| *** fnaval has quit IRC | 00:18 | |
| *** flwang_ has joined #openstack-ceilometer | 00:35 | |
| *** flwang_ has quit IRC | 00:40 | |
| *** amalagon has quit IRC | 01:01 | |
| *** julim has quit IRC | 01:03 | |
| *** _cjones_ has quit IRC | 01:22 | |
| *** _cjones_ has joined #openstack-ceilometer | 01:22 | |
| *** _cjones_ has quit IRC | 01:27 | |
| *** nosnos has joined #openstack-ceilometer | 01:44 | |
| *** amalagon has joined #openstack-ceilometer | 01:51 | |
| *** Ruetobas has quit IRC | 02:03 | |
| *** Ruetobas has joined #openstack-ceilometer | 02:11 | |
| *** giroro_ has joined #openstack-ceilometer | 02:13 | |
| *** Ruetobas has quit IRC | 02:15 | |
| *** ccrouch has joined #openstack-ceilometer | 02:28 | |
| *** changbl has joined #openstack-ceilometer | 02:30 | |
| *** flwang_ has joined #openstack-ceilometer | 02:36 | |
| *** flwang_ has quit IRC | 02:41 | |
| *** shardy has quit IRC | 03:01 | |
| *** Alice911 has joined #openstack-ceilometer | 03:12 | |
| *** nosnos has quit IRC | 03:20 | |
| *** nati_ueno has quit IRC | 03:22 | |
| *** renatoarmani has joined #openstack-ceilometer | 03:43 | |
| *** renatoarmani has quit IRC | 03:44 | |
| *** flwang has joined #openstack-ceilometer | 04:03 | |
| flwang | jd__: any lucky you there? | 04:04 |
|---|---|---|
| *** ityaptin has quit IRC | 04:08 | |
| *** nosnos has joined #openstack-ceilometer | 04:13 | |
| *** nosnos has quit IRC | 04:24 | |
| *** r3pl4y has quit IRC | 04:43 | |
| *** yatin has joined #openstack-ceilometer | 04:54 | |
| *** underyx|off is now known as underyx | 04:58 | |
| *** ildikov has quit IRC | 05:02 | |
| *** underyx is now known as underyx|off | 05:13 | |
| *** underyx|off is now known as underyx | 05:17 | |
| openstackgerrit | Nejc Saje proposed a change to openstack/ceilometer-specs: Arithmetic transformer spec https://review.openstack.org/105467 | 05:27 |
| *** underyx is now known as underyx|off | 05:28 | |
| *** ildikov has joined #openstack-ceilometer | 05:42 | |
| *** inc0 has joined #openstack-ceilometer | 05:45 | |
| *** changbl has quit IRC | 05:51 | |
| *** nacim has quit IRC | 06:01 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ceilometer: Imported Translations from Transifex https://review.openstack.org/104974 | 06:01 |
| *** amalagon has quit IRC | 06:30 | |
| *** ajc_ has joined #openstack-ceilometer | 06:35 | |
| *** ajc_ has left #openstack-ceilometer | 06:35 | |
| *** ajc_ has joined #openstack-ceilometer | 06:35 | |
| openstackgerrit | Stefano Zilli proposed a change to openstack/ceilometer: Parametrize table_prefix_separator in hbase https://review.openstack.org/100829 | 06:36 |
| *** Longgeek has joined #openstack-ceilometer | 06:36 | |
| *** flwang_ has joined #openstack-ceilometer | 06:38 | |
| *** r3pl4y has joined #openstack-ceilometer | 06:39 | |
| *** idegtiarov has quit IRC | 06:42 | |
| *** flwang_ has quit IRC | 06:43 | |
| *** idegtiarov has joined #openstack-ceilometer | 06:46 | |
| *** amalagon has joined #openstack-ceilometer | 06:47 | |
| *** underyx|off is now known as underyx | 07:10 | |
| *** eglynn has joined #openstack-ceilometer | 07:16 | |
| *** eglynn has quit IRC | 07:20 | |
| *** eglynn has joined #openstack-ceilometer | 07:21 | |
| *** shardy has joined #openstack-ceilometer | 07:47 | |
| *** ildikov has quit IRC | 07:55 | |
| *** ildikov has joined #openstack-ceilometer | 07:56 | |
| *** safchain has joined #openstack-ceilometer | 07:58 | |
| *** eglynn_ has joined #openstack-ceilometer | 07:58 | |
| *** eglynn has quit IRC | 07:59 | |
| *** nacim has joined #openstack-ceilometer | 08:04 | |
| *** flwang_ has joined #openstack-ceilometer | 08:39 | |
| *** flwang_ has quit IRC | 08:43 | |
| *** flwang_ has joined #openstack-ceilometer | 08:49 | |
| flwang_ | eglynn_: piing | 09:04 |
| eglynn_ | flwang_: 'sup? | 09:05 |
| flwang_ | eglynn_: quick question, just wanna confirm, if I just add a new pollster or enhance an existed one, should I create a new spec? | 09:06 |
| flwang_ | eglynn_: I think a small blueprint is deserved, but I'm not sure if a spec is needed | 09:06 |
| eglynn_ | flwang_: grey area for the new case, probably not required to enhance pre-existing | 09:07 |
| eglynn_ | flwang_: new BPs need a spec, so it's both or none really | 09:07 |
| flwang_ | eglynn_: yep, grey area, that's why I ask | 09:08 |
| flwang_ | my team are going to upsteam a cinder pollster we're using | 09:09 |
| *** cdent has joined #openstack-ceilometer | 09:09 | |
| eglynn_ | flwang_: are there very similar existing pollsters? ==> in which case, spec probably not needed | 09:09 |
| flwang_ | so is it ok without bp? just upsteam the patch? | 09:09 |
| eglynn_ | flwang_: new pollsters polling an previously unused API? ==> in which case, spec probably needed | 09:09 |
| flwang_ | yep, it's very similar with existed, just get the volume size | 09:09 |
| flwang_ | for billing | 09:10 |
| eglynn_ | flwang_: "just BP" isn't really an option anymore ... so BP+spec, or just a RFE-style bug | 09:10 |
| eglynn_ | flwang_: sounds like an RFE bug will suffice in this case | 09:11 |
| flwang_ | ok, got it | 09:11 |
| flwang_ | RFE = ? I know I have asked this question :) | 09:12 |
| openstackgerrit | liusheng proposed a change to openstack/ceilometer: Fix SQL exception getting statitics with metaquery https://review.openstack.org/102150 | 09:12 |
| eglynn_ | flwang_: RFE == "request for enhancement" | 09:13 |
| eglynn_ | flwang_: ... i.e. a bug that asks for a small feature to be added | 09:13 |
| flwang_ | awesome | 09:13 |
| flwang_ | I really like the term :) | 09:13 |
| * eglynn_ raps himself on knuckles for indiscriminate use of a TLA ... ;) | 09:14 | |
| flwang_ | eglynn_: cheers | 09:14 |
| eglynn_ | np! | 09:14 |
| cdent | eglynn needs a bigger jar for his bizspeak sins | 09:14 |
| eglynn_ | cdent: ... or start using paypal to pay into a virtual jar ;) | 09:18 |
| *** yatin_ has joined #openstack-ceilometer | 09:20 | |
| flwang_ | eglynn_: and here is another topic, as for this https://github.com/openstack/ceilometer/blob/master/ceilometer/network/notifications.py#L161 | 09:21 |
| flwang_ | eglynn_: i'm going to rename it as "traffic", since it's not a real bandwidth | 09:21 |
| flwang_ | eglynn_: and i will do some tranfomation to get the average 'bandwidth' from the 'traffic', any concern? | 09:22 |
| openstackgerrit | liusheng proposed a change to openstack/ceilometer: Fix SQL exception getting statitics with metaquery https://review.openstack.org/102150 | 09:23 |
| eglynn_ | flwang_: do you mean change the logic added here ... | 09:24 |
| flwang_ | eglynn_: i have confirmed with the neutron guy, safchain who is the original author of neutron metering agent | 09:24 |
| eglynn_ | https://github.com/openstack/ceilometer/commit/3636a044a6ec04eea4ace7cbb3b377d2f412598e#diff-26cb8f56ce7ad5b2acc90aaddd268e44R154 | 09:24 |
| *** yatin has quit IRC | 09:24 | |
| eglynn_ | so that meter name was added way back in the Havana timeframe, or? | 09:24 |
| flwang_ | yes | 09:25 |
| flwang_ | but the name is not correct, and we haven't release it in ceilometer measurements document | 09:25 |
| flwang_ | so I'd like to correct it before release it | 09:25 |
| eglynn_ | hmmm, not documented you say? | 09:26 |
| flwang_ | yes | 09:26 |
| eglynn_ | to what extent is "bandwidth" incorrect? | 09:26 |
| eglynn_ | completely misleading? | 09:26 |
| eglynn_ | or just personal terminology preference? | 09:26 |
| flwang_ | I think so | 09:26 |
| flwang_ | generally, the unit of bandwidth should be B/second | 09:27 |
| *** Longgeek has quit IRC | 09:27 | |
| flwang_ | or I would say current name will confuse the user | 09:27 |
| eglynn_ | a-ha, yes, I see that now ... unit='B' | 09:28 |
| flwang_ | safchain: any lucky you around? | 09:28 |
| eglynn_ | ... is type=sample.TYPE_DELTA accurate in that case? | 09:28 |
| eglynn_ | ... i.e. the number of bytes since the last what? | 09:28 |
| flwang_ | it's correct, because the volume/data is collected in the period, between two timestamps | 09:29 |
| eglynn_ | do we capture that timestamp range? | 09:29 |
| flwang_ | http://paste.openstack.org/show/85401/ | 09:31 |
| eglynn_ | one sec | 09:31 |
| flwang_ | yes | 09:32 |
| eglynn_ | flwang_: so for example in that paste ... 65353079.0 is accurately the number of bytes passed between 2013-12-23T09:44:43.592000 and 2013-12-23T09:39:14.232000? | 09:34 |
| *** yatin__ has joined #openstack-ceilometer | 09:35 | |
| eglynn_ | flwang_: i.e. the notifications from neutron accurately count data since the *last* notification | 09:35 |
| *** yatin__ is now known as yatin | 09:36 | |
| eglynn_ | flwang_: ... in that case, yep please go ahead and 1. change meter name 2. document this meter and 3. keep the delta type | 09:36 |
| flwang_ | 65353079.0 collected between 2013-12-23T09:34:13.668000 and 2013-12-23T09:39:14.232000 | 09:37 |
| *** yatin_ has quit IRC | 09:37 | |
| flwang_ | so yes, you're right | 09:37 |
| flwang_ | awesome | 09:38 |
| eglynn_ | flwang_: ... thank you sir! | 09:38 |
| flwang_ | glad to see we can make progress, cheers | 09:38 |
| flwang_ | btw, do you think we should provide an average bandwidth metric? | 09:39 |
| flwang_ | like using 65353079.0 / 300 sec | 09:39 |
| *** yatin has quit IRC | 09:40 | |
| flwang_ | just think aloud | 09:40 |
| *** yatin has joined #openstack-ceilometer | 09:41 | |
| eglynn_ | flwang_: possibly, but preferable to do such derivation in a transformer | 09:43 |
| eglynn_ | flwang_: ... as we do for cpu_util etc. | 09:44 |
| flwang_ | eglynn_: yep, make sense | 09:44 |
| eglynn_ | DinaBelova, Vadim: a heads-up on ... http://lists.openstack.org/pipermail/openstack-dev/2014-July/039709.html | 09:46 |
| *** vrovachev has joined #openstack-ceilometer | 09:46 | |
| *** yatin has quit IRC | 09:46 | |
| *** yatin has joined #openstack-ceilometer | 09:47 | |
| *** yatin has quit IRC | 09:54 | |
| *** yatin has joined #openstack-ceilometer | 09:54 | |
| * cdent has bought popcorn | 09:56 | |
| *** ildikov has quit IRC | 09:57 | |
| *** Longgeek has joined #openstack-ceilometer | 09:58 | |
| *** underyx is now known as underyx|off | 10:02 | |
| *** yatin_ has joined #openstack-ceilometer | 10:02 | |
| *** _nadya_ has joined #openstack-ceilometer | 10:05 | |
| *** Longgeek has quit IRC | 10:05 | |
| *** yatin has quit IRC | 10:05 | |
| *** _nadya_ has quit IRC | 10:06 | |
| eglynn_ | nsaje: "Multi meter arithmetic transformer" ... catchy title, /me likee :) | 10:08 |
| *** idegtiarov has quit IRC | 10:09 | |
| *** yatin_ has quit IRC | 10:13 | |
| *** ildikov has joined #openstack-ceilometer | 10:15 | |
| *** aignatov has joined #openstack-ceilometer | 10:19 | |
| nsaje | eglynn_: well, looked like the only one that described the functionality to me :) | 10:21 |
| *** flwang_ has quit IRC | 10:22 | |
| eglynn_ | nsaje: yep I'm copying'n'pasting that title directly into my Juno slide-deck :) | 10:22 |
| cdent | needs another T to make a nicely balance acronym | 10:23 |
| cdent | multi meter arithmetic transformer technology | 10:23 |
| cdent | your friend MMATT | 10:23 |
| nsaje | cdent: nice! | 10:23 |
| *** flwang_ has joined #openstack-ceilometer | 10:27 | |
| *** underyx|off is now known as underyx | 10:30 | |
| openstackgerrit | Zhai, Edwin proposed a change to openstack/ceilometer-specs: Add spec for IPMI support https://review.openstack.org/104460 | 10:31 |
| *** alexpilotti has joined #openstack-ceilometer | 10:35 | |
| DinaBelova | eglynn_, /me reading | 10:40 |
| DinaBelova | eglynn_, thanks for the meeting attending :D for me it's too late :D | 10:41 |
| eglynn_ | DinaBelova: yeah that PTLs meeting is rediculously late for MSK, I'm sure SergeyL loves it ... NOT! | 10:42 |
| DinaBelova | eglynn_, hehe, exactly :D | 10:42 |
| *** Alice911 has quit IRC | 10:42 | |
| DinaBelova | eglynn_, sometimes I try to keep my eye on-line for them, but it's still too late to be always available there :D | 10:43 |
| eglynn_ | ... as the centre of gravity of the community moves away from the west coast US, we really should push for saner scheduling of those upstream meetings | 10:43 |
| DinaBelova | eglynn_, +1 | 10:43 |
| SergeyLukjanov | eglynn_, yeah, it's a very nice meeting time | 10:56 |
| *** yatin_ has joined #openstack-ceilometer | 10:59 | |
| * cdent must go socialize with a nephew briefly | 11:03 | |
| cdent | biab | 11:03 |
| *** cdent has quit IRC | 11:03 | |
| *** yatin_ has left #openstack-ceilometer | 11:09 | |
| *** alexpilotti has quit IRC | 11:26 | |
| *** underyx is now known as underyx|off | 11:29 | |
| ekarlso | many moving from west coast ? :p | 11:29 |
| *** underyx|off is now known as underyx | 11:30 | |
| *** alexpilotti has joined #openstack-ceilometer | 11:30 | |
| eglynn_ | ekarlso: not need for *individuals* to move | 11:31 |
| eglynn_ | ekarlso: ... as we explicitly do not follow a BDFL model | 11:32 |
| *** alexpilotti has quit IRC | 11:35 | |
| *** Longgeek has joined #openstack-ceilometer | 11:38 | |
| *** cdent has joined #openstack-ceilometer | 11:41 | |
| *** promulo has quit IRC | 11:46 | |
| *** giroro_ has quit IRC | 12:01 | |
| *** flwang_ has quit IRC | 12:02 | |
| *** Ruetobas has joined #openstack-ceilometer | 12:07 | |
| *** jdob has joined #openstack-ceilometer | 12:09 | |
| *** Ruetobas has quit IRC | 12:11 | |
| *** idegtiarov has joined #openstack-ceilometer | 12:14 | |
| *** cdent has quit IRC | 12:17 | |
| *** Ruetobas has joined #openstack-ceilometer | 12:17 | |
| *** cdent has joined #openstack-ceilometer | 12:31 | |
| *** cdent has quit IRC | 12:36 | |
| *** jdob has quit IRC | 12:37 | |
| *** jdob has joined #openstack-ceilometer | 12:38 | |
| *** prad has quit IRC | 12:40 | |
| *** cdent has joined #openstack-ceilometer | 12:42 | |
| openstackgerrit | ling-yun proposed a change to openstack/ceilometer-specs: Add log translation hints for ceilometer https://review.openstack.org/105727 | 12:43 |
| *** _nadya_ has joined #openstack-ceilometer | 12:45 | |
| openstackgerrit | Nejc Saje proposed a change to openstack/python-ceilometerclient: Check if the alarm has time constraints field before displaying https://review.openstack.org/95713 | 12:48 |
| *** julim has joined #openstack-ceilometer | 12:48 | |
| openstackgerrit | ling-yun proposed a change to openstack/ceilometer-specs: Add log translation hints for ceilometer https://review.openstack.org/105727 | 12:51 |
| *** eglynn_ has quit IRC | 12:51 | |
| *** ajc_ has quit IRC | 12:52 | |
| *** rbowen has joined #openstack-ceilometer | 12:53 | |
| openstackgerrit | ling-yun proposed a change to openstack/ceilometer-specs: Add log translation hints for ceilometer https://review.openstack.org/105727 | 12:53 |
| *** _nadya_ has quit IRC | 12:54 | |
| *** _nadya_ has joined #openstack-ceilometer | 12:55 | |
| openstackgerrit | ling-yun proposed a change to openstack/ceilometer-specs: Add log translation hints for ceilometer https://review.openstack.org/105727 | 12:58 |
| *** _nadya_ has quit IRC | 12:59 | |
| *** thomasem has joined #openstack-ceilometer | 13:00 | |
| *** d0ugal has joined #openstack-ceilometer | 13:04 | |
| *** joesavak has joined #openstack-ceilometer | 13:08 | |
| *** promulo has joined #openstack-ceilometer | 13:08 | |
| *** changbl has joined #openstack-ceilometer | 13:10 | |
| *** zul has quit IRC | 13:11 | |
| openstackgerrit | Nejc Saje proposed a change to openstack/ceilometer-specs: Arithmetic transformer spec https://review.openstack.org/105467 | 13:11 |
| openstackgerrit | ling-yun proposed a change to openstack/ceilometer-specs: Add log translation hints for ceilometer https://review.openstack.org/105727 | 13:11 |
| *** ildikov has quit IRC | 13:14 | |
| *** nsaje is now known as nsaje-commute | 13:17 | |
| *** zul has joined #openstack-ceilometer | 13:18 | |
| *** eglynn_ has joined #openstack-ceilometer | 13:21 | |
| *** ityaptin has joined #openstack-ceilometer | 13:23 | |
| *** Longgeek has quit IRC | 13:25 | |
| *** _nadya_ has joined #openstack-ceilometer | 13:25 | |
| openstackgerrit | ling-yun proposed a change to openstack/ceilometer-specs: Add log translation hints for ceilometer https://review.openstack.org/105727 | 13:33 |
| *** Longgeek has joined #openstack-ceilometer | 13:35 | |
| *** nealph__ has joined #openstack-ceilometer | 13:40 | |
| *** eglynn-regus has joined #openstack-ceilometer | 13:41 | |
| *** inc0 has quit IRC | 13:41 | |
| *** shadower_ has joined #openstack-ceilometer | 13:43 | |
| *** ondergetekende_ has joined #openstack-ceilometer | 13:45 | |
| *** mitz- has joined #openstack-ceilometer | 13:47 | |
| *** tasdomas` has joined #openstack-ceilometer | 13:47 | |
| *** nealph_ has quit IRC | 13:48 | |
| *** gibi has quit IRC | 13:48 | |
| *** mitz has quit IRC | 13:48 | |
| *** nijaba has quit IRC | 13:48 | |
| *** greghaynes has quit IRC | 13:48 | |
| *** chrisf has quit IRC | 13:48 | |
| *** shadower has quit IRC | 13:48 | |
| *** tasdomas has quit IRC | 13:48 | |
| *** dragondm has quit IRC | 13:48 | |
| *** eglynn-offic-afk has quit IRC | 13:48 | |
| *** ondergetekende has quit IRC | 13:48 | |
| *** fc__ has quit IRC | 13:48 | |
| *** nijaba has joined #openstack-ceilometer | 13:56 | |
| *** nijaba has quit IRC | 13:56 | |
| *** nijaba has joined #openstack-ceilometer | 13:56 | |
| *** gibi has joined #openstack-ceilometer | 13:58 | |
| *** prad has joined #openstack-ceilometer | 13:59 | |
| *** dragondm has joined #openstack-ceilometer | 13:59 | |
| *** chrisf has joined #openstack-ceilometer | 14:01 | |
| *** flwang_ has joined #openstack-ceilometer | 14:03 | |
| *** flwang_ has quit IRC | 14:07 | |
| openstackgerrit | Chris Dent proposed a change to openstack/ceilometer: Implement consuming ipmi notifications from Ironic https://review.openstack.org/105486 | 14:10 |
| *** greghaynes has joined #openstack-ceilometer | 14:11 | |
| *** alexpilotti has joined #openstack-ceilometer | 14:14 | |
| rbowen | A reminder - eglynn will be doing a Google Hangout in about 40 minutes on what's coming for Ceilometer for Juno :: https://plus.google.com/events/c6e8vjjn8klrf78ruhkr95j4tas | 14:19 |
| *** idegtiarov has quit IRC | 14:22 | |
| *** idegtiarov has joined #openstack-ceilometer | 14:24 | |
| *** jaypipes has joined #openstack-ceilometer | 14:27 | |
| *** fnaval has joined #openstack-ceilometer | 14:42 | |
| *** r3pl4y has quit IRC | 14:44 | |
| *** fc__ has joined #openstack-ceilometer | 14:44 | |
| *** nacim has quit IRC | 14:46 | |
| *** cdent has quit IRC | 14:56 | |
| *** cdent has joined #openstack-ceilometer | 14:58 | |
| *** r3pl4y has joined #openstack-ceilometer | 15:01 | |
| *** idegtiarov has quit IRC | 15:02 | |
| *** _nadya_ has quit IRC | 15:10 | |
| *** Hao has joined #openstack-ceilometer | 15:11 | |
| *** jaypipes is now known as jay-house-hunter | 15:22 | |
| *** cdent_ has joined #openstack-ceilometer | 15:22 | |
| *** underyx is now known as underyx|off | 15:23 | |
| *** cdent_ has quit IRC | 15:23 | |
| *** cdent has quit IRC | 15:25 | |
| *** cdent has joined #openstack-ceilometer | 15:25 | |
| *** shadower_ is now known as shadower | 15:26 | |
| *** joesavak has quit IRC | 15:30 | |
| *** Longgeek has quit IRC | 15:36 | |
| *** Longgeek has joined #openstack-ceilometer | 15:37 | |
| *** Longgeek has quit IRC | 15:40 | |
| *** Longgeek has joined #openstack-ceilometer | 15:40 | |
| *** yjiang5_away is now known as yjiang5 | 15:45 | |
| *** alexpilotti has quit IRC | 15:45 | |
| *** _cjones_ has joined #openstack-ceilometer | 15:48 | |
| *** _nadya_ has joined #openstack-ceilometer | 15:50 | |
| *** safchain has quit IRC | 16:00 | |
| *** Ruetobas has quit IRC | 16:01 | |
| *** Ruetobas has joined #openstack-ceilometer | 16:03 | |
| *** flwang_ has joined #openstack-ceilometer | 16:04 | |
| *** jay-house-hunter has quit IRC | 16:05 | |
| rbowen | For those that missed eglynn_'s Ceilometer hangout, that video is at https://plus.google.com/events/c6e8vjjn8klrf78ruhkr95j4tas | 16:06 |
| *** Ruetobas has quit IRC | 16:08 | |
| *** flwang_ has quit IRC | 16:08 | |
| *** vrovachev has quit IRC | 16:11 | |
| cdent | nicely done eglynn | 16:12 |
| *** Hao has quit IRC | 16:14 | |
| *** nati_ueno has joined #openstack-ceilometer | 16:19 | |
| *** Ruetobas has joined #openstack-ceilometer | 16:21 | |
| *** Ruetobas has quit IRC | 16:25 | |
| *** _nadya_ has quit IRC | 16:29 | |
| *** _nadya_ has joined #openstack-ceilometer | 16:30 | |
| *** Ruetobas has joined #openstack-ceilometer | 16:31 | |
| *** _nadya_ has quit IRC | 16:34 | |
| *** Hao has joined #openstack-ceilometer | 16:42 | |
| *** Hao has quit IRC | 17:59 | |
| *** ildikov has joined #openstack-ceilometer | 17:59 | |
| *** Longgeek has quit IRC | 18:01 | |
| *** eglynn_ has quit IRC | 18:03 | |
| *** flwang_ has joined #openstack-ceilometer | 18:05 | |
| *** flwang_ has quit IRC | 18:10 | |
| *** fnaval has quit IRC | 18:25 | |
| *** eglynn_ has joined #openstack-ceilometer | 18:35 | |
| *** prad has quit IRC | 18:53 | |
| *** prad has joined #openstack-ceilometer | 18:54 | |
| *** Hao has joined #openstack-ceilometer | 19:02 | |
| *** Hao has quit IRC | 19:11 | |
| *** Hao has joined #openstack-ceilometer | 19:23 | |
| *** promulo has quit IRC | 19:29 | |
| *** Hao has quit IRC | 19:30 | |
| *** r3pl4y has quit IRC | 19:33 | |
| *** fnaval has joined #openstack-ceilometer | 19:50 | |
| *** prad has quit IRC | 19:59 | |
| *** flwang_ has joined #openstack-ceilometer | 20:06 | |
| *** flwang_ has quit IRC | 20:10 | |
| *** jergerber has joined #openstack-ceilometer | 20:14 | |
| *** prad has joined #openstack-ceilometer | 20:14 | |
| *** jogo has joined #openstack-ceilometer | 20:16 | |
| jogo | eglynn_: ping | 20:16 |
| eglynn_ | jogo: hola | 20:16 |
| jogo | so the branchless tempest stuff | 20:17 |
| eglynn_ | yep | 20:17 |
| *** nati_uen_ has joined #openstack-ceilometer | 20:18 | |
| jogo | I wanted to get some clarity on your response to Matt's email | 20:18 |
| eglynn_ | yeah, I just going ask whether my response to Matt made sense | 20:18 |
| jogo | when saying cross-serice communication should have an API contract that doesn't mean it has to be REST | 20:18 |
| jogo | notifcations can be a contractual API | 20:19 |
| eglynn_ | sure, though sadly they haven't been treated like that in openstack (traditionally) | 20:19 |
| jogo | that is because we have no testing etc. around them | 20:20 |
| jogo | ceilometer silently started consuming them without setting up ant contractual API that we enfoce | 20:20 |
| *** nati_ueno has quit IRC | 20:20 | |
| jogo | so from reading that thread I see several seperate questions: | 20:20 |
| eglynn_ | ceilometer silently consuming notifications is kinda orthogonal, no? | 20:21 |
| eglynn_ | I mean if not ceilo, something else would be consuming those notifications? | 20:22 |
| jogo | 1) who is the target audience for ceilometer | 20:22 |
| jogo | 2) how do I discover what my ceilometer supports | 20:22 |
| jogo | 3) how can we safely evolve / not break notifications | 20:22 |
| jogo | so notifications sort of grew organically and never had a tested contractual public API around them. | 20:23 |
| eglynn_ | the target audience ranges from cloud operators to admins to normal users of features such as heat autoscaling | 20:23 |
| jogo | great, so that is the target audience that I expected | 20:23 |
| jogo | so that brings us to item 2. how do i dscover what my ceilometer supports | 20:24 |
| eglynn_ | discovery via a capabilities API is possible for QoS provided by the storage driver | 20:24 |
| *** ccrouch has quit IRC | 20:24 | |
| eglynn_ | (e.g. the type of aggregate functions, querying, pagination etc.) | 20:24 |
| jogo | ?, how do I know that ceilometer will process a specific notification? | 20:25 |
| eglynn_ | we could potentially extend that capabilities API model to include a declaration of the types of meters ceilo generates | 20:25 |
| jogo | ++ | 20:25 |
| eglynn_ | jogo: as I meant to say in that email, there is no way of discovering that *currently* for notifications | 20:26 |
| eglynn_ | (other than inferring from docco) | 20:26 |
| jogo | right | 20:26 |
| jogo | well lets first look at where we want to be | 20:26 |
| eglynn_ | is there a way of discovering which notifications nova or cinder actually produce? | 20:26 |
| cdent | I think it would be a huge bummer if ceilometer were set up so that the default orientation is that it must know about the notifications it is supposed to consume. I think that is something the other services (and users) should get to decide _not_ ceilometer. | 20:27 |
| *** Hao has joined #openstack-ceilometer | 20:27 | |
| cdent | (by "were set up" I mean "continued to be") | 20:28 |
| eglynn_ | AFAICS the notification mechanism has no discoverability on either the producer or consumer side, and no versioning | 20:28 |
| *** flwang_ has joined #openstack-ceilometer | 20:28 | |
| jogo | using Sean's taxonomy, I think we should move the specific failure you hit (ceilometer not processing a cinder notifcatoin) closer to category 2. something that is discoverable | 20:28 |
| jogo | eglynn_: right, so that is the third question I have | 20:29 |
| jogo | we need full versioning/discoverability/contractual APIs/ around all cross service communication. we have this for REST APIs (mostly) but not for notifications | 20:30 |
| jogo | cdent: ? can you elaborate | 20:30 |
| eglynn_ | jogo: do I detect that you're coming from the PoV that the shortcomings in the notification system are due to ceilometer's usage of that mechanism? | 20:30 |
| eglynn_ | "without setting up ant contractual API ..." etc. | 20:31 |
| *** flwang_ has quit IRC | 20:31 | |
| jogo | eglynn_: correct. | 20:31 |
| cdent | jogo: to me ceilometer is a service for gathering metrics and then being able to do things with those metrics. anything with the proper creds ought to be able to send out a notification which is to be treated as a metric and ceilometer should consume it | 20:31 |
| *** Hao has quit IRC | 20:32 | |
| jogo | eglynn_: but I think that is somewhat decoupled from item 2) no discoverabilty of what ceilometer can process | 20:32 |
| jogo | cdent: so your saying ceilometer should be generic and not need to know bout the contents of notifications? | 20:32 |
| eglynn_ | jogo: that seems like an odd viewpoint ... "here's a mechanism for being notified about stuff, but you should fix it before you use it" ;) | 20:33 |
| eglynn_ | we're interleaving two separate conversations here, can we sequence the discussion? | 20:33 |
| jogo | eglynn_: yeah we are. lets go back to item #2 | 20:33 |
| jogo | we agree on item #1 | 20:33 |
| cdent | Sorry, I brought this up because I think it pertains to the idea that a contract of some form ought to be required. | 20:33 |
| eglynn_ | #topic "how do I discover what my ceilometer supports" | 20:34 |
| jogo | eglynn_: yup. so as a user of the cloud I want an answer that is better then read the docs | 20:34 |
| cdent | me too | 20:34 |
| eglynn_ | the capabilities API provides some scope possibly for doing this | 20:34 |
| eglynn_ | http://docs.openstack.org/developer/ceilometer/webapi/v2.html#capabilities | 20:35 |
| jogo | on a related note, this is a much bigger issue accross all of OpenStack we do not have uniform ways of doing discovery | 20:35 |
| eglynn_ | ^^^ mostly focused on API features that are predicated on storage-driver-level capabilities | 20:35 |
| eglynn_ | so say we extend that, to advertize the meters that we're capable of producing? | 20:36 |
| jogo | eglynn_: that should address the specific issue you hit right? | 20:36 |
| eglynn_ | without necessarily saying *how* we produce those meters | 20:36 |
| eglynn_ | i.e. without stating in capabilities ... "we consume volume.snapshot.create notifications" or whatever | 20:37 |
| cdent | when I first ran `ceilometer meter-list` the output I was expecting was "the meters we're capable of producing" | 20:37 |
| eglynn_ | cdent: nope it just says ... "what meters I currently have data for" | 20:38 |
| cdent | yeah, I know that _now_ | 20:38 |
| *** prad has quit IRC | 20:38 | |
| cdent | but I think it supports the value of a being able to get such a list | 20:38 |
| jogo | cdent eglynn_: sounds like both lists are needed | 20:38 |
| eglynn_ | jogo: so such a mechanism would allow that test to be skipped against stable/icehouse ... | 20:39 |
| eglynn_ | jogo: ... *if and only if* the extended capabilities mechanism was backported to icehouse | 20:39 |
| jogo | eglynn_: I am less concerend with fixing this *specific* bug and more interested in solving it for the future | 20:40 |
| jogo | but yes. | 20:40 |
| eglynn_ | well until we transition to Kepler, the future cases would also require backporting the mechanism to Icehouse, I think | 20:41 |
| * eglynn_ is not letting "Kepler" go, reardless of the trademark search ;) | 20:42 | |
| cdent | :) | 20:42 |
| cdent | instead of backporting can't you instead declare some kind of requirement for _new_ tests? | 20:42 |
| eglynn_ | cdent: but some of the new tests would be runnable against icehouse, no? | 20:43 |
| *** ccrouch has joined #openstack-ceilometer | 20:43 | |
| eglynn_ | cdent: i.e. new tests for old features, as opposed to new tests for new features | 20:43 |
| cdent | s/requirement/annotation/ ? | 20:44 |
| jogo | so I think the short term issue for icehouse has several possible solutions, that the tempest folks can better answer | 20:44 |
| *** prad has joined #openstack-ceilometer | 20:45 | |
| eglynn_ | jogo: so continue that aspect of the discussion (short term solutions for icehouse) on the ML thread to get further input from the Tempest folks? | 20:46 |
| eglynn_ | ... but move on item #3 here on IRC? | 20:47 |
| jogo | eglynn_: sounds good. yes so #3 | 20:47 |
| eglynn_ | k | 20:47 |
| jogo | have notification formats changed in the past breaking ceilometer? | 20:48 |
| eglynn_ | well, I have anecdata to that effect, but can't recall chapter & verse of specific instances | 20:49 |
| eglynn_ | ... digging thru' the fossil record would find specific instances | 20:49 |
| jogo | so it sounds like its a real issue | 20:49 |
| jogo | this goes back to what you said earler about notifications intially were this strange thing | 20:50 |
| jogo | that evolved without a contractual (and tested) stable API | 20:50 |
| jogo | they are very haphazard | 20:51 |
| jogo | and with ceilometer (or really anything) consuming them that is not acceptable | 20:51 |
| jogo | if each project had some tests making sure the correct notifications are emitted at the right time, that would be a good first whack at this | 20:52 |
| jogo | but presumably we need full versioning and discoverablity at some point as well | 20:53 |
| jogo | cdent: you had some thing to add about this? | 20:53 |
| cdent | when you say versioning do you mean of a specific notification from a specific service? | 20:53 |
| eglynn_ | ok, so a basic problem here is that notifications pre-existed ceilo, are outside of the control of ceilo, and are also being consumed by things other than ceilo (e.g. stacktach) | 20:53 |
| eglynn_ | so we can't simply wave a magic wand and make notifications discoverable and versioned | 20:53 |
| jogo | agree on all points | 20:54 |
| eglynn_ | cdent: "versioning" == in a.b you can expect this payload, whereas in version c.d you can expect this other payload | 20:54 |
| jogo | so lets at least make sure projcts don't break there notifcations | 20:54 |
| jogo | without knowing they are doing so | 20:54 |
| cdent | eglynn_: yes, but is that per notification or of any notification? | 20:54 |
| cdent | I think, if we try hard, we can imagine a future where people have migrated to a notification message format protocol. | 20:55 |
| eglynn_ | cdent: per-notification, we do somethign similar already for RPC within the nova services for example | 20:55 |
| jogo | eglynn_: the issue here IHMO is ceilometer made some assumptions about notifcations that were never really true. without trying to fix notifcations themselves | 20:55 |
| cdent | and ceilometer and anything else can talk that protocol... | 20:55 |
| jogo | so I think item #3 is about how do we make notifcations stable contractual APIs in all of OpenStack | 20:56 |
| jogo | so we don't break ceilometer | 20:56 |
| eglynn_ | jogo: the assumption in question being "here's a thing being emitted for other things to listen on, let's go use it"? | 20:56 |
| jogo | yes, I think that assumption is only partially correct. yes things were being emitted but they weren't a stable API they were quitly added without signficant testing for rax usage etc | 20:57 |
| eglynn_ | TBH ceilo initially had minimal leverage with the other projects to force a fixing of the notification mechanism before actually using it | 20:58 |
| jogo | eglynn_: I don't fully agree with that statement, but its a moot point now. We need to lock down notifcations | 20:58 |
| jogo | because if we emit them and say folks can use them (which we sort of do) they better be a stable contractual API) | 20:59 |
| cdent | s/lock down/standardize/ ? | 20:59 |
| eglynn_ | jogo: I'd love to see notifications fixed | 20:59 |
| cdent | there's a big different between those two | 20:59 |
| eglynn_ | jogo: so here's the least that needs to happen | 20:59 |
| jogo | cdent: lock down: don't change quietly without backwards compat etc | 20:59 |
| cdent | standardize: give people some rules they know they can follow, and if they follow them, they get to participate, if they don't, they don't | 21:00 |
| eglynn_ | 1. buy-in from the emitters of the notification stream (nova, glance, cinder, neutron etc.) | 21:00 |
| *** jdob has quit IRC | 21:00 | |
| eglynn_ | 2. acquisence from the "external" consumers such as stacktach | 21:01 |
| eglynn_ | *acquiescence | 21:01 |
| eglynn_ | i.e. we shouldn't, as good citizens, pull the rug out from under them | 21:01 |
| jogo | whats #3? | 21:02 |
| eglynn_ | 3. is implicit ... "internal" consumers such as ceilo also buy-in to the program | 21:02 |
| jogo | cdent: so standardizing would be great but, I think thats a later step. once we have a way to evolve notifcations without breaking things | 21:02 |
| jogo | eglynn_: hmm I am not sure if we are currently talking about the same thing | 21:03 |
| jogo | I would like to see several things happen to make notifcations a real API | 21:03 |
| eglynn_ | jogo: what are you talking about? | 21:03 |
| cdent | well, we have notifications on the bus now, you can "shim" hearing those notifications and translate and republish them | 21:03 |
| jogo | steps to make notifications a stable API | 21:03 |
| jogo | brb, just got a package | 21:04 |
| eglynn_ | jogo: sure, so am I (I think) | 21:04 |
| jogo | so I think the first step is to say that the current notifcations should not change unintentionally. | 21:06 |
| eglynn_ | jogo: so I was assuming that the fall-out from making notifications a stable, versioned, contract-bound API would impact on *both* producers and consumers? | 21:06 |
| jogo | this means each project should have a suite of tests to validate notifications | 21:06 |
| jogo | eglynn_: in the long run yes | 21:06 |
| *** prad has quit IRC | 21:06 | |
| jogo | but I would first like to prevent nova from accidentally breaking a notifcation | 21:06 |
| eglynn_ | OK so nova adds a bunch of tests to assert over notification payload | 21:07 |
| eglynn_ | then nova wants to change that payload | 21:07 |
| eglynn_ | which requires a version bump | 21:07 |
| eglynn_ | which surely requires that the consumers are version-aware, no? | 21:08 |
| jogo | or drop/add a notification | 21:08 |
| jogo | so once we prevent projects from accidentally changing notifcations we can sort out howto properly version them / make 'em discoverable etc. | 21:08 |
| jogo | so I think we can fix this in two steps | 21:09 |
| jogo | add tests to confirm what we have now. fix the future | 21:09 |
| jogo | eglynn_ cdent: thoughts? | 21:10 |
| *** prad has joined #openstack-ceilometer | 21:10 | |
| * cdent thinks | 21:10 | |
| eglynn_ | well, ok, but that does sound a little like kicking the hard, frictionful problem down the road by concentrating on the frictionless problem first | 21:10 |
| cdent | yes, that's exactly what I was going to say | 21:10 |
| jogo | they can be done in parallel | 21:10 |
| eglynn_ | OK, if there's buy-in from nova/glance/neutron to go add those tests, they would certainly have value | 21:11 |
| jogo | so speaking for nova I would push for buy in | 21:11 |
| *** julim has quit IRC | 21:12 | |
| eglynn_ | another way of approaching this would be to do what russellbryant did for RPC versioning in nova, IIRC | 21:13 |
| jogo | we would need those tests even if we had perfect versioned/discoverable notifications etc | 21:13 |
| jogo | eglynn_: hrm, that may work for the versioning etc things. I like it | 21:14 |
| eglynn_ | ... i.e. ACK that the past was broken, get buy-in for the future, take the pain over a release cycle | 21:14 |
| jogo | yeah | 21:14 |
| jogo | the notication problem is similar to nova's RPC issue | 21:15 |
| *** thomasem has quit IRC | 21:15 | |
| jogo | so I think thats a good model to use | 21:15 |
| jogo | or at least start with | 21:16 |
| eglynn_ | so we'd be talking about session(s) on the cross-project track in Paris in November? | 21:16 |
| * eglynn_ is thinking time-horizon for this ... | 21:16 | |
| jogo | yeah, | 21:16 |
| jogo | I do think we can get test in for the current RPC notifcations this cylce though | 21:17 |
| *** chrisf has quit IRC | 21:17 | |
| eglynn_ | so that would be in-tree test in each of he services, as opposed to tempest-level tests? | 21:17 |
| jogo | I was thinking in tree | 21:17 |
| jogo | for example: nova would make sure that x notifcatoins with y contents are produced for a nova boot | 21:17 |
| jogo | or something like that | 21:18 |
| eglynn_ | sounds good, but the proof would be the resourcing of this task in each project | 21:18 |
| eglynn_ | jogo reckons nova would be on-board, but we'd still need glance/cinder/neutron to step up to the plate | 21:19 |
| jogo | yeah, and then we would need folks to do all the work | 21:19 |
| jogo | eglynn_: sounds like we are in agreement on most things, want to follow up to the ML thread with some of this? | 21:19 |
| jogo | to take this back to the larger table | 21:19 |
| jogo | nova has similar tests right now for the API | 21:20 |
| jogo | we have basic API test that prevent us from changing the API without knowing about it | 21:20 |
| eglynn_ | yeah I can follow up on the ML with a summary of this discussion, but would be tmrw AM by the time I get to it | 21:21 |
| * eglynn_ has the end of a world cup match to watch now :) | 21:21 | |
| * cdent squints | 21:23 | |
| *** chrisf has joined #openstack-ceilometer | 21:25 | |
| jogo | works for me | 21:25 |
| eglynn_ | jogo: cool, thanks for the discussion & input | 21:25 |
| jogo | eglynn_: thanks, this has been very productive | 21:25 |
| cdent | Thanks for putting up with my blue-skying... | 21:25 |
| jogo | ^_^ | 21:26 |
| * cdent is still bewildered that notifications are RPC | 21:26 | |
| jogo | yeah RPC isn't really the correct term | 21:28 |
| *** Hao has joined #openstack-ceilometer | 21:29 | |
| cdent | but it is modelled that way: agent A calls a method on agent B via the bus | 21:29 |
| cdent | _sort of_ | 21:29 |
| cdent | it all seems a bit hinky | 21:29 |
| *** Hao has quit IRC | 21:34 | |
| *** fnaval has quit IRC | 21:35 | |
| *** fnaval has joined #openstack-ceilometer | 21:42 | |
| *** thomasem has joined #openstack-ceilometer | 21:46 | |
| *** thomasem has quit IRC | 21:47 | |
| *** rbowen has quit IRC | 21:48 | |
| *** thomasem has joined #openstack-ceilometer | 21:52 | |
| *** thomasem has quit IRC | 21:54 | |
| *** thomasem has joined #openstack-ceilometer | 21:55 | |
| *** jay-house-hunter has joined #openstack-ceilometer | 22:03 | |
| *** fnaval has quit IRC | 22:11 | |
| *** nacim has joined #openstack-ceilometer | 22:15 | |
| openstackgerrit | Fei Long Wang proposed a change to openstack/ceilometer: Rename bandwidth to network.traffic https://review.openstack.org/105896 | 22:16 |
| *** cdent_ has joined #openstack-ceilometer | 22:20 | |
| *** cdent has quit IRC | 22:23 | |
| *** cdent_ is now known as cdent | 22:23 | |
| *** prad has quit IRC | 22:26 | |
| *** thomasem has quit IRC | 22:27 | |
| *** nacim has quit IRC | 22:28 | |
| *** Hao has joined #openstack-ceilometer | 22:30 | |
| *** flwang_ has joined #openstack-ceilometer | 22:32 | |
| *** Hao has quit IRC | 22:36 | |
| *** flwang_ has quit IRC | 22:36 | |
| *** boris-42 has quit IRC | 22:37 | |
| *** jay-house-hunter has quit IRC | 22:38 | |
| *** amalagon has quit IRC | 22:38 | |
| *** boris-42 has joined #openstack-ceilometer | 22:39 | |
| *** ryanpetrello has joined #openstack-ceilometer | 22:43 | |
| ryanpetrello | anybody aware of any issues that would cause ceilometer tests to hang in the gate queue? | 22:44 |
| ryanpetrello | https://jenkins01.openstack.org/job/gate-pecan-tox-ceilometer-tip/47/console | 22:44 |
| ryanpetrello | this is the second time I’ve watched this test just sit and timeout and hang :\ | 22:44 |
| *** amalagon has joined #openstack-ceilometer | 23:00 | |
| *** eglynn_ has quit IRC | 23:03 | |
| *** jergerber has quit IRC | 23:08 | |
| *** Hao has joined #openstack-ceilometer | 23:32 | |
| *** Hao has quit IRC | 23:37 | |
| *** packet has joined #openstack-ceilometer | 23:40 | |
| *** ccrouch has quit IRC | 23:48 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!