has quit IRC14:48
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Adds support for dynamic event pipeline
*** eglynn has joined #openstack-ceilometer14:55
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Fixed event requeuing/ack on publisher failure
gordceglynn: the clarification on alarm deprecation is from last week's meeting correct?14:57
eglynngordc: yep14:58
gordceglynn: cool cool14:58
eglynngordc: (I didn't add anything new to the agenda for today)14:58
gordceglynn: ok. just wanted to check in case it was re-add... was just me being lazy and not cleaning last week14:59
eglynncool :)14:59
*** Ephur has joined #openstack-ceilometer15:02
gordcfor those interested... meeting is starting15:02
*** Ephur has quit IRC15:06
*** thorst has quit IRC15:12
openstackgerritRyota MIBU proposed openstack/aodh: Storage: add 'exclude' constraint to get_alarms()
openstackgerritChris Dent proposed openstack/ceilometer: integration: chown ceilometer directory properly
cdentgordc, jd__: if that goes through the integration job should start working15:21
*** kfox1111 has left #openstack-ceilometer15:22
*** hhuang has quit IRC15:23
*** haigang has joined #openstack-ceilometer15:24
cdentor at least get to the next break15:24
*** haigang has quit IRC15:24
*** haigang has joined #openstack-ceilometer15:25
*** vishwanathj has joined #openstack-ceilometer15:29
nadyagordc: please take a look after . I'm afraid of making changes in voting job but it looks rather safe15:30
openstackgerritMerged openstack/ceilometer: Updated from global requirements
openstackgerritMerged openstack/ceilometer: Remove version from os_auth_url in service_credentials
openstackgerritMerged openstack/ceilometer: Grenade plugin using devstack plugin for ceilometer
*** jwcroppe has quit IRC15:44
*** jwcroppe has joined #openstack-ceilometer15:45
openstackgerritRyota MIBU proposed openstack/aodh: Refactor api tests (_update_alarm)
*** haigang has quit IRC15:49
*** haigang has joined #openstack-ceilometer15:49
*** thorst has joined #openstack-ceilometer15:56
*** _cjones_ has joined #openstack-ceilometer15:59
*** _cjones_ has quit IRC15:59
*** _cjones_ has joined #openstack-ceilometer15:59
openstackgerritIgor Degtiarov proposed openstack/ceilometer: Avoid from storing samples with empty or not numerical volumes
*** hhuang has joined #openstack-ceilometer16:01
gordccdent: i have no idea what the danish reference is16:01
cdentjust looked danish to me16:01
gordci'm in americas... we do americas...16:01
* gordc steps out... giving a fun presentation on ceilometer16:02
*** Ephur has joined #openstack-ceilometer16:02
*** yassine has quit IRC16:04
*** ocaner has quit IRC16:05
*** edmondsw has quit IRC16:06
gordchmm. jumped the gun by 30mins16:08
cdentgordc: you're overwhelmed by all of this and plus your divorce from me. I think even if you aren't willing to do that therapy with me, you need to consider it for yourself16:10
cdentor at least a visit to the spa16:10
gordchahahahah! i've been told spas are nice.16:10
*** alexpilotti_ has joined #openstack-ceilometer16:11
*** haigang has quit IRC16:15
*** hhuang has quit IRC16:18
*** jasonamyers has quit IRC16:18
*** alexpilotti_ is now known as alexpilotti16:19
*** alexpilotti has quit IRC16:19
*** alexpilotti has joined #openstack-ceilometer16:19
*** alexpilotti has quit IRC16:19
*** alejandrito has joined #openstack-ceilometer16:22
openstackgerritgordon chung proposed openstack/ceilometer: fix event tests to use rbac properly
*** edmondsw has joined #openstack-ceilometer16:30
*** openstackgerrit has quit IRC16:31
*** openstackgerrit has joined #openstack-ceilometer16:31
*** eglynn has quit IRC16:32
*** exploreshaifali has joined #openstack-ceilometer16:35
*** ildikov has quit IRC16:36
*** ankita_wagh has joined #openstack-ceilometer16:36
*** ankita_wagh has quit IRC16:37
*** ankita_wagh has joined #openstack-ceilometer16:41
*** nadya has quit IRC16:53
*** yprokule has quit IRC17:03
*** boris-42 has joined #openstack-ceilometer17:11
*** vishwanathj has quit IRC17:15
*** jasonamyers has joined #openstack-ceilometer17:22
*** nijaba has joined #openstack-ceilometer17:36
*** nijaba has quit IRC17:36
*** nijaba has joined #openstack-ceilometer17:36
*** nijaba_ has quit IRC17:36
*** fawadkhaliq has quit IRC17:42
*** nadya has joined #openstack-ceilometer17:42
*** ankita_w_ has joined #openstack-ceilometer17:44
*** ildikov has joined #openstack-ceilometer17:46
*** Ephur_ has joined #openstack-ceilometer17:46
*** ankita_wagh has quit IRC17:47
*** srsakhamuri has quit IRC17:47
*** Ephur has quit IRC17:48
*** Ephur has joined #openstack-ceilometer17:50
*** Ephur_ has quit IRC17:51
openstackgerritgordon chung proposed openstack/ceilometer: fix event tests to use rbac properly
cdentgordc: it was nice of you to +A that chown thing but the integration job is still not ready to fly17:55
cdentthere will need to be yet another followup patch:17:55
cdenthorizon is hogging apache making gnocchi too hard to find17:55
cdenti'm looking into it now17:55
*** fawadkhaliq has joined #openstack-ceilometer17:56
david-lylecdent: what do you mean by hogging?17:56
david-lylejust curious17:56
gordccdent: should i remove +a?17:57
* gordc wonders if david-lyle has a script to track 'horizon'17:57
cdentdavid-lyle: I've not got all the details yet, but if horizon is on port 80 with mod_wsgi and another service also wants to be on port 80 at a different path, horizon existing config gets in the way17:58
gordcwow really? that's actually very efficient.17:58
david-lylegordc: my client handles it, so it's easy17:58
cdentat some point in the past I made gnocchi default to running on port 80 with prefix in devstack17:58
cdentbut apparently all my testing was with horizon disabled17:58
david-lylecdent: ok that's why I asked, I have a patch that will interest you17:59
cdentso now I'm trying to determine if it is possible to get horizon a bit more contained17:59
david-lyleit's blocked by a tempest change, but once that's the situation should be better17:59
*** fawadkhaliq has quit IRC18:00
*** fawadkhaliq has joined #openstack-ceilometer18:00
gordchmm... timing was oldly convenient18:00
david-lylewe had a conflict with keystone too18:01
*** ankita_wagh has joined #openstack-ceilometer18:02
david-lylesadly Horizon has been getting to this point for about 6 months18:02
cdentthat's just awesome18:02
david-lylefinally ready to land18:02
david-lyleso, hopefully I saved you some time18:02
cdentyes, very much so18:02
cdentso yeah, gordc, when that lands it looks like the integration stuff will inch one step forward18:03
cdentprobably producing yet another bug to solve18:03
* david-lyle goes back to lurking18:03
cdentthanks for the quick whatever that was david-lyle18:03
cdentvery helpful18:03
gordccdent: so leave it +a?18:04
cdentyeah, may as well18:04
*** ankita_w_ has quit IRC18:04
cdentwe'll just fix whatever the next problem is when it comes along18:04
rjaiswalgordc: replied to your comment on, please take a look18:06
*** ankita_w_ has joined #openstack-ceilometer18:06
cdentcheers thanks a lot gordc18:06
*** idegtiarov_ has joined #openstack-ceilometer18:08
prad_gordc, are you looking into making the multi meter logic fit other optional cases? or should i? I think i'll need that for cpu as well. where type and unit are not part of the payload and staticly defined18:08
*** nadya has quit IRC18:08
cdentoh this reminds me18:09
prad_gordc, so basically instead of we might want to do that look up based on the fields in multi option?18:09
*** ankita_wagh has quit IRC18:09
cdentgordc: what prad is saying recalls a conversation I had with him earlier today where I was trying to push the idea that the declarative syntax should stay simple and the complex things ought to be left in code as signals that they are dirty and wrong and should be shunned back to the projects from whence they came18:09
gordcrjaiswal: sure. sorry. been busy with a presentation. will get to it.18:10
rjaiswalgordc: thanks18:10
prad_gordc, in cpu case i need that for colume and name, but set the type and unit based on whats statically defined in yaml18:10
gordcprad_: yeah i think that was my plan18:10
gordccdent: i think that might be what is required.18:11
cdentit should be that we are doing it because it is too hard to do it otherwise18:11
prad_yea there are some that are quite wacky to represent in yaml18:11
cdentbut because clean syntax in the yaml is the end goal18:11
gordccdent: that said, the multi meter stuff isn't actually dirty/wrong... it's just presents the data differently18:11
cdentand being the yaml is cool18:11
cdentso if your stuff can't go in the yaml you must be doing it completely wrong and should have a rethink18:12
prad_i think muti meter is fine18:12
* cdent shrugs18:12
prad_what cdent might be referring to here is wacky match expressions like the ones in trove and dns18:12
cdentmostly that, yes prad_18:12
cdentbut generally speaking I think that the things producing the notifications should be held responsible for quality, not the other end18:13
gordcprad_: s/match/math/?18:13
prad_match in this case.. the timeutils delta is a python expression rather than a pure math18:13
gordcprad_: hmmm... it's definitely not straight forward...18:16
*** fawadkhaliq has quit IRC18:16
gordci wonder if we should be supporting some sort of plugin mechanism like we have in events which can derive values18:16
gordcthe time delta shouldn't be difficult and seems like a useful use case.18:16
prad_it might be easier to just handle it as a python expression in yaml perhaps .. do a !!python/ . i also looked at pyparsing that works well for straight math expressions18:18
cdentit's okay for there to be both kinds of meters: the easy ones in yaml18:18
cdentand the code ones18:18
prad_i agree, our motivation should be to make yaml look easy to use18:18
cdentthe code ones should just be python like they've always been18:19
prad_ it would be nice to convert as much as possible to keep everything in one place .. but i agree if its just wacky to convert may be best to leave it or find an alternative like move it to a transformer or something18:20
gordc simple enough no? a function call?18:20
gordcthe ipmi stuff can burn18:21
gordcbut i feel like the delta stuff might be manageable...meh18:21
gordccdent: that patch you did for ipmi notifications does some crazy ass sh*t18:22
cdentthat's because the input was crazy18:22
cdentbut remind me, gotta link?18:22
gordccdent: yeah i saw the input18:23
gordcand said f it18:23
gordci don't have patch... i don't think there's any way to fix it.18:23
cdentthat was back in the early days when I had no idea what I was doing and was getting a lot of pressure18:24
cdentnow I still have no idea what I'm doing and disregard the pressure18:24
gordclol nah, i saw the payload... it's craziness.18:25
gordci was going to go to ironic and ask why it was the way it was... but then decided to drive home.18:26
cdentthe long way18:28
gordcgas is still expensive... the longer than normal way18:30
*** jasonamyers has quit IRC18:38
*** jasonamy_ has joined #openstack-ceilometer18:38
*** exploreshaifali has quit IRC18:49
*** ankita_wagh has joined #openstack-ceilometer18:52
*** ankita_w_ has quit IRC18:55
*** jasonamyers has joined #openstack-ceilometer18:55
*** jasonamyers has quit IRC18:59
*** jasonamy_ has quit IRC18:59
*** fawadkhaliq has joined #openstack-ceilometer19:01
*** fawadkhaliq has quit IRC19:06
*** exploreshaifali has joined #openstack-ceilometer19:12
*** nadya has joined #openstack-ceilometer19:14
*** cdent has quit IRC19:25
*** kevinc_ has joined #openstack-ceilometer19:30
*** nadya has quit IRC19:33
*** alejandrito has quit IRC19:44
openstackgerritMerged openstack/ceilometer: integration: chown ceilometer directory properly
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometermiddleware: Updated from global requirements
gordcprad_: hmm. we need away to customise metadata20:16
prad_gordc, yea we could add attributes: with metadata we specifically want and filter out before passing sample generation20:18
prad_most cases we just store the whole payload from what i saw20:19
gordcso should we add a new attr?20:20
openstackgerritOpenStack Proposal Bot proposed openstack/python-ceilometerclient: Updated from global requirements
prad_gordc, we can keep it optional? and only look up if its defined if not we use the whole payload20:21
gordcsure. want me to include that in my patch or you want to do it separately?20:22
gordci don't have a preference20:23
prad_do you need it right away?20:23
prad_i can add it tomorrow morning20:24
prad_or later tonight if i find time20:24
*** nadya has joined #openstack-ceilometer20:25
gordcnot right away... just that i can't properly drop telemetry notification without it20:25
gordci'll work without it for now.20:26
gordcand then rebase on whatever you have20:26
nadyagate-ceilometer-dsvm-functional-postgresqlSUCCESS in 41m 02s :) I've almost managed with it20:27
* gordc does maths.20:27
gordcnadya: what time is it?lol20:27
nadya:) time to sleep20:27
gordci have a random question for you tomorrow... why is it so slow in gate?20:27
gordcbut yay to postgres!20:28
*** nadya has quit IRC20:28
*** alejandrito has joined #openstack-ceilometer20:28
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Add support to compute cpu meters in yaml
prad_gordc, i modified the multi meter logic a little to get the above working for cpu meters20:36
prad_we still need to make it more extendable20:37
prad_gordc, i changed the yaml a bit as well
gordcwhat's separator?20:45
gordci don't think it's right. your meter definitions are acting on the same values but somehow giving different data/units20:47
*** fawadkhaliq has joined #openstack-ceilometer20:48
prad_gordc, which part?20:51
prad_gordc, hmm i ran through the test_cpu.py20:52
*** julim has quit IRC20:52
prad_gordc, so you're saying its not generating all the meters?20:53
gordcprad_: no. it's generating too many20:55
gordcit doesn't deal with the conditional.20:55
gordcso both definitions will match all meters... and it'll build one set with unit=ns and another set with unit=%20:55
gordcyou need to make your jsonpath more restrictive.20:56
prad_ah i misread the original meters.. i thought there were two sets, what i did was intentional20:59
prad_gordc, good catch20:59
gordcprad_: i'd tell you the jsonpath to check sibling values... but i don't know it without googling21:00
*** thorst has quit IRC21:14
*** vishwanathj has joined #openstack-ceilometer21:17
*** idegtiarov_ has quit IRC21:19
*** slogan621 has quit IRC21:21
gordchmmm... maybe we shouldn't be declaring the telemetry notifications21:22
gordcif someone deletes/overrides it, everything goes to hell.21:23
gordcprad_: ^21:23
prad_gordc, well that could be said for others? maybe we put a note at the top saying: if you delete any of these meter you're screwed ?21:24
prad_gordc, is it getting too messy to convert?21:24
gordcprad_: it's pretty straight forward... it's just that you need to map message_id, and source, and timestamp21:25
prad_i dont really see a way to check for siblings .. donno what Desendents are trying that21:25
gordcthese particular are something i don't think we should surface... it's internal to ceilometer... and if you delete one, you delete all polling/api post21:26
*** ankita_wagh has quit IRC21:26
prad_yea i'm cool with leaving it .. i dont see much benefit to the user exposing telemetry ones anyway21:27
*** ankita_wagh has joined #openstack-ceilometer21:27
gordci'll add support for mutliple everything... (minus message_id, source, timestamp)... but yeah i don't think we should expose it now that i think abou it.21:27
prad_what are they going to tweak there anyway21:27
prad_yea thats fine21:27
gordcprad_: yeah, any check from user will explode something.21:27
gordcyeah i think the bigger problem is if they override meter.yaml they need to know to have it in their override version... and they won't.21:29
gordclet's leave it!21:29
*** ankita_wagh has quit IRC21:31
kevinc_It appears that ceilometer polls swift and glance on a regular interval for object and image usage (storage.object.size and image.size). However volume.size seems to be based on events. Is there any way for ceilometer to poll cinder on a regular basis to collect usage?21:37
gordckevinc_: i believe cinder generates metrics periodically.21:38
gordclet me check21:38
kevinc_gordc: Thanks! being able to get volume usage stats it the last piece of the puzzle that I need to be able to bill users21:39
*** alejandrito has quit IRC21:41
*** ankita_wagh has joined #openstack-ceilometer21:42
gordci'm not sure if this is what you were looking for.21:43
*** ankita_wagh has quit IRC21:43
prad_gordc, the thing with unit lookup is, its defined statically in yaml, its not part of the message.. so there is nothing to look for, we'll have to do it outside of jsonparse21:43
*** ankita_wagh has joined #openstack-ceilometer21:43
gordckevinc_: if that doesn't give you the info you need, we'll need to patch either ceilometer or cinder. not sure if it'll happen this late in development cycle21:45
gordcprad_: hmmm.. i think you might be able to... it should be fine to define static value in jsonpath.21:46
gordci need to think about it... after i write this multi everything code21:46
kevinc_gordc: Thank you! I will take a look at that. Basically we want to get stats so we can charge users by the GB/hour for their cinder volume usage21:47
gordckevinc_: that should work but feel free to raise a bug if not21:47
prad_ i dont see a way to do as part of the tree look up but we could just pass the static unit as part of the name path perhaps .. i'll poke around21:48
gordcprad_: kk, i'll let you know if i find something21:48
kevinc_gordc: Thank you! I think that will do what I need. It looks like the admin has the cron job configured incorrectly.21:53
gordckevinc_: cool cool.21:54
*** alexpilotti has joined #openstack-ceilometer22:04
*** exploreshaifali has quit IRC22:09
openstackgerritgordon chung proposed openstack/ceilometer: full multi-meter support
*** gordc has quit IRC22:25
*** rbak_ has joined #openstack-ceilometer22:31
*** rbak has quit IRC22:32
*** julim has joined #openstack-ceilometer22:32
*** edmondsw has quit IRC22:48
*** alexpilotti has quit IRC22:49
*** rbak_ has quit IRC22:58
*** prad_ has quit IRC23:08
*** jwcroppe has quit IRC23:42
*** jwcroppe has joined #openstack-ceilometer23:43
*** jwcroppe has quit IRC23:43
*** thorst has joined #openstack-ceilometer23:43
*** jwcroppe has joined #openstack-ceilometer23:43
*** julim has quit IRC23:47
*** jwcroppe has quit IRC23:48
*** fawadkhaliq has quit IRC23:50
*** julim has joined #openstack-ceilometer23:58

Generated by 2.14.0 by Marius Gedminas - find it at!