Monday, 2015-08-24

*** Kennan2 is now known as Kennan00:02
*** julim has quit IRC00:39
*** julim has joined #openstack-ceilometer00:40
*** liusheng has joined #openstack-ceilometer01:10
openstackgerritliusheng proposed openstack/ceilometer-specs: Improve Nova Instance metering
*** Kennan has quit IRC02:02
*** Kennan has joined #openstack-ceilometer02:02
*** gordc has joined #openstack-ceilometer03:26
gordcwill repost tomorrow but here is the starting etherpad for our survey question. please free to delete/add as you wish:
*** dikonoor has joined #openstack-ceilometer03:28
*** julim has quit IRC03:30
*** openstackgerrit has quit IRC03:31
*** openstackgerrit has joined #openstack-ceilometer03:31
*** ddieterly has quit IRC03:39
openstackgerritgordon chung proposed openstack/ceilometer: support custom timestamp
*** ankita_wagh has joined #openstack-ceilometer03:54
*** gordc has quit IRC03:56
*** yprokule has joined #openstack-ceilometer04:24
*** fawadkhaliq has joined #openstack-ceilometer04:38
*** ddieterly has joined #openstack-ceilometer04:40
openstackgerritMerged openstack/ceilometer: disable non-metric meter definitions
*** ddieterly has quit IRC04:44
openstackgerritDivya K Konoor proposed openstack/ceilometer: WIP Events RBAC needs scoped token
*** shardy_ has joined #openstack-ceilometer05:29
*** shardy has quit IRC05:30
*** shardy_ has quit IRC05:34
*** shardy has joined #openstack-ceilometer05:34
*** ankita_wagh has quit IRC05:41
*** ddieterly has joined #openstack-ceilometer05:41
*** ankita_wagh has joined #openstack-ceilometer05:41
*** ddieterly has quit IRC05:46
*** dikonoor has quit IRC06:04
*** lsmola has joined #openstack-ceilometer06:11
*** nadya has joined #openstack-ceilometer06:17
*** nadya has quit IRC06:22
*** ildikov has quit IRC06:38
*** ddieterly has joined #openstack-ceilometer06:42
*** ddieterly has quit IRC06:46
openstackgerritDivya K Konoor proposed openstack/ceilometer: WIP Events RBAC needs scoped token
*** ankita_wagh has quit IRC07:05
*** haigang has joined #openstack-ceilometer07:05
*** haigang has quit IRC07:06
*** I has joined #openstack-ceilometer07:06
*** I is now known as Guest6305707:07
*** Guest63057 has quit IRC07:07
*** haigang has joined #openstack-ceilometer07:07
*** haigang has quit IRC07:09
*** haigang has joined #openstack-ceilometer07:10
*** ildikov has joined #openstack-ceilometer07:17
*** cdent has joined #openstack-ceilometer07:28
*** Ala has joined #openstack-ceilometer07:34
*** ddieterly has joined #openstack-ceilometer07:42
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: integration: add some new tests
*** ddieterly has quit IRC07:47
openstackgerritMehdi Abaakouk (sileht) proposed openstack/gnocchi: DO NOT MERGE: integrated downscaling test
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: gnocchi: move to jsonpath_rw_ext
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Use jsonpath_rw_ext for meter/event definitions
*** fawadkhaliq has quit IRC07:53
*** nadya has joined #openstack-ceilometer07:59
*** haigang has quit IRC08:06
*** fawadkhaliq has joined #openstack-ceilometer08:07
nadyacdent, sileht, ildikov: hi filks! I've prepared a list of use-cases for customer survey . You may add your thoughts there or here08:11
* cdent looks08:11
ildikovhi nadya! Thanks!08:13
*** cdent has quit IRC08:34
*** ityaptin has joined #openstack-ceilometer08:36
openstackgerritNadya Shakhat proposed openstack/ceilometer: Add doc about coordination for Notification Agent
*** cdent has joined #openstack-ceilometer08:37
*** ddieterly has joined #openstack-ceilometer08:43
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: integration: add some new tests
*** ddieterly has quit IRC08:48
*** fawadkhaliq has quit IRC08:52
*** cdent has quit IRC09:27
*** nadya has quit IRC09:31
*** cdent has joined #openstack-ceilometer09:31
*** nadya has joined #openstack-ceilometer09:31
*** nadya has quit IRC09:32
*** nadya has joined #openstack-ceilometer09:32
*** nadya has quit IRC09:32
*** nadya has joined #openstack-ceilometer09:33
*** nadya has quit IRC09:33
*** nadya has joined #openstack-ceilometer09:33
*** fawadkhaliq has joined #openstack-ceilometer09:33
*** ddieterly has joined #openstack-ceilometer09:44
openstackgerritOpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements
openstackgerritChris Dent proposed openstack/gnocchi: Rudimentary support for influxdb in devstack plugin
*** ddieterly has quit IRC09:48
*** shardy_ has joined #openstack-ceilometer09:53
*** shardy has quit IRC09:54
*** shardy_ has quit IRC09:58
*** shardy has joined #openstack-ceilometer09:59
cdentjd__, sileht: less than awesome news on gnocchi+influxdb:
openstackLaunchpad bug 1488027 in Gnocchi "The gnocchi wsgi app experiences timeout errors when using influxdb" [Undecided,New]10:06
*** lan_ has quit IRC10:08
*** Kennan has quit IRC10:09
*** Kennan has joined #openstack-ceilometer10:10
*** lan_ has joined #openstack-ceilometer10:21
*** ekarlso has quit IRC10:32
*** ekarlso has joined #openstack-ceilometer10:32
*** cdent has quit IRC10:38
*** ddieterly has joined #openstack-ceilometer10:45
openstackgerritRoman Dobosz proposed openstack/ceilometer: Versioned object initial patch
*** ddieterly has quit IRC10:50
openstackgerritDivya K Konoor proposed openstack/ceilometer: Events RBAC needs scoped token
*** mstyne has quit IRC10:51
*** sudorandom has quit IRC10:56
*** dikonoo has joined #openstack-ceilometer10:56
*** dikonoor has joined #openstack-ceilometer10:56
*** dikonoo has quit IRC10:56
*** mstyne has joined #openstack-ceilometer10:56
*** jroll has quit IRC10:56
_gryfany aodh developer around?10:57
*** rex_lee has quit IRC10:57
*** deepthi has joined #openstack-ceilometer11:01
*** sudorandom has joined #openstack-ceilometer11:01
*** jroll has joined #openstack-ceilometer11:02
*** rex_lee has joined #openstack-ceilometer11:04
*** nadya has quit IRC11:13
*** jaosorior has joined #openstack-ceilometer11:19
openstackgerritliusheng proposed openstack/python-ceilometerclient: WIP: Don't try to get aodh endpoint if auth_url didn't provided
dikonoorsileht: hi. I am trying to run the ceilometer functional test namely test11:25
dikonoortest_event_scenarios and it doesn't seem to be running based on the instructions at
dikonoorsileht: I generally run sudo  tox -e py2711:26
dikonooror sudo  tox -e py27 ceilometer/tests/functional/api/v2/test_event_scenario11:26
dikonoorand I don't think the tests are actually getting run . IT always passes11:28
dikonoor(py27)ubuntu@ubuntu-VirtualBox:~/openstack-src/projects/ceilometer$ sudo  tox -e py27 ceilometer/tests/functional/api/v2/test_event_scenario11:28
dikonoorpy27 develop-inst-noop: /home/ubuntu/openstack-src/projects/ceilometer11:28
dikonoorpy27 runtests: PYTHONHASHSEED='262418146'11:28
dikonoorpy27 runtests: commands[0] | python testr --slowest --testr-args=ceilometer/tests/functional/api/v2/test_event_scenario11:28
dikonoorrunning testr11:28
dikonoorrunning=OS_STDOUT_CAPTURE=${OS_STDOUT_CAPTURE:-1} \11:28
dikonoorOS_TEST_TIMEOUT=${OS_TEST_TIMEOUT:-600} \11:28
dikonoor${PYTHON:-python} -m discover ${OS_TEST_PATH:-./ceilometer/tests} -t . --list11:28
dikonoorPASSED (id=17)11:28
dikonoorSlowest Tests11:28
dikonoorsileht : Does this run those functional tests?11:29
dikonoorildikov: not sureif sileht is actually around or sleeping in a different time zone11:31
ityaptindikonoor: Hi! For testing functional tests you should run tox -e functional and write a path to tests in python format. In your case: tox -e functional ceilometer.tests.functional.api.v2.test_event_scenario11:31
dikonoorityaptin : Hi ityaptin..Thanks..LEt me try that.. I didn't find any documentation for that anywhere11:32
dikonoorityaptin: Do you know how I can run debug on functional testcases..Would tox -e debug work (I tried that and it didn't)11:34
*** thorst has joined #openstack-ceilometer11:36
ityaptindikonoor: Hmm, I missed it :( I think it's bug. I'll register it and fix asap.11:39
dikonoorityaptin : you mean missed the documentation ?11:40
dikonoorityaptin: how about running debugger on functional tests..Does that work ?11:40
ityaptinmissed the run debug tox for functional tests11:41
ityaptindikonoor: And I'll clean the documentation lack.11:42
dikonoorityaptin : Do you want me to open a bug and assign it to you so that we don;t forget?11:43
dikonoorityaptin : you might be able to help me with debugging this >>
*** gordc has joined #openstack-ceilometer11:43
dikonoorityaptin: gordc : Hi . I am trying to understand why failed. As the functional tests don't have a debug working , I am hitting a wall here11:45
ityaptindikonoor: It will be good if you open a bug and assign to me and share the link. If you don't want I could make it.11:45
dikonoorityaptin : will do that..11:46
*** ddieterly has joined #openstack-ceilometer11:46
dikonoorityaptin: tox -e functional ceilometer.tests.functional.api.v2.test_event_scenario did not run for me :(11:47
dikonoorThe test run didn't actually run any tests11:48
dikonoorERROR: InvocationError: '/bin/bash -x /home/ubuntu/openstack-src/projects/ceilometer/ ceilometer/tests/functional/api/v2/test_event_scenario'11:48
liushengmaybe you can have a try with testr to debug
liushengjust guess ;)11:49
dikonoorityaptin: sorry, that has a spelling mistake..11:49
*** ddieterly has quit IRC11:50
dikonoorliusheng: I did try the instructions at and that didn't work..Has anyone tried this before for functional tests11:51
dikonoorityaptin: the right spelling also doesn't work :-|11:52
dikonoorThe test run didn't actually run any tests11:52
dikonoorERROR: InvocationError: '/bin/bash -x /home/ubuntu/openstack-src/projects/ceilometer/ ceilometer/tests/functional/api/v2/'11:52
dikonoorgordc: hello gordc. could you take a look at the changes here
ityaptindikonoor: about "tox -e functional ceilometer.tests.functional.api.v2.test_event_scenario" could you share all output?11:54
gordcdikonoor: you can run debug by tox -edebug -- <test>11:54
gordcdikonoor: or tox -debug-mysql -- <test>11:54
gordcwhere <test> is the test name11:54
gordcit works. i usually do that and put 'import pdb;pdb.set_trace()' in code so i can step through and see all the variables11:55
dikonoorgordc: it works for functional tests as well ? ityaptin was just mentioning that it might not11:57
dikonoorgordc: just tried for test_event_scenarios11:57
gordcdikonoor: it works :)11:57
gordcor your environment is broken.11:57
gordcor mine is broken (but in a good way)11:57
dikonoorgordc: I get this:11:57
dikonoorTests running...11:57
dikonoorRan 0 tests in 0.010s11:57
dikonoor+ clean_exit /tmp/CEILO-MONGODB-ZNcIf11:57
dikonoor+ local error_code=011:57
dikonoor+ test -n ''11:58
dikonoor+ rm -rf /tmp/CEILO-MONGODB-ZNcIf11:58
dikonoor++ jobs -p11:58
dikonoor+ kill 10448 10496 1049711:58
dikonoor+ return 011:58
gordcwhat's the full command you're running?11:58
dikonoor________________________________________________________________________________ summary _________________________________________________________________________________11:58
dikonoor  debug: commands succeeded11:58
dikonoor  congratulations :)11:58
dikonoorsudo tox -edebug ceilometer/tests/functional/api/v2/test_event_scenarios.py11:58
gordcit should be 'tox -edebug -- test_event_scenarios'11:59
gordcor tox -edebug -- ceilometer.tests.functional.api.v2.test_event_scenarios11:59
dikonoorgordc: ok..let me try that11:59
gordcor replace 'debug' with 'debug-mysql' if you don't want to run mongo11:59
*** pradk has quit IRC12:01
dikonoorgordc:  tox -edebug -- ceilometer.tests.functional.api.v2.test_event_scenarios did not work. Gave me same result as before12:01
dikonoorgordc: trying with debug-mysql12:01
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: WIP: Remove xfail tests fixed in wsme
*** deepthi has quit IRC12:02
dikonoorgordc: while I try that, I need some inputs on from you. Please see the last comment from edmondsw12:03
dikonoorgordc: Talks about making user_id and project_id default traits12:03
*** liusheng has quit IRC12:04
*** liusheng has joined #openstack-ceilometer12:05
*** shardy_ has joined #openstack-ceilometer12:05
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Move profiler meters to yaml
dikonoorgordc: Do you see any problems with something like this:12:05
dikonoor    DEFAULT_TRAITS = dict(12:05
dikonoor        service=dict(type='text', fields='publisher_id'),12:05
dikonoor        request_id=dict(type='text', fields='_context_request_id'),12:05
dikonoor        project_id=dict(type='text', fields=['payload.tenant_id',12:05
dikonoor                                            '_context_tenant']),12:05
dikonoor        user_id=dict(type='text', fields=['payload.user_id',12:05
dikonoor                                          '_context_user_id']),12:05
dikonoor        tenant_id=dict(type='text', fields=['payload.tenant_id',12:05
dikonoor                                            '_context_tenant']),12:05
dikonoor    )12:05
dikonoorgordc: As tenant_id is the old terminology, that would probably have to be deprecated (not sure how to do that though)12:06
gordcdikonoor: ... i think we can discuss it.12:06
*** shardy has quit IRC12:06
gordcdikonoor: i personally don't have anything against that plan.12:07
gordcdikonoor: keep in mind. code freeze is next week12:07
gordcif you do propose this, this needs to be done asap (ie. now)12:07
ildikovdikonoor: hi, soory, I'm on a workshop today, so I'm that responsive :(12:08
dikonoordikonoor: ok.. :-| Let me add those as default traits then and send that for review..these functional tests are killing me12:08
gordcdikonoor: regarding your patches, there seems to two different WIP patches... which one is should we review?12:09
*** nadya has joined #openstack-ceilometer12:09
dikonoorgordc: try this first >>
ildikovnadya: I added a few questions/comments to the etherpad12:10
dikonoorgordc: this is no longer wip12:10
*** shardy_ has quit IRC12:10
dikonoorgordc: is pretty much done..but I haven't been able to figure out why those tests fail12:11
gordcdikonoor: so this patch just checks to ensure there is USER/PROJECT_ID in request?12:11
*** shardy has joined #openstack-ceilometer12:11
gordcok. but it still requires admin?12:11
nadyagordc: hi Gordon! Have you seen this ?12:11
gordcwhat if a operator wants (for whatever reason) to have non-auth api?12:12
dikonoorgordc: removing admin and using rules are in patch
*** deepthi has joined #openstack-ceilometer12:12
gordcnadya: nope. but that's very cool12:13
gordcnadya: i had this:
gordcwe can copy paste your stuff in there?12:13
nadyagordc: yep, sure!12:13
gordci think it'd be good for our question to be maybe 5 answers and a 6th which is other?12:13
*** deepthi has quit IRC12:15
nadyagordc: do we have any limit about amount of options? Or you think that people may just skip the question with too many options :)?12:16
dikonoorgordc: wouldn't a non-auth api break the existing logic of looking for an admin role ?12:16
gordcnadya: 'skip because too many options'12:17
gordcnadya: i'm not sure there's a limit... i know in school we learned most people can only remember up to 7 options... so maybe 7 is our absolute max12:18
gordcdikonoor: i'm going to check code to see exactly what it looks at12:18
dikonoorgordc: when you say non-auth api, would it just mean that someone chooses to NOT have the authtoken middleware here >>
gordcdikonoor: yeah i guess based on the name it should be admin... but that's obviously too restrictive.12:20
gordcdikonoor: right... let me dig up a bp12:20
*** julim has joined #openstack-ceilometer12:22
nadyagordc: In your question you're asking about OpenStack resources monitoring. But we have several hardware pollsters. IMHO, Ceilo is not right tool for hw monitoring but it would be good to know customer's view12:22
gordcnadya: yeah, i think the idea is to understand what they're our data for, and we can debate whether to add more resources to those cases?12:24
ildikovnadya: gordc: if we want to limit the topics, then I guess it would be good to to focus on what Ceilometer can be used for today12:26
gordcthat was the bp i was thinking of. but now that i read it, i think it's not about api in general but specific calls.12:26
gordcdikonoor: so i think it's safe to carry on with assumption that authentication is required.12:27
nadyaildikov: gordc: so we should choose between "what is your use-case" and "what features do you want from Telemetry"12:28
gordcildikov: 'can be used for' can be a very big set of possibilities12:28
gordcsome that make sense, others that don't12:28
dikonoorgordc: ok..the changes for this bp got abandoned12:29
gordcnadya: what are the options you were thinking about for 'what features do you want'?12:29
gordcexisting features? or new features?12:29
nadyagordc: e.g. application lvl monitoring, trending analysis; new features mostly12:30
*** edmondsw has joined #openstack-ceilometer12:31
gordcsileht: do we need this bumped to 0.1.7?
silehtgordc, to support +/*- yes12:33
gordcnadya: tbh, i think analysis tools should be a separate project.12:33
gordcsileht: ok. i'll repropose.12:34
nadyagordc: yep, I agree; maybe it's better to rephrase "collect all data to make trending analyses possible"12:35
gordcnadya: "collect all data." -- i fixed it for you ;)12:36
ildikovgordc: I meant that in nadya's list there are a few options that Ceilo is not good for today12:37
ildikovI think it's better to focus on what we would like to improve further as we still don't have that many resources around to work on things12:38
nadyaildikov: yep, this list is smth like "all possible use cases for Telemetry"12:38
gordcildikov: good point12:38
gordcnadya: i would suggest we try to merge your existing list into shorter one line options and that can be a possible question12:39
gordcanother possible question is asking about gnocchi specifically but i think that's a bit too focused and will not give us information on rest of services.12:40
gordcsileht: updated
nadyagordc: yep, I think that customers don't care about the way of data storage: time series or not12:42
silehtgordc, thx12:42
nadyagordc: I'm ok with merge. Will start working on this shortly12:42
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Add cpu meters to yaml
*** ddieterly has joined #openstack-ceilometer12:46
ildikovnadya: thanks for the work on this!12:47
openstackgerritRoman Dobosz proposed openstack/ceilometer: Versioned object initial patch
gordcnadya: ^ much appreciated.12:47
*** shardy_ has joined #openstack-ceilometer12:48
openstackgerritRoman Dobosz proposed openstack/ceilometer: Versioned object for Meter and Sample model
*** shardy has quit IRC12:49
*** ddieterly has quit IRC12:51
openstackgerritRoman Dobosz proposed openstack/aodh: Versioned object initial patch
*** shardy_ has quit IRC12:53
*** shardy has joined #openstack-ceilometer12:54
*** ddieterly has joined #openstack-ceilometer12:59
*** annasort has joined #openstack-ceilometer13:00
*** ddieterly has quit IRC13:07
*** shardy_ has joined #openstack-ceilometer13:15
*** shardy has quit IRC13:17
*** ildikov has quit IRC13:18
*** ildikov has joined #openstack-ceilometer13:20
*** shardy_ has quit IRC13:21
*** shardy has joined #openstack-ceilometer13:21
*** jroll has quit IRC13:23
*** links has joined #openstack-ceilometer13:31
*** exploreshaifali has joined #openstack-ceilometer13:33
*** prad_ has joined #openstack-ceilometer13:34
*** prad_ is now known as pradk13:35
*** jroll has joined #openstack-ceilometer13:37
openstackgerritDivya K Konoor proposed openstack/ceilometer: WIP : Control Events RBAC from policy.json
*** r-mibu has quit IRC13:45
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: Use jsonpath_rw_ext for meter/event definitions
*** r-mibu has joined #openstack-ceilometer13:50
nadyagordc: - please see the first 7 items. I've tried to include all your cases, but 'health monitoring' looks similar to 'alarming' to me. Feel free to add/remove items13:59
*** boris-42 has joined #openstack-ceilometer14:00
openstackgerritMehdi Abaakouk (sileht) proposed openstack/ceilometer: gnocchi: move to jsonpath_rw_ext
gordcnadya: that looks like a good list.14:02
gordci'll build teh question from those.14:02
gordcwe got lucky that you were already working on this internally :)14:03
*** fawadkhaliq has quit IRC14:07
*** r-mibu has left #openstack-ceilometer14:08
*** Ala has quit IRC14:08
*** jasonamyers has joined #openstack-ceilometer14:09
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Move profiler meters to yaml
pradkgordc, re your tests comment, i'm going to modify this commit to validate the filter, expressions and prefix etc14:18
pradkgordc, rebased the profiler and cpu ones to use the extension14:19
gordcpradk: cool cool14:19
*** ahilan has joined #openstack-ceilometer14:20
*** alejandrito has joined #openstack-ceilometer14:37
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Add validation tests for arithmetic, string and prefix expressions
*** exploreshaifali has quit IRC14:46
*** bapalm- has joined #openstack-ceilometer14:46
gordcpradk: just curiuos but do we need all teh lookup changes still in this patch?
pradkgordc, i think lookup sounds better than multi imo14:52
*** nadya has quit IRC14:53
gordcpradk: asking more since we plan on dropping it?14:53
*** liusheng has quit IRC14:55
*** liusheng has joined #openstack-ceilometer14:55
pradkgordc,  yea i just updated whats needed to support ext and left the rest in place14:56
pradkgordc, we can drop it when we remove the support later? its just a rename so up to you14:59
gordcpradk: are you still working on support? if not, i can look at it.14:59
pradkgordc, i'm looking into updating the yaml to use $. syntax now.. i can look next15:00
*** ddieterly has joined #openstack-ceilometer15:03
pradkgordc, so we still want to convert paths to $. correct?15:09
*** jaypipes has joined #openstack-ceilometer15:09
*** cmyster has joined #openstack-ceilometer15:09
gordcif it's required.15:10
pradkthe parser seems to handle the difference in most cases15:10
gordcseems like sileht said it's smart enough to figure out paths in most in cases?15:10
pradkyea thats what i'm seeing15:10
*** thorst has quit IRC15:11
pradkif we explicitly want to check the path we can pass $.15:11
pradkfor expressions etc15:11
gordcpradk: i guess it's more consistent15:11
gordcif we apply it everywhere.15:11
*** thorst has joined #openstack-ceilometer15:11
pradkyea that part i agree, i dont think we need any special logic in our code.. just update the yaml to use this syntax15:12
gordci think we should still search for [*] to check multi value support (rather than use lookup/multi atttr)15:12
gordcpradk: cool cool. give it a try and see if it breaks :)15:12
*** jruano has joined #openstack-ceilometer15:12
pradksure.. so you are looking at the [*] support?15:13
pradki'll push this change in a bit15:13
gordcpradk: i could. but i haven't done anything yet15:13
*** cdent has joined #openstack-ceilometer15:14
silehtgordc, [*] is not the only expression to build array with jsonpath15:23
*** ildikov has quit IRC15:23
silehtyou can use $.foobar.[1:7] or [@.foobar > 5]15:24
gordcsileht: yeah, you're right.15:25
gordcsileht: do you suggest we keep the 'multi' attribute to mark which attributes we expect to be array?15:26
gordcsileht: kk15:27
gordci'm just thinking whether we need to know whether it's array or if we just need to know it's a path.15:27
silehtI have one last thing I haven't yet done in jsonpath_rw_ext is the object mapping thing15:27
* gordc looks at cocde15:27
gordcobject mapping?15:28
* sileht prepare an example15:28
*** r-mibu has joined #openstack-ceilometer15:28
gordcsileht: so it remaps attribute keys? i don't think we need that right now... although it might help with custom_metadatas15:35
silehtgordc, it creates a new array only remapped attributes15:35
silehtgordc, writing expression to build the sample will looks like a wierd big line15:36
openstackgerritAnna Sortland proposed openstack/ceilometer: require SQLAlchemy >=0.9.9
silehtthat why I prefer multi/lookup stuffs, it's more readable for the deployer15:36
gordcsileht:  i see.15:37
*** IBerezovskiy has joined #openstack-ceilometer15:37
gordcyeah, we should probably document what the multi/lookup value is.15:37
*** david-ly_ is now known as david-lyle15:38
*** mstyne has quit IRC15:38
*** mstyne has joined #openstack-ceilometer15:41
*** ddieterly has quit IRC15:42
*** exploreshaifali has joined #openstack-ceilometer15:44
openstackgerritChris Dent proposed openstack/gnocchi: Rudimentary support for influxdb in devstack plugin
*** exploreshaifali has quit IRC15:48
r-mibusileht, ping15:51
silehtr-mibu, pong15:51
r-mibusileht, thank you for reviews on my event-alarm patches15:51
*** rbak has joined #openstack-ceilometer15:51
r-mibusileht, I'd like ask you about the code placement15:52
*** alejandrito has quit IRC15:53
r-mibusileht, as you pointed, new evaluator and service inherit existing base classes15:54
r-mibusileht, due to difference of the evaluation sequence, there are non-used method that might be not clean15:55
r-mibusileht, should I write complete new class with bit duplication with the existing one?15:55
silehtr-mibu, I suggest to move common thing into a class named 'BaseEvaluator', the Evaluator and your evalutor inherit from BaseEvaluator15:58
silehtthey are no code duplication15:58
silehtabout the service class only the database connection will be duplicated, it's ok15:59
r-mibusileht, I agree your suggestion on service side.16:00
r-mibusileht, for evaluator, should I do the change in the current patch?16:01
r-mibusileht, I'd like to do that change after the current one merged16:01
*** fawadkhaliq has joined #openstack-ceilometer16:01
silehtI guess yes16:01
r-mibusileht, but I will remove setup.cfg entry16:02
silehtyour evaluator doesn't need any entry in setup.cfg (Or I have missed something)16:02
r-mibusileht, so that existing alarm service won't load new event-alarm evaluator16:03
r-mibusileht, yes if I follow your suggestion16:03
r-mibuok, thanks - then, I will amend my patches16:04
*** links has quit IRC16:08
*** thumpba has joined #openstack-ceilometer16:12
*** _cjones_ has joined #openstack-ceilometer16:19
cdenthas there been any further discussion about what do with resource ids and gnocchi?16:21
silehtcdent, that clearly a big issue !16:27
cdentI was just reviewing some of the jsonpath related changes, and it's creating resource ids in all kinds of interesting ways16:28
cdentand I was reminded16:28
silehtwe two choice fixing all openstack applications to add uuid to everything16:28
silehtor allow gnocchi to have string instead of uuid16:28
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Change json path's to start with $. for consistency
silehtmy heart prefer the first one, but second looks more feasable16:29
cdenti was wondering about some kind of translation layer...not really sure what that would mean though16:30
openstackgerritDivya K Konoor proposed openstack/ceilometer: Events RBAC needs scoped token
silehtcdent, a new db that store the mapping between a bad uuid to a good one ?16:30
cdentI was thinking something more along the lines of a a truncated sha256 checksum on the resource id, made to use the same number of bits as a uuid16:31
cdentbut that was a wild hair idea, didn't really put much thought into whether it was actually workable16:32
silehtcdent, nice workaround16:32
*** ildikov has joined #openstack-ceilometer16:33
openstackgerritRyota MIBU proposed openstack/aodh: Make event-alarm evaluator caching alarms
openstackgerritRyota MIBU proposed openstack/aodh: Add evaluator for event alarm
openstackgerritRyota MIBU proposed openstack/aodh: Add listener service for event alarm evaluation
openstackgerritChris Dent proposed openstack/gnocchi: Rudimentary support for influxdb in devstack plugin
*** IBerezovskiy has quit IRC16:39
cdentsileht: you been seeing any of this kind of failure in your gnocchi test runs:
*** r-mibu has left #openstack-ceilometer16:40
silehtcdent, I cross it while I was testing the integration tests, but not dig into16:40
silehtcdent, I guess that due to a oslo.db releases16:40
cdentah, that makes sense16:41
* cdent tries to repeat16:41
openstackgerritMerged openstack/aodh: doc: Fix the devstack configuration
silehtcdent, we need to upgrade to 2.4.1 I guess16:43
silehthum the tests already run 2.4.116:43
silehtcdent, I have to go, but I known that 2.4.0 was having trouble about alebmic migration, perhaps 2.4.1 still have some trouble16:46
silehtcdent, not all project to check running schema with the migration one16:47
silehtperhaps we caughth a new bug16:47
openstackgerritMerged openstack/ceilometer: Fix dependency for doc build
cdentmy local tests are 2.4.0 and they pass16:48
cdentupgrade to 2.4.1 and they break16:48
silehtcdent, the difference is only two commit :)16:49
openstackgerritIlya Tyaptin proposed openstack/ceilometer: Allow to run debug tox job for functional tests
*** ankita_wagh has joined #openstack-ceilometer16:53
openstackgerritMerged openstack/ceilometer: add flexible grouping key
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Adds support for dynamic event pipeline
cdentsileht: apparently it is because of . There is an actual problem in the migrations: "yes these FKs have ondelete=RESTRICT in them, need that on both sides"17:08
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Adds support for dynamic event pipeline
cdentjd__: you around? i'm ignorant about alembic17:19
gordccdent: this?
openstackLaunchpad bug 1480346 in Ceilometer "Gnocchi doesn't accept OpenStack id madness" [High,Confirmed] - Assigned to jiaxi (tjxiter)17:20
* cdent looks17:21
cdentgordc: yeah, that was what sileht and I were disussing above17:21
cdentas sileht says "clearly a big issue"17:22
gordccdent: i was going to open a openstack thread on that.17:22
gordcbut it's going to be hard as hell. all the id fields in openstack are set to accept any arbitrary string of 255char17:22
cdentyeah, thus my wondering about some kind of work around17:23
gordcmake resource ids a display id?17:24
*** ankita_wagh has quit IRC17:24
cdentthat was one of the workarounds, the other was have a one way translation layer: incoming resource id that is not uuid is sha256 hashed and then transformed into number of bits in a uuid17:25
cdent(which is crazy, but I wonder)17:25
gordci wonder what requires more computation, sha256 hash or string comparison.17:26
cdentthe idea on the latter wasn't to maintain a mapping17:26
cdentjust make ids17:26
cdentthat are one way computable from strings17:26
cdentbut it really is just a wild hair idea, I have no idea if it would actually work17:27
gordcbut it would happen every single record right?17:27
gordcdo it, if no ones complains, it ain't broke.17:27
cdentevery time you write a resource using the resource url17:27
cdentwhich could be made to be more rare than it currently is17:29
cdent(by determing and then using  the /v1/metric url )17:29
gordci see... not sure myself.17:30
gordci think it's pretty important for us to figure out though. because if you look at logs, it's dropping a lot of measurments because we're not giving it valid ids17:31
cdentoh yeah, definitely17:31
cdentthere are _many_ "bad" ids17:31
gordchow would we go about solving it in OpenStack? tell everyone to just use uuid4()?17:34
cdentthat's the ideal outcome but it seems unlikely17:35
cdentand people will wonder by gnocchi's being "difficult"17:36
gordcyeah, but i think we should start fixing it globally... considering it'll take 2+ cycles17:37
gordcbut yeah, currently without a workaround in place in gnocchi, it's pretty useless in terms of using it with openstack17:37
cdentsileht, jd__ :
openstackLaunchpad bug 1488190 in Gnocchi "gnocchi alembic migrations not compatible with alembic 0.8.1" [Undecided,New]17:38
* cdent does some experiments17:38
*** yprokule has quit IRC17:40
*** dikonoor has quit IRC17:40
*** ankita_wagh has joined #openstack-ceilometer17:43
*** Ephur has joined #openstack-ceilometer17:53
openstackgerritRyota MIBU proposed openstack/aodh: Make event-alarm evaluator caching alarms
openstackgerritRyota MIBU proposed openstack/aodh: Add listener service for event alarm evaluation
*** Ephur has quit IRC17:54
openstackgerritPradeep Kilambi proposed openstack/ceilometer: Change json path's to start with $. for consistency
*** rjaiswal has joined #openstack-ceilometer17:55
cdentgordc: will have a proof of concept in a few minutes, just cleaning up some loose  ends18:17
gordccdent: cool cool. is it a dispatcher change?18:18
cdentno, it is a gnocchi change18:19
cdentyou can't do it as a dispatcher change because then you are limited to what clients you can use18:19
*** bapalm- is now known as bapalm_18:19
*** ankita_w_ has joined #openstack-ceilometer18:27
*** ankita_wagh has quit IRC18:30
*** exploreshaifali has joined #openstack-ceilometer18:36
*** fawadkhaliq has quit IRC18:55
openstackgerritMerged openstack/aodh: Add evaluator for event alarm
*** ankita_wagh has joined #openstack-ceilometer18:59
*** ankita_w_ has quit IRC19:03
openstackgerritRohit Jaiswal proposed openstack/ceilometer: Adds support for dynamic event pipeline
*** nadya has joined #openstack-ceilometer19:09
openstackgerritgordon chung proposed openstack/ceilometer: support custom metadata
*** jaypipes has quit IRC19:18
openstackgerritChris Dent proposed openstack/gnocchi: WIP: Introduce a resource id translation shim
cdentgordc: I reckon that ^ is pretty insane but might stir some ideas.19:24
cdentgordc: another gnocchi limitation we need to think about is the way(s) in which it stores a _very_ limited (and fixed) set of metdata per resource19:26
gordccdent: are you suggesting it should store more?19:32
gordcor be more flexible?19:32
gordcor somethign else.19:32
cdentI'm saying that people do queries on metadata in the ceilometer api19:33
cdentthose queries won't just be different in gnocchi, they'll be impossible19:33
cdentand the question is: does it matter?19:35
gordcconsidering it's a completely new api, i'd say it matters less?...19:36
gordcwe definitely can't go back to storing everything.19:37
cdentyeah, no doubt19:37
gordci'm sort of in the extreme boat where i don't think we should have the metadata at all19:37
gordcand we should somehow leverage the events data. but i really don't have a fully realised plan for that.19:37
cdentI don't think we should have metadata either19:38
cdentand we should just have a way to look up any resource by uuid from the service that "owns" it19:38
cdentwhere by "we" I mean "client code"19:38
cdentBut I worry about some use case that we're not thinking of (because none of us are _actually_ ceilometer users) is going to get lost in the shuffle19:39
cdentanyway, look at the patch I posted above please, and tell me it is nuts19:40
gordci don't think i fully understand. is the idea that you can only query on for metrics of a resource and that's the only filtering capabilities possible?19:40
cdenthere's all the info stored with a resource in gnocchi:
cdent(beyond the generics, defined here:19:43
*** Ephur has joined #openstack-ceilometer19:46
*** ahilan has quit IRC19:48
cdentgordc: so which part is not fully understood?19:50
gordcdid you want to get rid of that?19:51
gordcwe also define attr here:
gordcwhich is somethign i don't like either (we have definitinos in two places it seems)19:52
cdentyeah, that's just a duplication19:52
cdentbut is somewhat necessary to maintain gnocchi as an indepdendent beastie but still having a strong contract19:52
cdentwhere "strong contract" means "indexed database"19:53
openstackgerritMerged openstack/python-ceilometerclient: Updated from global requirements
*** exploreshaifali has quit IRC19:53
gordcso what's the change you're thinking of?19:54
*** ahilan has joined #openstack-ceilometer19:55
cdentso there's two separate threads here19:55
cdenta) me being worried about whether the lack of metadata is going to be upset people's use cases19:55
cdentb) me agreeing with you about metadata being problematic and then going off into blue sky territory of metadata being a globally available datastore which is stored and mainted by each service (without stopping to think about how people might do queries)19:56
gordcfrom what i understand, it'll be hard to add more metadata... a migration is required?19:57
gordcah i got it.19:57
cdentb, cont) so I'm not thinking about something I actually want to do, just a way I wish the world was19:57
cdentadding more metadata, as things currently stand, requires a migration19:57
gordcyeah, i feel like that idea was floated once... if there was an easy way to search eavesdrop logs i think it's somewhere.19:57
gordcbut yeah, i'd like that world.19:58
cdentstoopid world19:58
cdentanyway, that sha256 stuff has made me tired so I'm going to go sit on the couch and do anything else19:58
cdentjd seemed to like it well enough19:58
gordcbut it seems like it'd require a very coordinated api system. since data is everywhere. but i'm just talking randomly right now19:58
*** jasonamyers has quit IRC19:59
cdentimagine if you could a sort of map reduce across all of openstack..19:59
gordcmap reduce is dead.19:59
gordcis jd__ on half pto or something?20:00
cdentdon't really know20:00
cdentbut it is august20:00
cdentand he is french20:00
gordclol i feel like the next sentence could get you in trouble20:01
cdentthankfully there is no next sentence20:01
gordchr would be proud of us.20:02
EmilienMcdent: hey !!20:03
EmilienMcdent: what about frenchies?20:03
EmilienMgordc: HR!20:03
cdentI've been led to believe20:03
cdentbut perhaps it's wrong20:03
cdentthat in August all french urbanites pack up and leave the city20:04
EmilienMcdent: it's 10pm now, I guess he just has a life? :)20:04
gordccdent: EmilienM's been on pto for 3 weeks20:04
cdenthah, no, not talking about today20:04
EmilienMgordc: no he worked remotely afik20:04
*** exploreshaifali has joined #openstack-ceilometer20:04
gordcit seems like he's travelling so i'm very surprised to see him reviewing/patching.20:05
EmilienMa lot of activities20:05
EmilienMhe was not PTO I think20:05
cdentin case it's not clear, we weren't complaining, just wondering20:05
gordci'm complaining that EmilienM was on pto20:06
cdentthat's another matter entirely20:06
cdenthe shouldn't be allowed any time off ever20:06
gordctrue. EmilienM you must work for your visa.20:07
EmilienMI'm always PTO man20:07
EmilienMgordc: you know I have troubles because of that?20:07
EmilienMgordc: to get the visa, I need XX days of work20:07
EmilienMand I took too much days this year so I'll have to ask for a second work permit lol20:08
gordcjust lie20:08
gordcsay you worked 400 days last year20:08
* cdent must go20:09
cdentdon't forget to not take PTO20:10
*** cdent has quit IRC20:10
openstackgerritFrederic Guillot proposed openstack/ceilometer-specs: Generic Timeseries Support
*** exploreshaifali has quit IRC20:13
openstackgerritFrederic Guillot proposed openstack/ceilometer: Add generic timeseries support
openstackgerritFrederic Guillot proposed openstack/ceilometer: Add generic timeseries support
*** nadya has quit IRC20:41
*** ddieterly has joined #openstack-ceilometer20:47
*** ddieterly has quit IRC20:48
*** shardy has quit IRC20:50
*** thorst has quit IRC21:03
openstackgerritMerged openstack/aodh: Add listener service for event alarm evaluation
*** ankita_w_ has joined #openstack-ceilometer21:04
*** ankita_wagh has quit IRC21:07
*** liusheng has quit IRC21:08
*** liusheng has joined #openstack-ceilometer21:09
*** gordc has quit IRC21:26
*** jruano has quit IRC21:45
*** ankita_wagh has joined #openstack-ceilometer21:53
*** ankita___ has joined #openstack-ceilometer21:53
*** ankita_w_ has quit IRC21:56
*** ankita_wagh has quit IRC21:57
*** pradk has quit IRC22:04
*** ankita___ has quit IRC22:11
*** ankita_wagh has joined #openstack-ceilometer22:11
*** srsakhamuri has joined #openstack-ceilometer22:16
*** srsakhamuri-2 has joined #openstack-ceilometer22:21
*** jruano has joined #openstack-ceilometer22:24
*** srsakhamuri2 has joined #openstack-ceilometer22:33
*** ankita_w_ has joined #openstack-ceilometer22:33
*** ankita_wagh has quit IRC22:36
*** srsakhamuri2 is now known as srini22:40
*** srini is now known as sinisak22:41
*** thumpba has quit IRC22:42
*** jaosorior has quit IRC22:45
*** edmondsw has quit IRC22:57
*** tongli has joined #openstack-ceilometer23:02
*** rbak has quit IRC23:04
*** thorst has joined #openstack-ceilometer23:08
*** Ephur has quit IRC23:22
*** EmilienM has quit IRC23:34
*** EmilienM has joined #openstack-ceilometer23:36
*** tongli has quit IRC23:54

Generated by 2.14.0 by Marius Gedminas - find it at!