| *** matsuhashi has joined #openstack-metering | 00:22 | |
| *** anteaya has quit IRC | 01:04 | |
| *** sandywalsh_ has quit IRC | 01:14 | |
| *** changbl has joined #openstack-metering | 01:29 | |
| *** gordc has joined #openstack-metering | 01:30 | |
| *** sandywalsh has joined #openstack-metering | 01:48 | |
| *** julim has joined #openstack-metering | 02:07 | |
| *** julim has quit IRC | 02:12 | |
| *** boris-42 has quit IRC | 02:13 | |
| *** sandywalsh has quit IRC | 02:50 | |
| *** nati_ueno has joined #openstack-metering | 02:53 | |
| openstackgerrit | Terri Yu proposed a change to openstack/ceilometer: Add example with return values in API v2 docs https://review.openstack.org/47668 | 03:49 |
|---|---|---|
| *** terriyu has quit IRC | 03:52 | |
| *** matsuhashi has quit IRC | 03:59 | |
| *** fnaval_ has joined #openstack-metering | 04:00 | |
| *** matsuhashi has joined #openstack-metering | 04:02 | |
| *** gordc has quit IRC | 05:07 | |
| *** nati_ueno has quit IRC | 05:26 | |
| *** nati_ueno has joined #openstack-metering | 05:45 | |
| *** eglynn has joined #openstack-metering | 05:49 | |
| openstackgerrit | Jenkins proposed a change to openstack/ceilometer: Imported Translations from Transifex https://review.openstack.org/47967 | 06:03 |
| *** SergeyLukjanov has joined #openstack-metering | 06:21 | |
| *** matsuhashi has quit IRC | 06:47 | |
| *** matsuhashi has joined #openstack-metering | 07:06 | |
| *** SergeyLukjanov has quit IRC | 07:24 | |
| *** eglynn has quit IRC | 07:24 | |
| *** SergeyLukjanov has joined #openstack-metering | 07:30 | |
| *** ashestakov_ has joined #openstack-metering | 07:39 | |
| *** SergeyLukjanov has quit IRC | 07:45 | |
| *** SergeyLukjanov has joined #openstack-metering | 07:46 | |
| openstackgerrit | A change was merged to openstack/ceilometer: tests: import pipeline config https://review.openstack.org/47200 | 07:47 |
| *** neumernace has joined #openstack-metering | 07:54 | |
| *** shang has quit IRC | 07:56 | |
| *** bogdando has quit IRC | 07:59 | |
| *** bogdando has joined #openstack-metering | 08:02 | |
| *** SergeyLu_ has joined #openstack-metering | 08:02 | |
| *** SergeyLukjanov has quit IRC | 08:04 | |
| *** ashestakov_ has quit IRC | 08:04 | |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/python-ceilometerclient: Allow to update an alarm partially https://review.openstack.org/48402 | 08:06 |
| *** briancline has quit IRC | 08:10 | |
| *** eglynn has joined #openstack-metering | 08:10 | |
| *** briancline has joined #openstack-metering | 08:17 | |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/python-ceilometerclient: Allow to update an alarm partially https://review.openstack.org/48402 | 08:18 |
| sileht | Hi, eglynn I have made a new change for alarming client, about your remarks on alarm update: https://review.openstack.org/48402 | 08:27 |
| eglynn | sileht: thanks! I just looked and made a minor suggestion, otherwise looks good | 08:28 |
| sileht | eglynn, this allow to have a working stack-update without any heat change. | 08:28 |
| sileht | eglynn, thx | 08:28 |
| *** nati_ueno has quit IRC | 08:29 | |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/python-ceilometerclient: Allow to update an alarm partially https://review.openstack.org/48402 | 08:31 |
| sileht | eglynn, done :) | 08:31 |
| eglynn | sileht: thanks! looking now ... | 08:31 |
| eglynn | sileht: cool, +2'd ... | 08:32 |
| *** shang has joined #openstack-metering | 08:37 | |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/python-ceilometerclient: Allow to update an alarm partially https://review.openstack.org/48402 | 08:48 |
| sileht | eglynn, I have missed the shell.py in my commit | 08:49 |
| *** Alexei_987 has joined #openstack-metering | 08:49 | |
| eglynn | sileht: d'oh! ;) | 08:49 |
| sileht | eglynn, I guess I have found a funny bug in alarm API | 08:49 |
| *** osphy has joined #openstack-metering | 08:50 | |
| eglynn | sileht: interesting ... what was the bug? | 08:50 |
| sileht | eglynn, when a alarm is disabled, it seems you can not GET it from the API | 08:50 |
| eglynn | sileht: hmmm, that's badness clearly | 08:51 |
| eglynn | sileht: ceiloclient issue, or in the API service layer? | 08:51 |
| sileht | eglynn, API service layer | 08:51 |
| eglynn | sileht: can you file a bug for this, sounds like a blocker for RC1 to me | 08:51 |
| sileht | eglynn, yes | 08:52 |
| eglynn | jd__: agree? ^^^ | 08:52 |
| * jd__ reads | 08:52 | |
| jd__ | I totally agree! | 08:52 |
| jd__ | add it as a milestone when you open the bug sileht | 08:52 |
| eglynn | cool | 08:52 |
| eglynn | sileht: need to set enabled kwarg in https://github.com/openstack/ceilometer/blob/master/ceilometer/api/controllers/v2.py#L1226 ? | 08:55 |
| eglynn | sileht: (otherwise looks like it's always defaulting to True, if I read the code correctly ...) | 08:56 |
| sileht | eglynn, exactly | 08:57 |
| * eglynn was probably just stating the obvious there ... ;) | 08:57 | |
| eglynn | sileht: or maybe better to just change the default value for enabled to None in https://github.com/openstack/ceilometer/blob/master/ceilometer/storage/base.py#L212 | 08:59 |
| eglynn | sileht: (and corresponding changes to the concrete storage drivers ...) | 09:00 |
| sileht | eglynn, the questions is 'GET /alarms' should return all alarms ? (I think yes), so enabled=None is the only good solution | 09:00 |
| eglynn | sileht: true that | 09:01 |
| eglynn | sileht: yep I thik it should return all alarms, regardless of enabled status | 09:01 |
| eglynn | sileht: the evaluators take care of ignoring the disabled one | 09:01 |
| eglynn | sileht: a lot of disabled alarms could potentially skew a partitioning somewhat I guess | 09:02 |
| lsmola | jd__, eglynn hello | 09:02 |
| eglynn | sileht: but generally they are going to be random distributed, so should be OK | 09:02 |
| eglynn | lsmola: hi | 09:02 |
| lsmola | jd__, eglynn could you please checkout the https://blueprints.launchpad.net/ceilometer/+spec/ipmi-inspector-for-monitoring-physical-devices ? | 09:02 |
| eglynn | lsmola: will do | 09:03 |
| eglynn | lsmola: I see you've pasted in some discussion with the IPMI domain experts | 09:03 |
| lsmola | jd__, eglynn seems we will have to poll Ironic API, so now I don't know which Agent will take care of that, (the central one?) | 09:03 |
| eglynn | lsmola: i.e. lifeless & devananda | 09:04 |
| jd__ | lsmola: this looks like the hardware blueprint we have already | 09:04 |
| eglynn | lsmola: v. good to get their take on it | 09:04 |
| sileht | eglynn, currently the evaluator control the enabled field but in reality ceiloclient.alarms.list return only enabled alarms, I will take a look if we can propose a other review the use alarms.list(enabled=True) | 09:04 |
| *** fnaval_ has quit IRC | 09:04 | |
| eglynn | sileht: makes sense | 09:04 |
| lsmola | eglynn, I have just mentioned it to them, they have pasted themselves, as well as they've created the Ironic BP :-) | 09:04 |
| eglynn | lsmola: | 09:05 |
| eglynn | lsmola: k | 09:05 |
| lsmola | jd__, you mean this one ? https://blueprints.launchpad.net/ceilometer/+spec/monitoring-physical-devices | 09:05 |
| jd__ | lsmola: yes | 09:06 |
| lsmola | jd__, the other one was just about IPMI inspector added to this agent | 09:06 |
| lsmola | jd__, though as we will have to query Ironic API, it may belong to another agent, than the hardware one | 09:06 |
| *** ashestakov has joined #openstack-metering | 09:07 | |
| lsmola | jd__, the hardware agent should be baked into each hardware device | 09:07 |
| jd__ | lsmola: typically, the hardware agent needs to know about resources you need to poll (e.g. IP for SNMP polling) | 09:07 |
| lsmola | jd__, well, at least that is how I understand it :-) | 09:07 |
| jd__ | lsmola: no that's not it | 09:07 |
| jd__ | the hardware agent is like the central one with a different of set of plugin for now | 09:08 |
| jd__ | though it might not make sense (code has not been merged yet) since we can extend the central one to do that too | 09:08 |
| jd__ | anyway I don't feel like I'm going to debate architecture on that right now, but it's just to give you an idea | 09:09 |
| jd__ | I'm still totally fine with your b | 09:09 |
| jd__ | bp | 09:09 |
| lsmola | jd__, ok, I see | 09:09 |
| eglynn | honestly I don't have enough domain knowledge on IPMI as yet, I need to educate myself on that ... | 09:10 |
| ashestakov | hi all | 09:10 |
| eglynn | (to make any kind of informed determination ...) | 09:10 |
| ashestakov | how i can do application monitoring using ceilometer? | 09:10 |
| lsmola | jd__, yeah we were talking with eglynn about this, at it seemed like we can have one Central Hardware agent, or Hardware agent in each baremetal | 09:10 |
| jd__ | eglynn: knowledge is so overrated :) | 09:10 |
| lsmola | hehe | 09:11 |
| eglynn | jd__: LOL :) | 09:11 |
| ashestakov | for example, i want to get statistics of mysql server (count of select/instert, count of sessions etc) | 09:11 |
| jd__ | ashestakov: write a pollster for the central agent doing that | 09:11 |
| eglynn | ashestakov: ceilo monitoring is more oriented to cloud resources (instances, images, volumes etc.) that applications | 09:11 |
| ashestakov | is it possible with ceilometer? | 09:11 |
| eglynn | s/that/than/ | 09:11 |
| *** zaneb has left #openstack-metering | 09:11 | |
| jd__ | ashestakov: it's possible though like eglynn I don't know if you really want to do that :) | 09:12 |
| ashestakov | i want thing like amazon cloudwatch for rds | 09:13 |
| *** haomeng_ has joined #openstack-metering | 09:14 | |
| jd__ | you can't always get what you want, they sang | 09:14 |
| jd__ | ashestakov: but you could do that indeed with Ceilometer | 09:15 |
| *** haomeng has quit IRC | 09:17 | |
| lsmola | jd__, ok, is this a possible thing to raise up one the next Ceilometer meeting? | 09:18 |
| eglynn | lsmola: just add it to the wiki | 09:18 |
| eglynn | lsmola: http://wiki.openstack.org/Meetings/MeteringAgenda (with your name against it ...) | 09:18 |
| lsmola | jd__, I am now installing a Tripleo and planning to test the SNMP agent | 09:18 |
| ashestakov | if i write custom agent which will be installed on my instance with Mysql and will reports metrics to collector, are any issues in this way? maybe performance | 09:18 |
| jd__ | lsmola: as eglynn said | 09:18 |
| lsmola | jd__, ok, cool, thank you | 09:19 |
| lsmola | eglynn, and thank you :-) | 09:19 |
| jd__ | lsmola: and to have more success you should prepare some content on how you plan to play with the hardware agent currently being proposed that'll make llu happy | 09:19 |
| lsmola | jd__, ok :-) | 09:21 |
| lsmola | eglynn, added to agenda, will it be aright this way? | 09:29 |
| eglynn | lsmola: looking ... | 09:30 |
| eglynn | lsmola: looks fine to me, further detail will obviously come out in the discussion itself ... | 09:31 |
| swann | ashestakov: IIRC you can still use the cloudwatch api of Heat which can 'forwards' metrics to ceilometer : cfn-push-stats (in instance) -> cw-api -> ceilo | 09:34 |
| swann | ashestakov: but cfn-push-stats is very limited (no mysql) | 09:34 |
| ashestakov | for my case i need custom plugins which can capture custom metrics | 09:35 |
| ashestakov | can ceilometer get metrics from statsd? | 09:36 |
| lsmola | jd__, I will try to talk with tripleo guys about the SNMP also, now sure if it make sense to put it in the Ironic too, as it will know the network info, also I have no idea how SNMP authentication works, if there is any | 09:36 |
| lsmola | jd__, as far as i understand, the SNMP daemon will need to run on every Baremetal (hardware) | 09:37 |
| lsmola | eglynn, yeah I do expect that :-) | 09:38 |
| *** SergeyLu_ has quit IRC | 09:39 | |
| *** eglynn is now known as eglynn-fuse-summ | 09:40 | |
| *** eglynn-fuse-summ is now known as eglynn-fuse-f2f | 09:40 | |
| swann | ashestakov: you can if you write a statsd backend for ceilometer : statsd -> ceilo-collector-udp | 09:40 |
| jd__ | lsmola: yup | 09:40 |
| ashestakov | swann: is statsd support in ceilometer roadmap? | 09:42 |
| swann | ashestakov: i don't think so | 09:42 |
| eglynn-fuse-f2f | jd__: can you give https://review.openstack.org/48069 another look, now that the global reqirements revert has landed? | 09:59 |
| jd__ | eglynn-fuse-f2f: done | 10:01 |
| jd__ | eglynn-fuse-f2f: what's you nick about? | 10:01 |
| eglynn-fuse-f2f | jd__: thank you sir! | 10:01 |
| eglynn-fuse-f2f | jd__: oh, just presenting at a face-to-face meeting in Dublin of the FUSEsource folks (OSGi, ActiveMQ goodness) ... so semi-offline for the morning | 10:02 |
| *** boris-42 has joined #openstack-metering | 10:04 | |
| jd__ | ok :) | 10:04 |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/python-ceilometerclient: Allow to update an alarm partially https://review.openstack.org/48402 | 10:05 |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/ceilometer: Allow to get a disabled alarm https://review.openstack.org/48418 | 10:07 |
| *** neumernace has quit IRC | 10:19 | |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/python-ceilometerclient: Allow to update an alarm partially https://review.openstack.org/48402 | 10:23 |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/ceilometer: Don't load into alarms evaluators disabled alarms https://review.openstack.org/48423 | 10:24 |
| lsmola | llu, hello | 11:04 |
| *** julim has joined #openstack-metering | 11:20 | |
| *** julim has quit IRC | 11:32 | |
| openstackgerrit | A change was merged to openstack/ceilometer: add more test cases to improve the test code coverage #2 https://review.openstack.org/47629 | 11:34 |
| *** eglynn-fuse-f2f has quit IRC | 11:45 | |
| *** SergeyLukjanov has joined #openstack-metering | 11:55 | |
| *** beav has quit IRC | 11:59 | |
| *** eglynn-fuse-f2f has joined #openstack-metering | 12:01 | |
| *** sdake_ has quit IRC | 12:02 | |
| *** sdake has quit IRC | 12:02 | |
| *** sdake has joined #openstack-metering | 12:03 | |
| *** sdake_ has joined #openstack-metering | 12:04 | |
| *** sdake_ has quit IRC | 12:04 | |
| *** sdake_ has joined #openstack-metering | 12:04 | |
| *** ekarlso has quit IRC | 12:20 | |
| jd__ | dhellmann_: around? | 12:25 |
| *** julim has joined #openstack-metering | 12:32 | |
| *** matsuhashi has quit IRC | 12:34 | |
| openstackgerrit | A change was merged to openstack/ceilometer: Add example with return values in API v2 docs https://review.openstack.org/47668 | 12:41 |
| openstackgerrit | Stefano Zilli proposed a change to openstack/python-ceilometerclient: Added support to --os-cacert https://review.openstack.org/46456 | 12:42 |
| *** bpokorny has joined #openstack-metering | 12:57 | |
| *** thomasm has joined #openstack-metering | 12:59 | |
| *** Ruetobas has quit IRC | 12:59 | |
| *** eglynn-fuse-f2f has quit IRC | 13:00 | |
| thomasm | Hey all | 13:02 |
| *** eglynn-fuse-f2f has joined #openstack-metering | 13:14 | |
| *** beav has joined #openstack-metering | 13:15 | |
| *** gordc has joined #openstack-metering | 13:17 | |
| *** Ruetobas has joined #openstack-metering | 13:20 | |
| *** julim has quit IRC | 13:24 | |
| *** gordc has quit IRC | 13:26 | |
| *** shang has quit IRC | 13:27 | |
| *** julim has joined #openstack-metering | 13:27 | |
| *** shang has joined #openstack-metering | 13:32 | |
| *** gordc has joined #openstack-metering | 13:38 | |
| *** matsuhashi has joined #openstack-metering | 13:43 | |
| *** litong has joined #openstack-metering | 13:48 | |
| litong | hi, folks, anybody has trouble to run-tests.sh successfully? after I pull the latest code, it fails with random number of test cases each time I ran it. | 13:59 |
| *** julim has quit IRC | 14:02 | |
| litong | @jd__, julien, u there? do you mind take a quick look at this patch https://review.openstack.org/#/c/48354/ | 14:07 |
| *** changbl has quit IRC | 14:09 | |
| *** julim has joined #openstack-metering | 14:10 | |
| *** matsuhashi has quit IRC | 14:13 | |
| *** eglynn-fuse-f2f has quit IRC | 14:14 | |
| *** eglynn-fuse-f2f has joined #openstack-metering | 14:26 | |
| *** eglynn-fuse-f2f has quit IRC | 14:28 | |
| *** eglynn has joined #openstack-metering | 14:28 | |
| *** anteaya has joined #openstack-metering | 14:39 | |
| litong | anyone out there? | 14:39 |
| litong | extremely quiet today. | 14:39 |
| thomasm | litong, Heyy I'm here | 14:41 |
| thomasm | litong, I haven't tried. Let me pull and try | 14:41 |
| litong | yeah, the problem is in python-keystoneclient. | 14:41 |
| thomasm | oh :( | 14:41 |
| litong | it has been changed and we are using the older class member. that makes our test cases fail if you have latest python-keystoneclient. | 14:42 |
| thomasm | Ah | 14:42 |
| litong | I have opend a bug against us and providing the fixes now. | 14:42 |
| thomasm | Okay, cool. | 14:42 |
| thomasm | litong, I opened a couple of bugs yesterday - one with devstack, since they took away notification_driver when CM service is enabled, and one for neutronclient's quantumclient stable, because it had a mismatch in requirements with global-requirements. | 14:43 |
| thomasm | so, there's been a lot of dependency clashes lately | 14:43 |
| thomasm | Now I have to open another with CI to resolve the fact that the gate jobs for quantumclient are attempting operations assuming neutronclient rather than quantum. | 14:44 |
| thomasm | probably need to change the devstack config. | 14:44 |
| thomasm | Not entirely sure, though. | 14:45 |
| openstackgerrit | litong01 proposed a change to openstack/ceilometer: keystone client changes in AuthProtocol made our test cases failing https://review.openstack.org/48465 | 14:45 |
| litong | @thomasm, that is right, I had to blow out my env yesterday to get neutron to work. | 14:46 |
| litong | @thomasm, that is actually a common problem since we rely on many other projects and when they change their interface, then we have to change. | 14:47 |
| litong | the worst part is that they do not inform us. | 14:47 |
| thomasm | litong, ^^ And that's my problem. =] | 14:47 |
| litong | @thomasm, yeah, this is a problem we will have to live with, I do not see how it gets better in the future. | 14:48 |
| thomasm | litong, haha, we can just work on communicating better - that's about all we can do. | 14:48 |
| thomasm | In a system with so many moving parts, all we can do is improve our communication. | 14:48 |
| litong | what I do not really like is they do not seem to maintain the interface. that is really a bad thing. | 14:48 |
| thomasm | And get better testing around it to find the API issues before they effect us. | 14:49 |
| litong | @thomasm, true. | 14:49 |
| litong | exactly, I think that is why it is so important to have 100% unittest code coverage. | 14:49 |
| thomasm | agreed | 14:50 |
| litong | @thomasm, probably take a look at this patchset https://review.openstack.org/#/c/48465/ | 14:50 |
| thomasm | litong, Just pulled it up. | 14:50 |
| litong | k | 14:50 |
| thomasm | litong, So, they don't break it up into protocol/host anymore? | 14:51 |
| thomasm | It's just request_uri | 14:51 |
| litong | @thomasm, that is right, | 14:51 |
| thomasm | okay | 14:51 |
| litong | if you look at their latest code, that class now only have local vars for host, port, auth_protocol, but no longer the class member. | 14:51 |
| thomasm | litong, gotcha | 14:52 |
| litong | the change was made around august 12. | 14:52 |
| litong | I bet we have not pulling their latest code recently. | 14:52 |
| litong | I mean on our jenkins systems, so the gate jobs have not really caught the problem. | 14:52 |
| litong | need to inform these guys who take care of jenkins. | 14:53 |
| thomasm | good call | 14:53 |
| litong | @thomasm, do not seem to hear anything from cores here. | 14:55 |
| litong | are they all sleeping today? :-) | 14:55 |
| thomasm | Partied too hard last night | 14:56 |
| thomasm | I wonder why they removed the class member status. | 14:58 |
| thomasm | Seems a bit pointless. | 14:58 |
| thomasm | Considering you get the uri anyway - why not make it easy? | 14:58 |
| *** bpokorny has quit IRC | 14:59 | |
| thomasm | litong, Here's the keystone AuthProtocol change: https://github.com/openstack/python-keystoneclient/commit/20e166fd8a943ee3f91ba362a47e9c14c7cc5f4c#diff-bd6505432da8629a1e85b25349a8d5d0 | 15:00 |
| *** ryanpetrello has joined #openstack-metering | 15:01 | |
| litong | right. | 15:02 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: Remove MANIFEST.in https://review.openstack.org/48469 | 15:02 |
| litong | I found that in my machine by run git log command. | 15:02 |
| thomasm | Ahh | 15:02 |
| thomasm | Yeah, my git foo is incredibly junior. =] | 15:03 |
| thomasm | Considering I hadn't used git until about 5 months ago | 15:03 |
| thomasm | or any CLI based VCS. | 15:03 |
| thomasm | for that matter | 15:03 |
| litong | totally understood. | 15:05 |
| thomasm | litong, On a second look, I think that auth_uri would be backwards compatible with the PyPi release of keystoneclient. | 15:09 |
| *** bpokorny has joined #openstack-metering | 15:09 | |
| litong | which is same as auth_protocol? | 15:09 |
| litong | does not seem to be right. | 15:10 |
| thomasm | litong, LR373 - LR384: https://github.com/openstack/python-keystoneclient/commit/20e166fd8a943ee3f91ba362a47e9c14c7cc5f4c#L1R373 | 15:10 |
| thomasm | litong, then again, they could be different...https://github.com/openstack/python-keystoneclient/blob/master/keystoneclient/middleware/auth_token.py#L375 | 15:11 |
| litong | @thomasm, from line 373 (latest code), I think it is safe to use request_uri since it was made by using auth_protocol. | 15:12 |
| litong | auth_uri points to request_uri. | 15:13 |
| thomasm | litong, yeah, I was thinking of eglynn's comment about the latest version in PyPi. | 15:13 |
| thomasm | Sec, let me pull it down and inspect | 15:14 |
| litong | auth_uri may not be same as the request_uri , they are the same only when auth_uri is not set. | 15:14 |
| thomasm | litong, Yep, that's why I mentioned they could be different. :P | 15:15 |
| *** changbl has joined #openstack-metering | 15:16 | |
| thomasm | litong, Because 0.3.2 has self.auth_protocol exposed. | 15:16 |
| litong | right. | 15:17 |
| thomasm | But, devstack installs from git, not pypi | 15:18 |
| thomasm | :\ | 15:18 |
| *** nati_ueno has joined #openstack-metering | 15:19 | |
| eglynn | litong: if you run the tests via tox -epy27 (as opposed to run-tests.sh directly), the 0.3.2 version of keystoneclient should be used, right? | 15:28 |
| eglynn | litong: i.e. you'd only see this issue if running against the master of python-keystoneclient? | 15:29 |
| thomasm | eglynn, litong: correct: (py27)stack@tem-devstack-01:~/ceilometer/.tox/py27$ pip freeze | grep keystoneclient | 15:29 |
| thomasm | python-keystoneclient==0.3.2 | 15:29 |
| litong | @eglynn, I installed brand new devstack last night, then found this problem. | 15:29 |
| eglynn | litong: ok, so if the purpose of spelling out versioned requirements on dependencies like this is to avoid the churn of chasing trunk | 15:30 |
| eglynn | litong: ... probably best to just to use tox to run the tests? | 15:30 |
| litong | since I can not finish run-tests.sh successfully. by looking at the latest keystoneclient code, I found the removal of the member we are using in our tests. | 15:30 |
| eglynn | litong: but wouldn't https://review.openstack.org/48465 break any test runs against python-keystoneclient 0.3.2? | 15:31 |
| eglynn | litong: (which would be the "normal" mode for running the tests IIUC) | 15:32 |
| litong | @eglynn, yes, just relazed that using the new var break older version, need to find a way to make test run successfully against older and new version. | 15:32 |
| litong | I will take a look at that. | 15:33 |
| eglynn | litong: or just always test against the latest released version of keystoneclient (as opposed to supporting both new & old at once ...) | 15:33 |
| thomasm | litong, That's why I was thinking of auth_uri, but that obviously won't work in the case where auth_uri is already set to something else. | 15:33 |
| thomasm | Hmmm | 15:34 |
| litong | @thomasm, that is right. | 15:34 |
| thomasm | Not to mention fundamentally different things | 15:34 |
| litong | just need to check if the object has the member, then do an assert. | 15:34 |
| thomasm | Tests with branch patterns in unit tests are sort of an anti-pattern. | 15:36 |
| eglynn | sileht: I'd appreciate another pair of eyes on https://review.openstack.org/48069 if you've got a minute? | 15:36 |
| eglynn | litong: I don't think we need to go down that road | 15:36 |
| litong | @eglynn, so we ignore it? | 15:36 |
| litong | wait for the next release? | 15:37 |
| eglynn | litong: no we protect ourselves against churn by testing against the latest release only | 15:37 |
| eglynn | litong: then when the next release is cut, we update the code to only work with the latest | 15:37 |
| eglynn | litong: and also update the lower bound on the dependency in the requirements.txt at the same time | 15:37 |
| *** boris-42 has quit IRC | 15:38 | |
| litong | @eglynn, so right now, just leave that bug alone, we can fix that later, is that what you proposed? | 15:38 |
| eglynn | litong: exactly ... it only really becomes a bug when the next keystoneclient release is cut | 15:39 |
| eglynn | litong: in the sense of breaking tests run in the way that they're intended to be run | 15:39 |
| eglynn | litong: (IIUC) | 15:39 |
| litong | @eglynn, that is fine, the only problem is that when you have a fresh devstack install, and run-tests.sh, our test fails. | 15:40 |
| litong | @eglynn, that is fine. I can live with that. | 15:40 |
| eglynn | litong: why not use "tox -epy27" to run the tests instead? | 15:40 |
| eglynn | litong: (which will create an isolated virtual env with exactly the right dependencies installed) | 15:41 |
| litong | @eglynn, yeah, I did, it runs very slow for some reason. | 15:41 |
| eglynn | litong: then your tests run against python-keystoneclient ... happy days, no? | 15:41 |
| eglynn | litong: yeah it is slow to download the first time | 15:41 |
| eglynn | litong: but it's a one-time cost | 15:41 |
| litong | @eglynn, ok, let's just leave the bug open and the patch set hanging. | 15:42 |
| eglynn | litong: i.e. subsequent test runs will use the cached .tox directory, so will complete puch faster | 15:42 |
| litong | we can come back to it later. | 15:42 |
| jd__ | eglynn: the problem I see is that once Havana is out and someone upgrade keystoneclient, ceilometer api is going to be broken until we release a stable release | 15:43 |
| jd__ | eglynn: well, I'm being dramatic, only the tests will be broken, but is that different? :) it will annoy packagers for example | 15:44 |
| eglynn | jd__: could we put an upper bound on the dependency also? ... well just python-keystoneclient==0.3.2 | 15:44 |
| litong | @jd__, I think only tests break, since we are saying we require >=0.3.2 | 15:44 |
| jd__ | eglynn: poor distro packager, so they can't provide a newer version? :( | 15:45 |
| litong | we are not 100% correct. | 15:45 |
| *** nati_ueno has quit IRC | 15:45 | |
| *** SergeyLukjanov has quit IRC | 15:45 | |
| jd__ | eglynn: without saying you need to make a global requirement change for that and I don't think it'll be ok to have all projects pinned to that version :) | 15:46 |
| *** SergeyLukjanov has joined #openstack-metering | 15:46 | |
| eglynn | jd__: seems like we'd be opening a real can of worms trying to anticipate non-backward compat changes coming down the pike in all dependencies | 15:46 |
| eglynn | jd__: maybe we ask the keystone folks to maintain support for the old attributes also? (for backward compat) | 15:47 |
| jd__ | eglynn: hum in that case we could blame python-keystonclient for breaking the compatibility, but since it only affects our unit tests, I am not sure it's worth it | 15:47 |
| jd__ | fixing it in advance, because luckily litong saw this, seems like the best plan to make everybody happy | 15:48 |
| litong | @jd__, @eglynn, so we leave that bug open, take no further action? | 15:48 |
| eglynn | jd__: ok, fair enough ... though at least we'd need to keep support for the old attrs in the test also? | 15:48 |
| jd__ | I'm ok with letting your patch in litong , but it fails anyway currently | 15:48 |
| litong | just want to get a conclusion on this, not trying to push it one way or the other. | 15:48 |
| jd__ | eglynn: ah yeah the tests to pass present and future | 15:49 |
| jd__ | eglynn: ah yeah the tests need to pass present and future | 15:49 |
| eglynn | jd__: fair nuffski | 15:49 |
| eglynn | litong: so sounds like you'll need to update the patch | 15:49 |
| litong | @jd__, since it is using the new var which does not exist in the older version, need to branch it a little, but thomasm does not seem like that approach. | 15:49 |
| eglynn | litong: with conditional code based on getattr etc. | 15:49 |
| jd__ | what! I want code that is present, past and future proof, isn't that easy enough guys? | 15:49 |
| eglynn | LOL :) | 15:50 |
| jd__ | AND I WANT IT YESTERDAY | 15:50 |
| gordc | jd__: you're too easy on them. | 15:50 |
| litong | @eglynn, @jd__, are you guys ok that I do a object member check and then assert? | 15:50 |
| litong | that way , we are not failing anybody. | 15:50 |
| jd__ | litong: yeah branching wouldn't please me either | 15:50 |
| jd__ | litong: be imaginative | 15:50 |
| jd__ | litong: impress me! | 15:50 |
| eglynn | :) | 15:50 |
| litong | @jd__, haha. | 15:51 |
| jd__ | I want ponies and unicorns | 15:51 |
| * jd__ spent a day in WSME code so he's losing it a little | 15:51 | |
| eglynn | jd__: you're in an election cycle, you should be promising the ponies & unicorns ;) | 15:51 |
| jd__ | right | 15:51 |
| jd__ | my round of ponies for everybody, put on my note | 15:51 |
| gordc | dragons & unicorns... i see ponies all the time. | 15:52 |
| eglynn | :) | 15:52 |
| lsmola | llu, hello | 15:52 |
| lsmola | eglynn, do you happen to know, that is llu timezone? | 15:54 |
| *** SergeyLukjanov has quit IRC | 15:56 | |
| eglynn | lsmola: China I think | 15:56 |
| jd__ | I think so too | 15:58 |
| * thomasm hands jd__ a micro-pony | 15:58 | |
| jd__ | thomasm: what? should I talk into the pony? | 15:59 |
| *** shardy is now known as shardy_afk | 15:59 | |
| thomasm | LOL | 15:59 |
| thomasm | jd__, Nahh, I meant just a very very small pony - like one that could fit in your hand. | 15:59 |
| jd__ | ;-) | 15:59 |
| thomasm | Although? that would be interesting. A voice modulator that makes you sound like a horse. | 16:00 |
| *** Ruetobas has quit IRC | 16:01 | |
| thomasm | :P | 16:01 |
| lsmola | eglynn, ok thanks | 16:03 |
| *** Ruetobas has joined #openstack-metering | 16:03 | |
| lsmola | lol | 16:04 |
| *** Ruetobas has quit IRC | 16:08 | |
| *** fnaval_ has joined #openstack-metering | 16:08 | |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/ceilometer: Don't load into alarms evaluators disabled alarms https://review.openstack.org/48423 | 16:11 |
| sileht | eglynn, done | 16:11 |
| *** tasdomas has quit IRC | 16:12 | |
| *** tasdomas has joined #openstack-metering | 16:12 | |
| *** Shaan7 has quit IRC | 16:15 | |
| *** Ruetobas has joined #openstack-metering | 16:15 | |
| litong | can anyone take a look at this patch and give an opinion if that is the right thing to do? | 16:15 |
| litong | https://review.openstack.org/#/c/48354/ | 16:15 |
| litong | @eglynn, @jd__, @gordc, please. | 16:15 |
| jd__ | damn I wrote can haz | 16:17 |
| jd__ | too many lolcats | 16:17 |
| jd__ | I'm out. | 16:17 |
| *** fnaval_ has quit IRC | 16:18 | |
| *** fnaval_ has joined #openstack-metering | 16:19 | |
| litong | @jd__, u still there? | 16:31 |
| *** Shaan7 has joined #openstack-metering | 16:32 | |
| litong | do not think I understand what you are suggesting in your comment in this patch. https://review.openstack.org/#/c/48354/ | 16:33 |
| litong | @jd__, can you explain a bit more? | 16:34 |
| openstackgerrit | litong01 proposed a change to openstack/ceilometer: keystone client changes in AuthProtocol made our test cases failing https://review.openstack.org/48465 | 16:38 |
| *** zigo_ has quit IRC | 16:42 | |
| *** zigo has joined #openstack-metering | 16:42 | |
| *** eglynn has quit IRC | 16:44 | |
| *** Alexei_987 has quit IRC | 16:47 | |
| *** SergeyLukjanov has joined #openstack-metering | 16:49 | |
| litong | @thomasm, I took your advice and used auth_uri. | 17:01 |
| litong | so it should work for both old and new versions of keystoneclient. | 17:01 |
| *** beav has quit IRC | 17:02 | |
| *** terriyu has joined #openstack-metering | 17:12 | |
| *** dhellmann_ is now known as dhellmann | 17:13 | |
| *** beav has joined #openstack-metering | 17:17 | |
| *** nati_ueno has joined #openstack-metering | 17:20 | |
| thomasm | litong, Can we still guarantee that the protocol is the same, since that's what the test appears to be testing? | 17:21 |
| *** eglynn has joined #openstack-metering | 17:21 | |
| *** beav has quit IRC | 17:25 | |
| *** beav has joined #openstack-metering | 17:25 | |
| thomasm | litong, Can we still guarantee that the protocol is the same, since that's what the test appears to be testing? | 17:27 |
| *** litong has quit IRC | 17:37 | |
| *** changbl has quit IRC | 17:55 | |
| *** krtaylor has quit IRC | 17:55 | |
| *** ruhe has joined #openstack-metering | 18:01 | |
| *** changbl has joined #openstack-metering | 18:09 | |
| thomasm | So, can someone help me understand what the entry_points section of setup.cfg is about? https://github.com/openstack/ceilometer/blob/master/setup.cfg#L30-L54 | 18:14 |
| thomasm | What are those mapping? | 18:14 |
| thomasm | I see topics defined in each components notifications.py, so I'm thinking that is where we tell the collection what external notifications to process, so is there where we define what they turn into as samples or something? | 18:15 |
| *** sdake has quit IRC | 18:37 | |
| *** krtaylor has joined #openstack-metering | 18:37 | |
| *** sdake has joined #openstack-metering | 18:37 | |
| *** ruhe has quit IRC | 18:50 | |
| *** krtaylor has quit IRC | 19:01 | |
| *** dhellmann has quit IRC | 19:42 | |
| *** dhellmann has joined #openstack-metering | 19:43 | |
| *** litong has joined #openstack-metering | 20:00 | |
| litong | @thomasm, ping, hi, yes, the auth_uri was not set in the conf. so we are safe. | 20:01 |
| thomasm | litong, Ahhh, okay cool | 20:01 |
| thomasm | litong, thanks =] | 20:01 |
| litong | lost connection earlier. | 20:02 |
| *** evanjfraser has joined #openstack-metering | 20:13 | |
| *** ekarlso has joined #openstack-metering | 20:13 | |
| *** changbl_ has joined #openstack-metering | 20:17 | |
| *** changbl has quit IRC | 20:18 | |
| * gordc wonders why SQLAlchemy 0.8.0b2 is picked up when i run tests... | 20:21 | |
| *** beav has quit IRC | 20:24 | |
| *** eglynn has quit IRC | 20:24 | |
| *** changbl_ has quit IRC | 20:26 | |
| *** changbl has joined #openstack-metering | 20:26 | |
| *** evanjfraser has quit IRC | 20:28 | |
| *** julim has quit IRC | 20:37 | |
| *** boris-42 has joined #openstack-metering | 20:41 | |
| openstackgerrit | litong01 proposed a change to openstack/ceilometer: keystone client changes in AuthProtocol made our test cases failing https://review.openstack.org/48465 | 20:47 |
| *** eglynn has joined #openstack-metering | 20:48 | |
| *** krtaylor has joined #openstack-metering | 20:50 | |
| openstackgerrit | litong01 proposed a change to openstack/ceilometer: keystone client changes in AuthProtocol made our test cases failing https://review.openstack.org/48465 | 20:51 |
| *** boris-42 has quit IRC | 20:53 | |
| *** litong has quit IRC | 20:53 | |
| *** boris-42 has joined #openstack-metering | 20:57 | |
| *** gordc has quit IRC | 21:41 | |
| openstackgerrit | A change was merged to openstack/ceilometer: Remove MANIFEST.in https://review.openstack.org/48469 | 21:44 |
| *** thomasm has quit IRC | 21:51 | |
| *** bpokorny has quit IRC | 21:51 | |
| *** eglynn has quit IRC | 22:00 | |
| *** ryanpetrello has quit IRC | 22:09 | |
| *** changbl has quit IRC | 22:11 | |
| *** SergeyLukjanov has quit IRC | 22:38 | |
| *** boris-42 has quit IRC | 23:00 | |
| *** terriyu has quit IRC | 23:05 | |
| *** nati_ueno has quit IRC | 23:23 | |
| *** changbl has joined #openstack-metering | 23:31 | |
| *** thomasm has joined #openstack-metering | 23:47 | |
| *** fnaval_ has quit IRC | 23:52 | |
| *** nati_ueno has joined #openstack-metering | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!