| *** shaneduan is now known as shaneduan[afk] | 00:13 | |
| *** dina_belova has joined #openstack-metering | 00:14 | |
| *** shaneduan[afk] is now known as shaneduan | 00:15 | |
| *** eglynn-is-tired has quit IRC | 00:16 | |
| *** dina_belova has quit IRC | 00:19 | |
| *** nosnos has joined #openstack-metering | 00:36 | |
| *** osphy has quit IRC | 00:55 | |
| *** dina_belova has joined #openstack-metering | 01:15 | |
| *** changbl has joined #openstack-metering | 01:17 | |
| *** dina_belova has quit IRC | 01:19 | |
| *** shaneduan is now known as shaneduan[afk] | 01:22 | |
| *** shakayumi has quit IRC | 01:22 | |
| *** d34dh0r53 has joined #openstack-metering | 01:59 | |
| *** d34dh0r53 has quit IRC | 02:01 | |
| *** shaneduan[afk] is now known as shaneduan | 02:02 | |
| *** dina_belova has joined #openstack-metering | 02:15 | |
| *** dina_belova has quit IRC | 02:20 | |
| *** shaneduan is now known as shaneduan[afk] | 02:31 | |
| *** evanjfraser_ has joined #openstack-metering | 02:33 | |
| *** evanjfraser has quit IRC | 02:34 | |
| *** evanjfraser_ has quit IRC | 02:38 | |
| openstackgerrit | Haomeng,Wang proposed a change to openstack/ceilometer: Change resource.resource_metadata column type to Text to store entire JSON string without truncating. https://review.openstack.org/45064 | 02:55 |
|---|---|---|
| openstackgerrit | Lianhao Lu proposed a change to openstack/ceilometer: Adding hardware-agent https://review.openstack.org/43072 | 03:06 |
| openstackgerrit | Haomeng,Wang proposed a change to openstack/ceilometer: Change resource.resource_metadata column type to Text https://review.openstack.org/45064 | 03:06 |
| *** dina_belova has joined #openstack-metering | 03:16 | |
| *** dina_belova has quit IRC | 03:20 | |
| *** sandywalsh has joined #openstack-metering | 03:26 | |
| *** sandywalsh_ has quit IRC | 03:27 | |
| *** terriyu has quit IRC | 03:46 | |
| *** evanjfraser has joined #openstack-metering | 03:51 | |
| *** sdake_ has joined #openstack-metering | 04:00 | |
| *** sdake_ has quit IRC | 04:00 | |
| *** sdake_ has joined #openstack-metering | 04:00 | |
| *** dina_belova has joined #openstack-metering | 04:16 | |
| *** gordc has joined #openstack-metering | 04:20 | |
| *** dina_belova has quit IRC | 04:21 | |
| *** dina_belova has joined #openstack-metering | 05:17 | |
| *** evanjfraser has quit IRC | 05:17 | |
| openstackgerrit | Haomeng,Wang proposed a change to openstack/ceilometer: Change resource.resource_metadata column type to Text https://review.openstack.org/45064 | 05:20 |
| *** dina_belova has quit IRC | 05:21 | |
| *** tasdomas_afk is now known as tasdomas | 05:21 | |
| *** shang has quit IRC | 05:22 | |
| *** gordc has quit IRC | 05:39 | |
| *** SergeyLukjanov has joined #openstack-metering | 05:40 | |
| *** dina_belova has joined #openstack-metering | 05:57 | |
| openstackgerrit | Jenkins proposed a change to openstack/ceilometer: Imported Translations from Transifex https://review.openstack.org/45175 | 06:07 |
| openstackgerrit | Lianhao Lu proposed a change to openstack/ceilometer: Fixed nova nofitier unit test https://review.openstack.org/45176 | 06:09 |
| llu | 06:10 | |
| llu | if anyone wakes up, please look at the patch https://review.openstack.org/45176, our Jenkin's got stuck by the latest nova changes. | 06:11 |
| *** eglynn-is-tired has joined #openstack-metering | 06:16 | |
| *** dina_belova has quit IRC | 06:36 | |
| *** evanjfraser has joined #openstack-metering | 06:42 | |
| *** SergeyLukjanov has quit IRC | 06:45 | |
| *** dina_belova has joined #openstack-metering | 06:49 | |
| openstackgerrit | Lianhao Lu proposed a change to openstack/ceilometer: Fixed nova nofitier unit test https://review.openstack.org/45176 | 06:49 |
| *** evanjfraser has quit IRC | 06:50 | |
| *** dina_belova has quit IRC | 06:53 | |
| openstackgerrit | Fengqian.gao proposed a change to openstack/ceilometer: Add pagination support for sqlalchemy database https://review.openstack.org/35454 | 06:55 |
| openstackgerrit | Fengqian.gao proposed a change to openstack/ceilometer: Change pagination query method in mongodb https://review.openstack.org/41869 | 06:55 |
| *** anteaya has quit IRC | 07:04 | |
| *** shardy_afk is now known as shardy | 07:05 | |
| *** eglynn-is-tired has quit IRC | 07:07 | |
| *** shang has joined #openstack-metering | 07:37 | |
| *** eglynn-is-tired has joined #openstack-metering | 07:47 | |
| *** boris-42 has quit IRC | 07:51 | |
| *** graflu0 has joined #openstack-metering | 07:56 | |
| *** lexx has joined #openstack-metering | 07:57 | |
| *** eglynn-is-tired is now known as eglynn | 07:58 | |
| *** graflu0 has quit IRC | 07:58 | |
| sileht | eglynn, hi | 08:01 |
| eglynn | sileht: hey | 08:01 |
| sileht | eglynn, I work a bit on meta-alarm, dhellmann_ have a interresting remarks about this | 08:02 |
| sileht | eglynn, if you have some times to take a look: https://review.openstack.org/#/c/41971/13/ceilometer/api/controllers/v2.py | 08:02 |
| eglynn | sileht: looking now ... | 08:02 |
| llu | hi guys, please look at the patch https://review.openstack.org/45176, our Jenkin's got stuck by the latest nova changes. | 08:03 |
| jd__ | llu: can you fix the comments we made | 08:06 |
| llu | jd__: just saw it. doing it right now. | 08:07 |
| *** Alexei_987 has joined #openstack-metering | 08:07 | |
| jd__ | I'll approve https://review.openstack.org/#/c/44465/ then | 08:07 |
| sileht | jd__, I have updated this for the mongod issue to use the workaround only for test https://review.openstack.org/#/c/44465/ | 08:12 |
| jd__ | I saw | 08:12 |
| jd__ | I'm waiting for llu's patch to go in first as I don't think yours will work on its own wrt our nova notifier failure | 08:13 |
| sileht | jd__, sure | 08:13 |
| openstackgerrit | Lianhao Lu proposed a change to openstack/ceilometer: Fixed nova notifier unit test https://review.openstack.org/45176 | 08:18 |
| llu | i've updated the nova notifier one | 08:19 |
| Alexei_987 | sileht: ping | 08:21 |
| sileht | Alexei_987, o/ | 08:23 |
| Alexei_987 | sileht: regarding this pooling feature - do you plan to keep this workaround? or it's just a temporary solution to get tests working properly? | 08:23 |
| sileht | Alexei_987, I'm looking for more details about how we can fix our tests to really resolve the mongod maxconn issue | 08:23 |
| Alexei_987 | sileht: ah ok I'll try to prepare patch for it today | 08:24 |
| Alexei_987 | __jd sileht and I also have a stupid question: what is db2 backend and why it uses mongo client? | 08:27 |
| sileht | Alexei_987, db2 is a mongodb like from ibm, but we some differences | 08:28 |
| sileht | Alexei_987, I just don't understand why when in test the teardown is called and then the conn.close() of pymongo, the socket is not closed. | 08:29 |
| sileht | Alexei_987, perhaps I need to wait your patch to understand :p I'm just curious | 08:30 |
| Alexei_987 | sileht: the problem is that there are several references on MongoClient object held somewhere | 08:33 |
| Alexei_987 | and object is not destroyed | 08:33 |
| Alexei_987 | and the client object keeps sockets open for later use | 08:33 |
| Alexei_987 | so we need either ensure that MongoClient object is destroyed | 08:34 |
| Alexei_987 | or share this object between tests | 08:34 |
| Alexei_987 | sileht: I plan to go with option 2 since 1 requires huge amount of debuging | 08:34 |
| Alexei_987 | sileht: right now we have our own pooling of mongoclient | 08:35 |
| Alexei_987 | sileht: but it's based on full connection URL | 08:35 |
| Alexei_987 | sileht: which is ok for production deployments | 08:35 |
| Alexei_987 | sileht: but for tests we generate a random database for each test - e.g. URL is different | 08:35 |
| Alexei_987 | sileht: so we need to improve our pooling to use host:port pair without the database part | 08:36 |
| Alexei_987 | sileht: in such case it will work correctly for tests | 08:36 |
| openstackgerrit | A change was merged to openstack/ceilometer: Fixed nova notifier unit test https://review.openstack.org/45176 | 08:50 |
| openstackgerrit | Svetlana Shturm proposed a change to openstack/ceilometer: Fix wrong migrations https://review.openstack.org/45196 | 08:53 |
| jd__ | sileht: https://review.openstack.org/#/c/44465/ needs rebasing | 08:53 |
| openstackgerrit | Mehdi Abaakouk proposed a change to openstack/ceilometer: Disable the pymongo pooling feature for tests https://review.openstack.org/44465 | 08:54 |
| openstackgerrit | Svetlana Shturm proposed a change to openstack/ceilometer: Fix wrong migrations https://review.openstack.org/44539 | 08:55 |
| openstackgerrit | Svetlana Shturm proposed a change to openstack/ceilometer: Fix wrong index in 10 migration for mysql https://review.openstack.org/44539 | 08:57 |
| sileht | Alexei_987, I have already write some code to list the test that seems keeping ref to the object | 08:57 |
| openstackgerrit | A change was merged to openstack/ceilometer: Disable the pymongo pooling feature for tests https://review.openstack.org/44465 | 09:04 |
| *** sandywalsh has quit IRC | 09:10 | |
| openstackgerrit | Lianhao Lu proposed a change to openstack/ceilometer: Added hardware agent's inspector and snmp implementation https://review.openstack.org/43073 | 09:16 |
| openstackgerrit | Lianhao Lu proposed a change to openstack/ceilometer: Added pollsters for the hardware agent https://review.openstack.org/43074 | 09:16 |
| openstackgerrit | Svetlana Shturm proposed a change to openstack/ceilometer: Add testing of migrations from oslo https://review.openstack.org/43872 | 09:19 |
| openstackgerrit | Svetlana Shturm proposed a change to openstack/ceilometer: Implement testing of migrations https://review.openstack.org/44809 | 09:20 |
| openstackgerrit | A change was merged to openstack/ceilometer: Update to tox 1.6 and setup.py develop https://review.openstack.org/45124 | 09:21 |
| openstackgerrit | Alexei Kornienko proposed a change to openstack/ceilometer: Improved MongoClient pooling to avoid out of connections error https://review.openstack.org/45202 | 09:22 |
| openstackgerrit | Yuuichi Fujioka proposed a change to openstack/ceilometer: doc: Fix service name for devstack https://review.openstack.org/45204 | 09:33 |
| openstackgerrit | Alexei Kornienko proposed a change to openstack/ceilometer: Improved MongoClient pooling to avoid out of connections error https://review.openstack.org/45202 | 09:37 |
| openstackgerrit | Shuangtai Tian proposed a change to openstack/ceilometer: Handle the metrics sent by nova notifier https://review.openstack.org/42838 | 09:40 |
| jd__ | we are feature frozen | 09:42 |
| *** asalkeld has quit IRC | 10:00 | |
| openstackgerrit | Eoghan Glynn proposed a change to openstack/ceilometer: Clean-ups related to alarm history patches https://review.openstack.org/45135 | 10:02 |
| *** asalkeld has joined #openstack-metering | 10:02 | |
| *** mmcardle has joined #openstack-metering | 10:18 | |
| openstackgerrit | Swann Croiset proposed a change to openstack/ceilometer: Update openstack.common.policy from oslo-incubator https://review.openstack.org/45210 | 10:26 |
| *** Ruetobas has quit IRC | 10:44 | |
| *** Ruetobas has joined #openstack-metering | 10:44 | |
| openstackgerrit | Swann Croiset proposed a change to openstack/ceilometer: Update openstack.common.policy from oslo-incubator https://review.openstack.org/45210 | 10:45 |
| openstackgerrit | Julien Danjou proposed a change to openstack/python-ceilometerclient: alarm: rename counter_name to meter_name https://review.openstack.org/45076 | 11:00 |
| jd__ | wtf | 11:01 |
| jd__ | https://review.openstack.org/45210 failure scares me | 11:02 |
| lsmola | jd__, hello, can I have two quick questions? | 11:06 |
| *** nosnos has quit IRC | 11:24 | |
| openstackgerrit | Fengqian.gao proposed a change to openstack/ceilometer: Change test case get_alarm_history_on_create https://review.openstack.org/45219 | 11:25 |
| openstackgerrit | Fengqian.gao proposed a change to openstack/ceilometer: Add pagination support for sqlalchemy database https://review.openstack.org/35454 | 11:25 |
| openstackgerrit | Fengqian.gao proposed a change to openstack/ceilometer: Change pagination query method in mongodb https://review.openstack.org/41869 | 11:25 |
| jd__ | lsmola: sure | 11:25 |
| eglynn | dumb question ... | 11:29 |
| eglynn | so if I've got a scenario test failing against sqlalchemy, how can I access the sqlite log to debug the issue? | 11:29 |
| eglynn | (or even suppress deletion of the temporary sqlite DB, so that I can look at it after the fact ...) | 11:30 |
| eglynn | bbiab | 11:31 |
| *** eglynn is now known as hungry-eglynn | 11:31 | |
| swann | jd__: I didn't understand errors, you're ok that's an other reason than the change introduce in policy.py? | 11:33 |
| jd__ | hungry-eglynn: no se | 11:33 |
| *** dhellmann_ is now known as dhellmann | 11:33 | |
| lsmola | jd__, first one about BP workflow in client, when I find something missing in client, can i create BP for this? | 11:33 |
| jd__ | swann: I'm afraid so | 11:33 |
| dhellmann | good morning, team | 11:33 |
| jd__ | hi dhellmann | 11:33 |
| lsmola | jd__, like this https://blueprints.launchpad.net/python-ceilometerclient/+spec/alarm-audit-client-api | 11:34 |
| jd__ | lsmola: sure | 11:34 |
| lsmola | jd__, ok cool, I need to track dependencies, so bp is the best | 11:34 |
| lsmola | jd__, second one about Alerting | 11:34 |
| dhellmann | jd__: Where do we stand? I'm catching up on the backlog, looks like there were some nova-related issues? | 11:35 |
| lsmola | jd__, I walked through BP's and seems there is only Alarms for alerting, meaning when I want to track soemthing, I need to set an Alarm | 11:35 |
| jd__ | dhellmann: yes, this has been fixed, the mongodb connection pooling too, and milestone-proposed as been cut so havana's done | 11:35 |
| jd__ | holy crap | 11:36 |
| dhellmann | well, then | 11:36 |
| jd__ | I can reproduce the failure from https://review.openstack.org/45210 | 11:36 |
| jd__ | so Ceilometer is broken *again* | 11:36 |
| jd__ | good news, removing the nova notifier is now on top of my TODO list | 11:36 |
| dhellmann | yeah | 11:37 |
| * jd__ frowns | 11:37 | |
| dhellmann | shall we turn off the tests? | 11:37 |
| lsmola | jd__, what about notifications like disk failures, network failures, failures in general... Is there some API for getting alerts like this? Or it should all be done by Alarms? | 11:37 |
| dhellmann | delete the whole thing? | 11:37 |
| jd__ | lsmola: if failure can have a threshold, probably | 11:38 |
| jd__ | lsmola: if you consider failure to be 0 when the normal value is 1, that can be done | 11:38 |
| *** dina_belova has joined #openstack-metering | 11:38 | |
| lsmola | jd__, yeah some of them could probably be simulated through alarms | 11:38 |
| jd__ | lsmola: we just don't have code metering this meters | 11:38 |
| jd__ | s/this/these/ | 11:38 |
| jd__ | dhellmann: I think we should fix the test first :) | 11:39 |
| dhellmann | fix it? | 11:39 |
| dhellmann | if we're going to remove it? | 11:39 |
| dhellmann | the notifier, I mean | 11:39 |
| lsmola | jd__, so there are notifications about the failures, you just don't save them for now, right? | 11:39 |
| dhellmann | lsmola: there are no specific notifications for disk or network failures | 11:40 |
| dhellmann | lsmola: however, if you can derive a failure state from the data that is being collected, you could specify an alarm based on those rules | 11:41 |
| jd__ | dhellmann: err I don't think we can remove for Havana and H3 is cut now, so… | 11:41 |
| dhellmann | if h3 is *broken* then it's a bug, and we can fix it | 11:42 |
| jd__ | yes, I just warned ttx about it, I opened a bug https://bugs.launchpad.net/ceilometer/+bug/1221173 | 11:42 |
| dhellmann | ftr, this is why I wanted that notifier to talk to our agent via rpc | 11:42 |
| jd__ | dhellmann: but removing the nova notifier means replacing it actually :) | 11:42 |
| jd__ | and I don't see that happens in a snap | 11:43 |
| dhellmann | no | 11:43 |
| jd__ | no as I'm right or I'm wrong? | 11:43 |
| dhellmann | you're right, it won't be replaced quickly | 11:43 |
| jd__ | ok :) | 11:44 |
| dhellmann | we could build a new notifier that uses rpc to talk to us without depending on nova internals, or we can make nova collect the stats we want | 11:44 |
| jd__ | (that makes me half happy 'cause I'm right and half sad because I wish I was wrong) | 11:44 |
| dhellmann | yeah :-/ | 11:44 |
| lsmola | dhellmann, ok, cool, what about statistics like disk health? I am just askin because we want to show something like this in tuskar-ui | 11:45 |
| lsmola | dhellmann, though is will use hardware agent | 11:45 |
| dhellmann | lsmola: we have a full list of the data we collect at http://docs.openstack.org/developer/ceilometer/measurements.html | 11:46 |
| dhellmann | lsmola: could we do this via email? we're trying to clean up a release candidate right this second | 11:46 |
| lsmola | dhellmann, https://blueprints.launchpad.net/ceilometer/+spec/monitoring-physical-devices | 11:46 |
| lsmola | dhellmann, sorry :-) | 11:46 |
| dhellmann | lsmola: that's for bare metal monitorin? | 11:47 |
| dhellmann | monitoring | 11:47 |
| lsmola | dhellmann, yes | 11:47 |
| lsmola | dhellmann, I suppose it doesn't make sense to check hardware,network... failures for VM's | 11:47 |
| dhellmann | lsmola: someone is going to have to write new data collection plugins for ceilometer, or something else, to send data to us -- we don't have most of what you'll need about the hypervisors, we just collect data about the vms for now | 11:48 |
| dhellmann | IOW, if it's not already on the list of measurements we've published, we don't have it yet | 11:48 |
| *** shang has quit IRC | 11:48 | |
| lsmola | dhellmann, ok, thank you very much | 11:51 |
| *** shang has joined #openstack-metering | 12:00 | |
| *** boris-42 has joined #openstack-metering | 12:04 | |
| *** mmcardle has quit IRC | 12:09 | |
| *** hungry-eglynn is now known as eglynn | 12:11 | |
| eglynn | jd__: "no se" == "not possible"? | 12:11 |
| jd__ | eglynn: "i don't know" in spanish | 12:15 |
| eglynn | jd__: a-ha, cool | 12:15 |
| *** dina_belova has quit IRC | 12:16 | |
| sileht | eglynn, thx for you review | 12:17 |
| eglynn | np, did it make sense? | 12:17 |
| sileht | eglynn, yes I think, I will just take a look into wsme to ensure this is possible without hackism things (I want ensure type check/validation are done correctly) | 12:20 |
| eglynn | sileht: cool! | 12:20 |
| sileht | eglynn, sandy have pointed to me a insteresting code into nova scheduler: https://github.com/openstack/nova/blob/master/nova/scheduler/filters/json_filter.py | 12:22 |
| sileht | eglynn, I think about using a json string into a rules attibutes instead of having combination_alarms and combination_operator | 12:22 |
| eglynn | hmmm, that is an interesting idea | 12:23 |
| eglynn | certainly more flexible | 12:23 |
| *** shang has quit IRC | 12:25 | |
| sileht | I will finish my patch about correct default for normal alarm first, and (If dhellmann is ok) I will create the the new attribute to identify a alarm. | 12:26 |
| dhellmann | sileht: you don't mean posting json encoded in a string in a json payload, do you? | 12:26 |
| sileht | dhellmann, from the API point of view this just a flexible json structure: ie: alarm have a attributes 'rules' with a real json not a string :) | 12:28 |
| dhellmann | sileht: I guess I need to see an example to understand what you mean | 12:29 |
| sileht | dhellmann, sure, example is comming | 12:31 |
| *** zul has quit IRC | 12:32 | |
| *** mmcardle has joined #openstack-metering | 12:37 | |
| *** zul has joined #openstack-metering | 12:37 | |
| sileht | dhellmann, http://paste.openstack.org/show/45801/ I have take the syntax from the nova json_filter | 12:39 |
| sileht | eglynn, ^ | 12:39 |
| dhellmann | sileht: what would the WSME structure for that look like? | 12:40 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: nova_notifier: fix tests wrt instance_update_and_get_original proto change https://review.openstack.org/45222 | 12:40 |
| sileht | dhellmann, good question, If I correctly remember wsme needs to have a static schema of the json ? | 12:41 |
| dhellmann | sileht: yes | 12:41 |
| *** dina_belova has joined #openstack-metering | 12:41 | |
| dhellmann | the idea is to describe the API in a way that is easy to document | 12:41 |
| dhellmann | sileht: I was hoping the "rules" would also apply to the threshold stuff, so we were consistent | 12:42 |
| dhellmann | it seems like there's a rule type, an operator, and then some arguments | 12:42 |
| sileht | dhellmann, in my example, I have missed to removed statistic and threshold | 12:43 |
| dhellmann | threshold, =, [200] | 12:43 |
| dhellmann | alarm_id, in, [(uuid)] | 12:43 |
| dhellmann | alarm_id, in, [(uuid), (uuid)] | 12:43 |
| dhellmann | so I think you need an enum for the rule type, an enum for the operator, and a list of string arguments that are given to the rule | 12:44 |
| dhellmann | your flexibility comes in the arguments | 12:44 |
| dhellmann | the rules can't really be nested, that way, though | 12:44 |
| sileht | threshold is particular because it needs a meter_name, a period and a evaluation_period to be computed | 12:44 |
| dhellmann | yeah | 12:44 |
| dhellmann | well, does evaluation_period not apply in the other cases? | 12:44 |
| sileht | no, | 12:45 |
| dhellmann | I wonder if we should just create a little DSL and not use json | 12:45 |
| dhellmann | that doesn't really solve the problem so much as work around WSME, though | 12:45 |
| sileht | a other problem is the complexity of the evaluator | 12:47 |
| dhellmann | they rules wouldn't be any more complicated | 12:48 |
| dhellmann | *the | 12:48 |
| *** sandywalsh has joined #openstack-metering | 12:51 | |
| sileht | dhellmann, we have two use case: do something if the statistic of a meter reach a threshold, and do something if state of many alarms change | 12:53 |
| sileht | dhellmann, separating both allow to compute it easly without make a complex evaluator | 12:54 |
| openstackgerrit | Julien Danjou proposed a change to openstack/python-ceilometerclient: alarm: rename counter_name to meter_name https://review.openstack.org/45076 | 12:55 |
| dhellmann | if you push all of the settings down into rules, then each rule type can have its own evaluation class | 12:55 |
| dhellmann | jd__: I think I have a fix for the nova notifier issue | 12:55 |
| jd__ | dhellmann: ^^^ | 12:55 |
| sileht | dhellmann, yes but you need to reevalute every thing each time | 12:56 |
| jd__ | dhellmann: https://review.openstack.org/45222 | 12:56 |
| dhellmann | ah | 12:56 |
| dhellmann | jd__: fails pep8 | 12:56 |
| jd__ | pfff | 12:56 |
| jd__ | damn me | 12:56 |
| dhellmann | sileht: you're going to have to do that anyway, right? | 12:56 |
| eglynn | yep the evaluator has a separate strategy per rule type | 12:57 |
| eglynn | as suggested in comment on https://review.openstack.org/#/c/41971/13/ceilometer/api/controllers/v2.py | 12:57 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: nova_notifier: fix tests https://review.openstack.org/45222 | 12:57 |
| sileht | dhellmann, actually the time to get the next state of a alarm is constant | 12:57 |
| eglynn | that should simplify the evaluation logic | 12:57 |
| eglynn | tho' maybe we should re-name threshold_eval to alarm_eval | 12:58 |
| dhellmann | sileht: what is being "reevaluated" every time? I don't understand the objection | 12:58 |
| eglynn | (as combination alarms do not have a threshold associated) | 12:58 |
| dhellmann | eglynn: an alarm type of "alarm_eval"? threshold seems more descriptive | 12:58 |
| eglynn | dhellmann: no I meant the code that implements the evaluation cycle is currently in a module called threshold_evaluation | 12:59 |
| dhellmann | eglynn: got it, yeah, renaming that might make sense | 12:59 |
| eglynn | dhellmann: (whereas for the combination rule type, there is no threshold to evaluate against) | 12:59 |
| dhellmann | right | 12:59 |
| eglynn | what is being re-evaluated each time is the combination of current alarms states, IIUC | 13:00 |
| dhellmann | that's necessary, right? | 13:01 |
| eglynn | absolutely | 13:01 |
| dhellmann | ok, that's what I expected | 13:01 |
| eglynn | not sure if sileht was objecting to it, more just pointing it out? | 13:01 |
| dhellmann | why doesn't a combination alarm have an evaluation period? | 13:01 |
| dhellmann | ok | 13:01 |
| sileht | I'm ok | 13:01 |
| eglynn | dhellmann: evaluation_periods * period = window over which statistics are compared against threshold for normal alarms | 13:02 |
| dhellmann | how does the system know how often to evaluate a combination alarm? | 13:02 |
| eglynn | dhellmann: whereas for combination alarms, stats are not queried | 13:02 |
| dhellmann | ah, got it, that applies to the stats | 13:02 |
| eglynn | dhellmann: it uses the configured eval interval | 13:02 |
| dhellmann | and "period" is how often to check the given alarm | 13:03 |
| eglynn | no period is the timeslot over which the stats are aggregated | 13:03 |
| dhellmann | ok, clearly these docstrings need to be expanded :-) | 13:03 |
| eglynn | so say period = 300s, eval periods = 3 | 13:03 |
| eglynn | and eval interval = 60s | 13:03 |
| *** gordc has joined #openstack-metering | 13:04 | |
| dhellmann | I don't see "interval" in the data structure | 13:04 |
| dhellmann | is that global? | 13:04 |
| eglynn | then every 60s, we look back 15mins into the past to get 3 data points each aggregated over 5 mins | 13:04 |
| dhellmann | across all alarms? | 13:04 |
| eglynn | yep, global config | 13:04 |
| dhellmann | ah, ok, that's what threw me | 13:04 |
| eglynn | we use cloud-watch like terminology | 13:05 |
| eglynn | for the evaluation_periods and period | 13:05 |
| dhellmann | I'm trying to sort out which fields are needed for which types of alarms so I can help with the WSME structure | 13:05 |
| eglynn | which isn't the clearest I admit | 13:05 |
| dhellmann | yeah, it's just some of the docstrings say "foo is the foo for the alarm" | 13:06 |
| dhellmann | i.e., threshold :-) | 13:06 |
| eglynn | I took a first stab at suggesting which fields need to go where in that gerrit comment | 13:06 |
| eglynn | https://review.openstack.org/#/c/41971/13/ceilometer/api/controllers/v2.py | 13:06 |
| dhellmann | the problem is, we can't have 2 types of rule structures | 13:06 |
| eglynn | (may be incomplete tho' ...) | 13:06 |
| sileht | dhellmann, L945 they are 3 list | 13:06 |
| *** litong has joined #openstack-metering | 13:06 | |
| dhellmann | WSME doesn't do subclasses | 13:06 |
| dhellmann | at least not at the API layer | 13:07 |
| eglynn | a-ha, ok | 13:07 |
| dhellmann | so you need an AlarmRule that is flexible enough | 13:07 |
| sileht | dhellmann, Does wsme ComplexType is usable for this ? | 13:07 |
| eglynn | I see | 13:07 |
| dhellmann | I want to add a union type to the complex type, but that won't happen in time for us to use it | 13:07 |
| sileht | dhellmann, oh ok I have understand | 13:07 |
| dhellmann | so if we go back to what I was saying a while ago, with rule_type, operator, and a list of args, can we express the rules that way | 13:09 |
| dhellmann | threshold, =, [stats_name, value] | 13:09 |
| dhellmann | combination, any, [alarm1, alarm2] | 13:10 |
| dhellmann | combination, all, [alarm1, alarm2] | 13:10 |
| dhellmann | the problem is expressing the search metadata for that threshold alarm, it's so much more complex than the combination alarm | 13:10 |
| dhellmann | btw, the matching_metadata needs to support field types so we can do the search properly | 13:11 |
| dhellmann | hmm | 13:11 |
| *** lexx has quit IRC | 13:11 | |
| dhellmann | it needs to be a Query object I think | 13:11 |
| eglynn | the rule needs to be Query object? | 13:12 |
| eglynn | is that rich enough to capture a threshold-oriented rule? | 13:12 |
| dhellmann | I don't mean the rule, I mean the field "matching_metadata" | 13:12 |
| eglynn | i.e. not just a field, op, value but also period etc.? | 13:12 |
| eglynn | dhellmann: a-ha, gotcha | 13:13 |
| dhellmann | a dict of {str:str} is not sufficiently rich to describe a query | 13:13 |
| dhellmann | in fact, you'll never match any numerical values | 13:13 |
| eglynn | yep, that's fair | 13:13 |
| *** graflu0 has joined #openstack-metering | 13:14 | |
| dhellmann | noted on the review for posterity | 13:14 |
| jd__ | https://review.openstack.org/#/c/45222/ is ready to fix the gate btw | 13:15 |
| dhellmann | jd__: approved to fast-track | 13:15 |
| jd__ | thanks dhellmann | 13:16 |
| *** tasdomas is now known as tasdomas_afk | 13:16 | |
| *** tasdomas_afk is now known as tasdomas | 13:17 | |
| dhellmann | eglynn, sileht : changing matching_metadata to a [Query] means you can drop the meter_name field (that's consistent with the statistics API) | 13:20 |
| dhellmann | eglynn, sileht : you may also want to think about whether the object the user posts to the API is the same as the one they get back then they ask for details | 13:20 |
| dhellmann | maybe separate Alarm and AlarmStatus structures make sense, since there are a lot of fields that the user either can't control or don't make sense to set when posting a new alarm | 13:21 |
| sileht | dhellmann, the user is allowed to set the alarm status for debugging | 13:23 |
| openstackgerrit | gordon chung proposed a change to openstack/ceilometer: add tests for _query_to_kwargs func https://review.openstack.org/43796 | 13:23 |
| dhellmann | sileht: so they can trigger the alarm by changing its state? | 13:23 |
| dhellmann | maybe that should be a separate endpoing in the API? | 13:24 |
| sileht | dhellmann, no trigger only set the current state | 13:24 |
| dhellmann | changing the state of the alarm does not trigger it? | 13:24 |
| eglynn | dhellmann: that's correct | 13:25 |
| dhellmann | ok | 13:25 |
| eglynn | dhellmann: the actions are only triggered for a true state transition | 13:25 |
| dhellmann | let me spend some time today putting together a proposal | 13:26 |
| *** anteaya has joined #openstack-metering | 13:26 | |
| dhellmann | I'll write something up in the wiki and send an email | 13:26 |
| dhellmann | because I think what we have now is just exposing the schema for alarms, and is not really providing an API for working with alarms in discrete ways | 13:26 |
| dhellmann | and maybe I can help with the rule expression thing, too | 13:26 |
| jd__ | anyone up for reviewing https://review.openstack.org/#/c/45196/ ? | 13:27 |
| dhellmann | jd__: I looked at that, but don't understand what's going on | 13:27 |
| dhellmann | oh, wait, that's a new one | 13:27 |
| eglynn | gordc: I'd proposed those query tests in https://review.openstack.org/#/c/45135 (as discussed on IRC yesterday) | 13:27 |
| jd__ | that seems like a better version of what I tried to do | 13:27 |
| dhellmann | I thought that was yours | 13:28 |
| dhellmann | yeah | 13:28 |
| eglynn | gordc: d'oh! I hadn't realized you were on it also | 13:28 |
| gordc | eglynn: oh whoops, i didn't see discussion on irc. | 13:28 |
| gordc | eglynn: did you want to pick up the changes i did to the func as well? i'll just drop it then. | 13:29 |
| * eglynn looking ... | 13:29 | |
| gordc | eglynn: relating to this bug https://bugs.launchpad.net/ceilometer/+bug/1217129 | 13:30 |
| openstackgerrit | A change was merged to openstack/ceilometer: nova_notifier: fix tests https://review.openstack.org/45222 | 13:31 |
| dhellmann | jd__: crossing my fingers and giving a +2 | 13:31 |
| eglynn | gordc: k, that's being a bugfix probably best to keep separate from a clean-up patch | 13:32 |
| *** thomasm has joined #openstack-metering | 13:33 | |
| thomasm | Hey all | 13:34 |
| gordc | eglynn: i'll put mine as WIP since yours is related to alarm stuff. | 13:35 |
| eglynn | cool | 13:35 |
| gordc | hey thomasm | 13:37 |
| thomasm | how's the day treating you? | 13:37 |
| gordc | thomasm: just woke up.lol | 13:37 |
| thomasm | gordc, lol | 13:38 |
| gordc | thomasm: i think you can apply some of the comments i made to your patch -- just skip over the ones you weren't sure of. | 13:39 |
| thomasm | gordc, Sure thing. I'll do that this morning | 13:39 |
| gordc | thomasm: cool cool | 13:40 |
| openstackgerrit | A change was merged to openstack/ceilometer: Fix wrong migrations https://review.openstack.org/45196 | 13:41 |
| litong | @thomasm, I added some comments as well, if you need or have any question, please let me know. | 13:42 |
| thomasm | litong, Will do. Thank you. Just getting my code downloaded to work locally. :) | 13:43 |
| *** lexx has joined #openstack-metering | 13:44 | |
| *** graflu0 has quit IRC | 13:45 | |
| thomasm | gordc, regarding those comments in the impl_hbase changes, I guess the reason I went for more readability was because I prefer a more declarative approach. If I can abstract away the imperative, I can just read the code like a book. | 13:47 |
| thomasm | gordc, Of course, each line of code is a culprit, so I dunno - that's just something I do sometimes. :P | 13:47 |
| openstackgerrit | Julien Danjou proposed a change to openstack/python-ceilometerclient: alarm: rename counter_name to meter_name https://review.openstack.org/45076 | 13:48 |
| *** Fengqian has joined #openstack-metering | 13:49 | |
| gordc | thomasm: i'm ok with the breaking them into functions, i'm of the opposite philosophy but its the same logic at the end of the day so no problem for me. | 13:49 |
| thomasm | gordc, No worries, I'm actually just curious. I like to hear the rationale from developers more experienced than I. | 13:50 |
| zul | jd__: ping | 13:51 |
| jd__ | zul: pong | 13:51 |
| zul | jd__: mongodb should be the hitting the cloud archive this week | 13:51 |
| jd__ | zul: aaaaaaaaaaaahhhhhhhhhhhhhhh | 13:51 |
| sileht | zul, \o/ | 13:51 |
| sileht | zul, thanks ! | 13:51 |
| * jd__ runs like crazy | 13:51 | |
| thomasm | Somebody catch 'em! | 13:52 |
| gordc | thomasm: you probably won't want to quote me if it's base on experience then.lol maybe more ceilometer experience? | 13:52 |
| thomasm | gordc, Lol, fair. Both are valuable. I've been doing this type of work for 5 months. Before that I was a front end engineer for 2 years, never touched databases or mid-tier code. | 13:54 |
| thomasm | So, I learn quickly, but I always love to get a bunch of different perspectives to figure out what I really think about it. | 13:54 |
| thomasm | anywhooo, yeah. I guess experience has really shown that oftentimes the abstraction is where the root cause is. | 13:55 |
| thomasm | So, you have to know how it's doing what it's doing. | 13:55 |
| sandywalsh | eglynn, do we have any detailed docs on the alarm stuff (other than the blueprints)? I'm going to spend the day digging into that whole thing. | 14:01 |
| eglynn | sandywalsh: just the blueprints and the code so far | 14:02 |
| eglynn | sandywalsh: "digging in" ... in what sense? | 14:02 |
| dhellmann | eglynn, sileht : https://wiki.openstack.org/wiki/Ceilometer/blueprints/alarm-api | 14:02 |
| dhellmann | sandywalsh: ^^ that's my proposed set of changes to the Alarm structures in the API | 14:03 |
| * sileht is looking | 14:04 | |
| *** graflu0 has joined #openstack-metering | 14:04 | |
| dhellmann | it doesn't include a generic rule, but I think it at least clears up which settings are needed when | 14:04 |
| gordc | thomasm: agreed, always different valid ways to approach stuff. | 14:05 |
| sandywalsh | eglynn, cool ... I'll give them a re-read. I want to understand the requirements better so I can understand how the code maps to it. I can't really make good suggestions for refactoring without really knowing the requirements. | 14:05 |
| sandywalsh | dhellmann, thanks | 14:05 |
| gordc | thomasm: some of the comments i put are probably wrong sometimes so always good to question it if you think it makes no sense. | 14:05 |
| *** krtaylor has joined #openstack-metering | 14:06 | |
| thomasm | gordc, Sure, same here. =] | 14:07 |
| sileht | dhellmann, does I understand that on PUT operation, the user needs to provide the full alarm ? | 14:07 |
| dhellmann | sileht: WSME is going to give you an instance of the class specified as the argument type for the operation, with whatever fields the user provides | 14:08 |
| dhellmann | it doesn't make sense to ask them to PUT part of an Alarm | 14:08 |
| eglynn | dhellmann: how is the mutual exclusivity between threshold_rule and combination_rule captured? | 14:08 |
| dhellmann | eglynn: there's a type setting on the Alarm | 14:09 |
| dhellmann | during icehouse maybe we can add a union type to wsme, and then we can nest the type and rule fields in that | 14:09 |
| eglynn | dhellmann: sure I saw that, but the representation itself could have both set (I thought the idea was to do something union-like) | 14:09 |
| eglynn | k | 14:10 |
| dhellmann | yeah, the threshold thing is just too much more complex | 14:10 |
| dhellmann | let me write down what I was trying to express in irc before... | 14:10 |
| dhellmann | eglynn: https://wiki.openstack.org/wiki/Ceilometer/blueprints/alarm-api#Alternate_Rule_Spec | 14:13 |
| eglynn | hmmm the alternate more generic rule spec isn't very user friendly | 14:15 |
| dhellmann | eglynn: right | 14:15 |
| dhellmann | the two rule types are just *so* different, there's not much in common between them | 14:16 |
| dhellmann | but at least by splitting the two rule types out as in the first alternate, the user can see which fields are needed in which case | 14:16 |
| eglynn | yep | 14:16 |
| eglynn | yes I think that make the mutual exclusivity much clearer | 14:16 |
| eglynn | *makes | 14:17 |
| *** terriyu has joined #openstack-metering | 14:17 | |
| sileht | And this allow to easly extends to new type of rule without breaking API | 14:18 |
| eglynn | true also | 14:18 |
| eglynn | I like | 14:18 |
| eglynn | I'm nore sure we need to change the action representation though | 14:18 |
| eglynn | (as part of the combination work) | 14:19 |
| dhellmann | see also the new DSL section at the bottom, although that's not very well thought out yet | 14:19 |
| dhellmann | I really do think we should go with the classes, but wanted to capture that idea, too | 14:19 |
| eglynn | cool, worth considering, but agreed | 14:19 |
| dhellmann | eglynn: are we likely to add more action types at any point? more states? | 14:19 |
| sileht | dhellmann, sure | 14:19 |
| eglynn | dhellmann: I can't see a real need TBH | 14:20 |
| dhellmann | ok, I wasn't sure if this was a MVP with plans to extend | 14:20 |
| dhellmann | eglynn: I can go along with separate lists of URLs, then | 14:22 |
| eglynn | dhellmann: cool | 14:22 |
| dhellmann | eglynn: I'll update the page to reflect that | 14:23 |
| eglynn | thanks! | 14:23 |
| dhellmann | eglynn: done | 14:25 |
| sileht | dhellmann, eglynn thanks, I'm sure the code will be more clear and more concise now | 14:27 |
| sileht | dhellmann, by force the user to PUT the full alarm, all the default value can be in the wsme model (I have see somehing like wsproperty(xxxx, default='') in the documentation) | 14:28 |
| sileht | dhellmann, eglynn, so can I start to work on this reworker API ? | 14:30 |
| dhellmann | sileht: new alarms should be created with POST, because the user does not control the alarm id. Updates via PUT would require the entire alarm to be provided, in order to be idempotent. If you want to let them change individual pieces of the alarm, then some other mechanism is needed | 14:30 |
| dhellmann | sileht: that other mechanism could be PUT to just the field in question (/alarms/<uuid>/state) | 14:31 |
| sileht | dhellmann, I clearly prefer use wsme tools, and force it to PUT the full alarm | 14:31 |
| dhellmann | sure, that makes sense for updating the definition | 14:31 |
| eglynn | sileht: do you mean, should you start on reworking the alarm API as discussed above? | 14:32 |
| sileht | dhellmann, cool a new endpoint :p, this can be done later | 14:32 |
| sileht | eglynn, yes | 14:32 |
| eglynn | sileht: absolutely! :) | 14:32 |
| dhellmann | sileht: right, if we find we need to make some operations more convenient | 14:32 |
| eglynn | dhellmann: offtopic ... I tried to address your non-blocking suggestions for the alarm history patches here: https://review.openstack.org/45135 | 14:32 |
| dhellmann | eglynn: looking | 14:34 |
| *** shang has joined #openstack-metering | 14:40 | |
| openstackgerrit | Julien Danjou proposed a change to openstack/python-ceilometerclient: alarm: rename counter_name to meter_name https://review.openstack.org/45076 | 14:42 |
| *** dina_belova has quit IRC | 14:43 | |
| *** dina_belova has joined #openstack-metering | 14:45 | |
| *** dina_bel_ has joined #openstack-metering | 14:47 | |
| *** dina_belova has quit IRC | 14:47 | |
| *** graflu0 has quit IRC | 14:52 | |
| openstackgerrit | Eoghan Glynn proposed a change to openstack/ceilometer: Alarm history storage implementation for sqlalchemy https://review.openstack.org/45244 | 14:57 |
| *** Fengqian has quit IRC | 15:06 | |
| *** lexx has quit IRC | 15:10 | |
| *** dina_bel_ has quit IRC | 15:10 | |
| *** SergeyLu_ has joined #openstack-metering | 15:23 | |
| *** changbl has quit IRC | 15:23 | |
| *** SergeyLu_ has quit IRC | 15:23 | |
| *** SergeyLu_ has joined #openstack-metering | 15:24 | |
| *** shang has quit IRC | 15:29 | |
| *** SergeyLu_ has quit IRC | 15:29 | |
| *** SergeyLukjanov has joined #openstack-metering | 15:30 | |
| *** tasdomas is now known as tasdomas_afk | 15:35 | |
| eglynn | meh, we're seeing that IndexError from wsme/type.py again | 15:47 |
| eglynn | i.e. ... if bases[0] is not object and '__registry__' not in dct: | 15:47 |
| eglynn | e.g. in https://jenkins02.openstack.org/job/gate-ceilometer-python26/471/consoleFull | 15:48 |
| eglynn | I thought that had been resolved by constraining the six dependency version not to clash with WSME? | 15:48 |
| jd__ | O_o | 15:50 |
| jd__ | eglynn: this patch changes six bounding | 15:50 |
| eglynn | grrr! | 15:51 |
| *** boris-42 has quit IRC | 15:51 | |
| jd__ | be careful of devstack, it seems it changes these file magically now | 15:51 |
| eglynn | (yep, I took it from devstack) | 15:51 |
| jd__ | dumb idea IMHO | 15:51 |
| eglynn | sorry for the noise, fixing now ... | 15:51 |
| jd__ | :) | 15:51 |
| eglynn | yeah, agreed, dumb and v. annoying ... | 15:51 |
| *** shaneduan[afk] is now known as shaneduan | 15:53 | |
| openstackgerrit | Eoghan Glynn proposed a change to openstack/ceilometer: Alarm history storage implementation for sqlalchemy https://review.openstack.org/45244 | 15:55 |
| *** Ruetobas has quit IRC | 16:01 | |
| *** shang has joined #openstack-metering | 16:01 | |
| *** Ruetobas has joined #openstack-metering | 16:04 | |
| *** shang has quit IRC | 16:07 | |
| *** shang has joined #openstack-metering | 16:08 | |
| *** Ruetobas has quit IRC | 16:08 | |
| *** SergeyLukjanov has quit IRC | 16:13 | |
| openstackgerrit | A change was merged to openstack/ceilometer: Clean-ups related to alarm history patches https://review.openstack.org/45135 | 16:13 |
| openstackgerrit | gordon chung proposed a change to openstack/ceilometer: add MAINTAINERS file https://review.openstack.org/45264 | 16:13 |
| *** Ruetobas has joined #openstack-metering | 16:15 | |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: Fix wrong migrations https://review.openstack.org/45265 | 16:15 |
| openstackgerrit | Julien Danjou proposed a change to openstack/ceilometer: nova_notifier: fix tests https://review.openstack.org/45266 | 16:15 |
| openstackgerrit | gordon chung proposed a change to openstack/ceilometer: add MAINTAINERS file https://review.openstack.org/45264 | 16:19 |
| openstackgerrit | Thomas Maddox proposed a change to openstack/ceilometer: Fix to return latest resource metadata https://review.openstack.org/44277 | 16:22 |
| *** shaneduan is now known as shaneduan[afk] | 16:30 | |
| *** shang has quit IRC | 16:34 | |
| *** shaneduan[afk] is now known as shaneduan | 16:40 | |
| *** mmcardle has quit IRC | 16:42 | |
| openstackgerrit | Alex Gaynor proposed a change to openstack/python-ceilometerclient: Added support for running the tests under PyPy with tox https://review.openstack.org/45276 | 16:57 |
| *** boris-42 has joined #openstack-metering | 16:59 | |
| *** Alexei_987 has quit IRC | 17:04 | |
| *** SergeyLukjanov has joined #openstack-metering | 17:10 | |
| *** yjiang5_away is now known as yjiang5 | 17:11 | |
| *** dhellmann is now known as dhellmann_ | 17:11 | |
| *** nealph has joined #openstack-metering | 17:15 | |
| *** changbl has joined #openstack-metering | 17:16 | |
| *** shang has joined #openstack-metering | 17:20 | |
| *** dina_belova has joined #openstack-metering | 17:21 | |
| *** dina_belova has quit IRC | 17:25 | |
| *** dina_belova has joined #openstack-metering | 17:26 | |
| *** mmcardle has joined #openstack-metering | 17:42 | |
| openstackgerrit | Thomas Maddox proposed a change to openstack/ceilometer: Fix to return latest resource metadata https://review.openstack.org/44277 | 17:48 |
| *** mmcardle has quit IRC | 17:50 | |
| thomasm | Hmmm? I didn't see any tests actually run in the two gate-tempest tests that failed on my patch? are there known issues with these tests? | 17:52 |
| litong | @thomasm, I was surprised as well seeing that results. the last two tests are new to me. | 18:02 |
| thomasm | litong, It looks like keystone didn't start for devstack and failed the build. :\ | 18:03 |
| thomasm | For both of them | 18:04 |
| litong | yeah, that was from the log file, but I do not remember seeing the last two tests for other patchsets though. | 18:04 |
| thomasm | Yeah, I think they were just added today | 18:04 |
| litong | hope that the tests will run successfully this time. | 18:05 |
| thomasm | Me too! *crosses fingers* | 18:05 |
| litong | I will be back online in about one hour. | 18:06 |
| thomasm | Okay | 18:06 |
| *** litong has quit IRC | 18:06 | |
| openstackgerrit | Eoghan Glynn proposed a change to openstack/ceilometer: Alarm history storage implementation for sqlalchemy https://review.openstack.org/45244 | 18:11 |
| thomasm | Passed this time. Yay! | 18:16 |
| *** mmcardle has joined #openstack-metering | 18:17 | |
| *** mmcardle has quit IRC | 18:21 | |
| *** sdake has quit IRC | 18:22 | |
| *** eglynn has quit IRC | 18:22 | |
| *** sdake has joined #openstack-metering | 18:22 | |
| *** sdake has joined #openstack-metering | 18:22 | |
| openstackgerrit | gordon chung proposed a change to stackforge/pycadf: ensure unique id and timestamp defaults https://review.openstack.org/45301 | 18:40 |
| *** dina_belova has quit IRC | 18:40 | |
| *** litong has joined #openstack-metering | 18:40 | |
| thomasm | litong, They passed this time | 18:43 |
| litong | @thomasm, good job. | 18:43 |
| litong | glad it did. | 18:43 |
| thomasm | likewise | 18:44 |
| openstackgerrit | gordon chung proposed a change to stackforge/pycadf: ensure unique id and timestamp defaults https://review.openstack.org/45301 | 18:44 |
| openstackgerrit | A change was merged to stackforge/pycadf: ensure unique id and timestamp defaults https://review.openstack.org/45301 | 18:53 |
| openstackgerrit | A change was merged to openstack/python-ceilometerclient: Added support for running the tests under PyPy with tox https://review.openstack.org/45276 | 18:53 |
| *** SergeyLukjanov has quit IRC | 18:54 | |
| *** eglynn has joined #openstack-metering | 18:58 | |
| *** sdake_ has quit IRC | 18:59 | |
| *** sdake_ has joined #openstack-metering | 19:01 | |
| *** sdake_ has quit IRC | 19:01 | |
| *** sdake_ has joined #openstack-metering | 19:01 | |
| *** shang has quit IRC | 19:04 | |
| openstackgerrit | Dan Prince proposed a change to openstack/ceilometer: Force MySQL to use InnoDB. https://review.openstack.org/45306 | 19:04 |
| *** sdake_ has quit IRC | 19:10 | |
| *** krtaylor has quit IRC | 19:12 | |
| *** sdake_ has joined #openstack-metering | 19:13 | |
| *** sdake_ has quit IRC | 19:13 | |
| *** sdake_ has joined #openstack-metering | 19:13 | |
| *** mmcardle has joined #openstack-metering | 19:17 | |
| *** krtaylor has joined #openstack-metering | 19:17 | |
| *** mmcardle has quit IRC | 19:22 | |
| *** eglynn has quit IRC | 19:23 | |
| *** sdake_ has quit IRC | 19:33 | |
| *** dina_belova has joined #openstack-metering | 19:41 | |
| *** dina_belova has quit IRC | 19:45 | |
| *** dina_belova has joined #openstack-metering | 19:51 | |
| *** dina_belova has quit IRC | 19:55 | |
| thomasm | gordc: I was thinking we could start applying a builder pattern to that whole driver, since there's so much boilerplate query filtering and what-not with variations that could be built off of the basic filtered object? | 19:57 |
| thomasm | (mongodb) | 19:57 |
| openstackgerrit | Thomas Maddox proposed a change to openstack/ceilometer: Fix to return latest resource metadata https://review.openstack.org/44277 | 19:59 |
| *** sandywalsh has quit IRC | 20:00 | |
| openstackgerrit | Thomas Maddox proposed a change to openstack/ceilometer: Fix to return latest resource metadata https://review.openstack.org/44277 | 20:01 |
| thomasm | Man? I'm just being a baddie today. Forgot the commit message change too. | 20:03 |
| thomasm | gordc, Updated that patch. Sorry for the missed changes! | 20:09 |
| *** anteaya has quit IRC | 20:11 | |
| *** sandywalsh has joined #openstack-metering | 20:14 | |
| *** mmcardle has joined #openstack-metering | 20:19 | |
| *** sandywalsh has quit IRC | 20:20 | |
| openstackgerrit | A change was merged to openstack/ceilometer: add MAINTAINERS file https://review.openstack.org/45264 | 20:22 |
| *** dhellmann_ is now known as dhellmann | 20:23 | |
| *** mmcardle has quit IRC | 20:23 | |
| *** sdake_ has joined #openstack-metering | 20:25 | |
| *** sdake_ has quit IRC | 20:25 | |
| *** sdake_ has joined #openstack-metering | 20:25 | |
| *** mmcardle has joined #openstack-metering | 20:28 | |
| *** sandywalsh has joined #openstack-metering | 20:33 | |
| *** evanjfraser has joined #openstack-metering | 20:45 | |
| *** mmcardle has quit IRC | 20:45 | |
| *** dina_belova has joined #openstack-metering | 20:51 | |
| *** dina_belova has quit IRC | 20:56 | |
| *** shaneduan is now known as shaneduan[afk] | 21:04 | |
| *** shaneduan[afk] is now known as shaneduan | 21:05 | |
| *** mmcardle has joined #openstack-metering | 21:15 | |
| gordc | thomasm: whoops, was on another machine. | 21:15 |
| thomasm | gordc, No worries | 21:16 |
| gordc | thomasm: builder pattern sounds interesting... something we can explore in Icehouse. | 21:16 |
| gordc | thanks for applying fixes btw | 21:16 |
| thomasm | gordc, Sure thing. Yeah, the way we apply sorting/pagination is a little mixed, it feels like | 21:17 |
| thomasm | gordc, I'd like to be able to do query.sort(sort_keys).paginate(limit=100) | 21:18 |
| thomasm | or something like that | 21:18 |
| gordc | thomasm: :) yeah there's a lot of discussion around pagination if you look at the current patchsets | 21:18 |
| thomasm | I noticed that. I also saw one discussion around removing it completely. | 21:18 |
| *** mmcardle has quit IRC | 21:20 | |
| openstackgerrit | gordon chung proposed a change to openstack/ceilometer: add tests for _query_to_kwargs func https://review.openstack.org/43796 | 21:20 |
| thomasm | Of course, as soon as I mention that, I can't find the discussion I'm referring to. | 21:20 |
| gordc | thomasm: lol, yeah looks like we'll have time to talk about pagination since it's blocked by feature freeze | 21:22 |
| thomasm | gordc, gotcha | 21:23 |
| openstackgerrit | Dan Prince proposed a change to openstack/ceilometer: Force MySQL to use InnoDB! https://review.openstack.org/45306 | 21:33 |
| *** shaneduan is now known as shaneduan[afk] | 21:46 | |
| *** shaneduan[afk] is now known as shaneduan | 21:47 | |
| *** thomasm has quit IRC | 21:47 | |
| *** changbl has quit IRC | 21:51 | |
| *** dina_belova has joined #openstack-metering | 21:52 | |
| *** dina_belova has quit IRC | 21:55 | |
| *** shaneduan is now known as shaneduan[afk] | 22:04 | |
| *** shaneduan[afk] is now known as shaneduan | 22:05 | |
| *** mmcardle has joined #openstack-metering | 22:16 | |
| *** mmcardle has quit IRC | 22:21 | |
| *** shaneduan is now known as shaneduan[afk] | 22:26 | |
| *** shaneduan[afk] is now known as shaneduan | 22:30 | |
| *** shaneduan is now known as shaneduan[afk] | 22:31 | |
| *** shaneduan[afk] is now known as shaneduan | 22:31 | |
| *** evanjfraser_ has joined #openstack-metering | 22:45 | |
| *** evanjfraser has quit IRC | 22:49 | |
| *** thomasm has joined #openstack-metering | 22:51 | |
| *** dina_belova has joined #openstack-metering | 22:52 | |
| *** dina_belova has quit IRC | 22:57 | |
| *** evanjfraser_ has quit IRC | 23:03 | |
| *** shaneduan is now known as shaneduan[afk] | 23:03 | |
| openstackgerrit | gordon chung proposed a change to openstack/ceilometer: validate counter_type when posting samples https://review.openstack.org/45344 | 23:09 |
| *** nijaba has quit IRC | 23:12 | |
| *** nijaba has joined #openstack-metering | 23:13 | |
| *** evanjfraser has joined #openstack-metering | 23:15 | |
| *** boris-42 has quit IRC | 23:16 | |
| *** mmcardle has joined #openstack-metering | 23:17 | |
| *** mmcardle has quit IRC | 23:22 | |
| *** shaneduan[afk] is now known as shaneduan | 23:26 | |
| openstackgerrit | gordon chung proposed a change to openstack/ceilometer: validate counter_type when posting samples https://review.openstack.org/45344 | 23:31 |
| *** evanjfraser has quit IRC | 23:32 | |
| *** gordc has quit IRC | 23:32 | |
| *** evanjfraser has joined #openstack-metering | 23:42 | |
| *** dina_belova has joined #openstack-metering | 23:53 | |
| *** dina_belova has quit IRC | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!