| *** alexpilotti has quit IRC | 00:00 | |
| *** promulo_ has quit IRC | 00:03 | |
| *** ankita_wagh has quit IRC | 00:08 | |
| *** ddieterly has joined #openstack-ceilometer | 00:17 | |
| *** ddieterly has quit IRC | 00:22 | |
| *** ddieterly has joined #openstack-ceilometer | 00:24 | |
| *** lexholden has quit IRC | 00:26 | |
| *** r-daneel has joined #openstack-ceilometer | 00:38 | |
| *** r-daneel has quit IRC | 00:44 | |
| *** Kennan has quit IRC | 01:13 | |
| *** htruta has quit IRC | 01:14 | |
| *** htruta has joined #openstack-ceilometer | 01:15 | |
| *** Kennan has joined #openstack-ceilometer | 01:16 | |
| *** krotsche_vaca is now known as krotscheck | 01:28 | |
| *** htruta_ has quit IRC | 01:34 | |
| *** pauldix has quit IRC | 01:53 | |
| openstackgerrit | Ryota MIBU proposed openstack/aodh: Refactor alarm scenario tests (RuleThreshold) https://review.openstack.org/204439 | 02:07 |
|---|---|---|
| openstackgerrit | Ryota MIBU proposed openstack/aodh: Refactor alarm scenario tests (RuleThreshold) https://review.openstack.org/204439 | 02:11 |
| *** ddieterly has quit IRC | 02:17 | |
| *** liusheng has joined #openstack-ceilometer | 02:25 | |
| openstackgerrit | liusheng proposed openstack/aodh: Use user_id/project_id from service_credentials in alarm_change https://review.openstack.org/205782 | 02:27 |
| openstackgerrit | liusheng proposed openstack/aodh: Avoid getting alarm change notifier repeatedly https://review.openstack.org/205875 | 02:27 |
| *** ddieterly has joined #openstack-ceilometer | 02:56 | |
| *** ddieterly has quit IRC | 03:01 | |
| *** ddieterly has joined #openstack-ceilometer | 03:10 | |
| *** changbl has quit IRC | 03:19 | |
| openstackgerrit | Jie Li proposed openstack/ceilometer: Replace isotime() with utcnow() and isoformat() https://review.openstack.org/205884 | 03:24 |
| *** changbl has joined #openstack-ceilometer | 03:32 | |
| *** pauldix has joined #openstack-ceilometer | 03:41 | |
| *** pauldix has quit IRC | 03:46 | |
| openstackgerrit | liusheng proposed openstack/aodh: void getting alarm change notifier repeatedly https://review.openstack.org/205875 | 03:54 |
| *** liusheng has quit IRC | 04:00 | |
| openstackgerrit | liusheng proposed openstack/aodh: Avoid getting alarm change notifier repeatedly https://review.openstack.org/205875 | 04:03 |
| *** liusheng has joined #openstack-ceilometer | 04:03 | |
| *** deepthi has joined #openstack-ceilometer | 04:13 | |
| *** fawadkhaliq has joined #openstack-ceilometer | 04:17 | |
| *** Kennan2 has joined #openstack-ceilometer | 04:20 | |
| *** Kennan has quit IRC | 04:20 | |
| *** ankita_wagh has joined #openstack-ceilometer | 04:25 | |
| *** yprokule has joined #openstack-ceilometer | 04:28 | |
| openstackgerrit | liusheng proposed openstack/aodh: Use user_id/project_id from service_credentials in alarm_change https://review.openstack.org/205782 | 05:01 |
| *** fawadkhaliq has quit IRC | 05:13 | |
| *** fawadkhaliq has joined #openstack-ceilometer | 05:13 | |
| *** _nadya_ has joined #openstack-ceilometer | 05:29 | |
| *** _nadya_ has quit IRC | 05:33 | |
| *** _nadya_ has joined #openstack-ceilometer | 05:35 | |
| *** _nadya_ has quit IRC | 05:36 | |
| *** _nadya_ has joined #openstack-ceilometer | 05:37 | |
| *** _nadya_ has quit IRC | 05:41 | |
| *** _nadya_ has joined #openstack-ceilometer | 05:42 | |
| *** jwcroppe has joined #openstack-ceilometer | 06:00 | |
| *** _nadya_ has quit IRC | 06:01 | |
| *** _nadya_ has joined #openstack-ceilometer | 06:01 | |
| *** pauldix has joined #openstack-ceilometer | 06:02 | |
| *** _nadya_ has quit IRC | 06:06 | |
| *** fawadk has joined #openstack-ceilometer | 06:07 | |
| *** fawadkhaliq has quit IRC | 06:08 | |
| *** lsmola has joined #openstack-ceilometer | 06:12 | |
| openstackgerrit | Jie Li proposed openstack/ceilometer: Replace isotime() with utcnow() and isoformat() https://review.openstack.org/205884 | 06:18 |
| *** alexpilotti has joined #openstack-ceilometer | 06:26 | |
| *** lan has quit IRC | 06:27 | |
| *** fawadk has quit IRC | 06:30 | |
| *** fawadkhaliq has joined #openstack-ceilometer | 06:30 | |
| *** r-mibu has joined #openstack-ceilometer | 06:39 | |
| openstackgerrit | liusheng proposed openstack/aodh: Use user_id/project_id from service_credentials in alarm_change https://review.openstack.org/205782 | 06:39 |
| *** ddieterly has quit IRC | 06:51 | |
| *** fawadk has joined #openstack-ceilometer | 07:13 | |
| *** fawadkhaliq has quit IRC | 07:13 | |
| *** lsmola has quit IRC | 07:21 | |
| *** lsmola has joined #openstack-ceilometer | 07:21 | |
| *** jwcroppe has quit IRC | 07:30 | |
| *** jwcroppe has joined #openstack-ceilometer | 07:31 | |
| *** safchain has joined #openstack-ceilometer | 07:34 | |
| openstackgerrit | Kamil Rykowski proposed openstack/ceilometer: Versioned object for Event model https://review.openstack.org/181338 | 07:36 |
| *** ankita_wagh has quit IRC | 07:41 | |
| *** cmyster has joined #openstack-ceilometer | 07:42 | |
| *** fawadkhaliq has joined #openstack-ceilometer | 07:44 | |
| *** fawadk has quit IRC | 07:44 | |
| *** ankita_wagh has joined #openstack-ceilometer | 07:49 | |
| *** Ala has joined #openstack-ceilometer | 08:01 | |
| *** jwcroppe has quit IRC | 08:01 | |
| *** jwcroppe has joined #openstack-ceilometer | 08:02 | |
| *** jwcroppe has quit IRC | 08:06 | |
| *** ankita_wagh has quit IRC | 08:13 | |
| *** jwcroppe has joined #openstack-ceilometer | 08:15 | |
| *** fawadkhaliq has quit IRC | 08:20 | |
| *** _nadya_ has joined #openstack-ceilometer | 08:28 | |
| *** promulo_ has joined #openstack-ceilometer | 08:33 | |
| *** pauldix has quit IRC | 08:34 | |
| *** lexholden has joined #openstack-ceilometer | 08:53 | |
| *** promulo_ has quit IRC | 09:06 | |
| openstackgerrit | Jie Li proposed openstack/ceilometer: Replace isotime() with utcnow() and isoformat() https://review.openstack.org/205884 | 09:23 |
| *** afazekas has joined #openstack-ceilometer | 09:28 | |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: Improves send_test_data tools https://review.openstack.org/203366 | 09:30 |
| *** _nadya_ has quit IRC | 09:33 | |
| *** _nadya_ has joined #openstack-ceilometer | 09:36 | |
| openstackgerrit | Julien Danjou proposed openstack/aodh: tests.api: remove unused argument/config option https://review.openstack.org/205969 | 09:41 |
| openstackgerrit | Julien Danjou proposed openstack/aodh: api: remove force_canonical option https://review.openstack.org/205970 | 09:41 |
| openstackgerrit | Julien Danjou proposed openstack/aodh: api: remove global conf and local pecan config https://review.openstack.org/205971 | 09:41 |
| openstackgerrit | Julien Danjou proposed openstack/aodh: rpc: remove global conf usage from notifier https://review.openstack.org/205972 | 09:41 |
| openstackgerrit | Julien Danjou proposed openstack/aodh: evaluator: remove global conf usage from threshold evaluator https://review.openstack.org/205973 | 09:41 |
| openstackgerrit | Julien Danjou proposed openstack/aodh: mongodb: replace custom retry code by retrying https://review.openstack.org/205974 | 09:41 |
| openstackgerrit | Julien Danjou proposed openstack/aodh: mongodb: stop relying on global conf object in utils https://review.openstack.org/205975 | 09:41 |
| jd__ | yeah I was bored yesterday on the train hihi | 09:41 |
| openstackgerrit | Merged openstack/aodh: Use user_id/project_id from service_credentials in alarm_change https://review.openstack.org/205782 | 09:48 |
| *** jwcroppe has quit IRC | 09:54 | |
| *** jwcroppe has joined #openstack-ceilometer | 09:56 | |
| *** dobson` has quit IRC | 10:21 | |
| *** Ala has quit IRC | 10:21 | |
| *** ildikov has quit IRC | 10:21 | |
| *** ildikov has joined #openstack-ceilometer | 10:22 | |
| openstackgerrit | Ilya Tyaptin proposed openstack/ceilometer: [MongoDB] Use a aggregate pipeline in statistics https://review.openstack.org/172340 | 10:25 |
| *** dobson has joined #openstack-ceilometer | 10:25 | |
| *** promulo_ has joined #openstack-ceilometer | 10:29 | |
| *** promulo_ has quit IRC | 10:29 | |
| *** _nadya_ has quit IRC | 10:31 | |
| *** _nadya_ has joined #openstack-ceilometer | 10:32 | |
| *** Ala has joined #openstack-ceilometer | 10:34 | |
| *** _nadya_ has quit IRC | 10:36 | |
| *** cdent has joined #openstack-ceilometer | 10:56 | |
| *** fawadkhaliq has joined #openstack-ceilometer | 10:59 | |
| openstackgerrit | liusheng proposed openstack/aodh: Keep alarm other attrs constantly after evaluating an alarm https://review.openstack.org/205732 | 11:08 |
| openstackgerrit | liusheng proposed openstack/aodh: Avoid getting alarm change notifier repeatedly https://review.openstack.org/205875 | 11:11 |
| *** fawadkhaliq has quit IRC | 11:21 | |
| *** fawadk has joined #openstack-ceilometer | 11:21 | |
| *** thorst1 has joined #openstack-ceilometer | 11:22 | |
| *** rbak has joined #openstack-ceilometer | 11:23 | |
| openstackgerrit | liusheng proposed openstack/aodh: Delete its corresponding history data when deleting an alarm https://review.openstack.org/204828 | 11:27 |
| *** rbak has quit IRC | 11:28 | |
| *** jwcroppe_ has joined #openstack-ceilometer | 11:30 | |
| *** _nadya_ has joined #openstack-ceilometer | 11:30 | |
| *** deepthi has quit IRC | 11:33 | |
| *** jwcroppe has quit IRC | 11:33 | |
| openstackgerrit | Merged openstack/ceilometer: Replace isotime() with utcnow() and isoformat() https://review.openstack.org/205884 | 11:41 |
| *** jwcroppe_ has quit IRC | 11:46 | |
| *** dikonoor has joined #openstack-ceilometer | 11:52 | |
| openstackgerrit | Merged openstack/aodh: Avoid getting alarm change notifier repeatedly https://review.openstack.org/205875 | 12:00 |
| *** jaosorior has joined #openstack-ceilometer | 12:03 | |
| openstackgerrit | Ilya Tyaptin proposed openstack/ceilometer: Initial separating unit and functional tests https://review.openstack.org/204128 | 12:07 |
| openstackgerrit | Ilya Tyaptin proposed openstack/ceilometer: Initial separating unit and functional tests https://review.openstack.org/204128 | 12:09 |
| *** fawadk has quit IRC | 12:13 | |
| jasonamyers | good morning | 12:23 |
| * cdent waves at jasonamyers | 12:24 | |
| *** gordc has joined #openstack-ceilometer | 12:26 | |
| *** edmondsw has joined #openstack-ceilometer | 12:30 | |
| *** pauldix has joined #openstack-ceilometer | 12:43 | |
| *** jruano has joined #openstack-ceilometer | 12:50 | |
| *** ddieterly has joined #openstack-ceilometer | 12:53 | |
| *** ityaptin has quit IRC | 12:56 | |
| *** ddieterly has quit IRC | 12:56 | |
| *** yprokule has quit IRC | 12:59 | |
| *** yprokule has joined #openstack-ceilometer | 13:00 | |
| *** eglynn has joined #openstack-ceilometer | 13:02 | |
| *** fawadkhaliq has joined #openstack-ceilometer | 13:03 | |
| *** julim has joined #openstack-ceilometer | 13:05 | |
| *** edmondsw has quit IRC | 13:08 | |
| *** tongli has joined #openstack-ceilometer | 13:08 | |
| *** pauldix has quit IRC | 13:12 | |
| *** thorst1 has quit IRC | 13:16 | |
| *** thorst1 has joined #openstack-ceilometer | 13:16 | |
| *** r-mibu has quit IRC | 13:22 | |
| *** r-mibu has joined #openstack-ceilometer | 13:22 | |
| openstackgerrit | gordon chung proposed openstack/ceilometer: add flexible grouping key https://review.openstack.org/204140 | 13:29 |
| openstackgerrit | Merged openstack/aodh: Use stevedore directive to document plugins https://review.openstack.org/205693 | 13:45 |
| cdent | gordc: recall last week me saying the collector was using a fair bit of cpu while idling? | 13:45 |
| cdent | a) it is just under mongo as we thought b) it appears to be in pymongo not ceilo code | 13:46 |
| openstackgerrit | Merged openstack/aodh: Fix the service entry of evaluator and notifier https://review.openstack.org/205746 | 13:46 |
| *** alexpilotti has quit IRC | 13:46 | |
| cdent | gordc: can you remind me of the pollster notifier combo we talked about investigating? | 13:47 |
| cdent | (for performance) | 13:47 |
| *** pauldix has joined #openstack-ceilometer | 13:50 | |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: storage: deprecates mongodb_replica_set option https://review.openstack.org/206073 | 13:51 |
| gordc | pollster notifier combo? | 13:51 |
| cdent | gordc: yeah, you said something was using more cpu and I said I'd look into it | 13:52 |
| cdent | but I don't recall the details | 13:52 |
| gordc | i think the two things i was found was having pollster spiked the CPU load when sending individual samples vs batch | 13:52 |
| gordc | ~2x the CPU load on compute agent and a longer CPU load on agent (the latter would be expected) | 13:53 |
| gordc | and the other item was that notification agent would never idle on my machine | 13:53 |
| cdent | and the assumption is that this is o.messaging overhead, yeah? | 13:54 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements https://review.openstack.org/205650 | 13:54 |
| gordc | cdent: first item or second? | 13:54 |
| gordc | brb. meeting | 13:54 |
| cdent | both? | 13:55 |
| *** julim_ has joined #openstack-ceilometer | 13:57 | |
| *** edmondsw has joined #openstack-ceilometer | 13:57 | |
| *** julim has quit IRC | 14:00 | |
| *** boris-42 has joined #openstack-ceilometer | 14:02 | |
| *** exploreshaifali has joined #openstack-ceilometer | 14:03 | |
| * sileht ETOOMUCH_JD_PATCHES | 14:05 | |
| *** yprokule has quit IRC | 14:05 | |
| *** yprokule has joined #openstack-ceilometer | 14:06 | |
| *** prad has joined #openstack-ceilometer | 14:11 | |
| jd__ | lol sileht | 14:15 |
| *** ityaptin has joined #openstack-ceilometer | 14:23 | |
| openstackgerrit | Michael Bayer proposed openstack/gnocchi: Switch to facade https://review.openstack.org/204669 | 14:24 |
| *** jruano has quit IRC | 14:25 | |
| *** titilambert is now known as Taytay | 14:25 | |
| *** Taytay is now known as titilambert | 14:25 | |
| *** exploreshaifali_ has joined #openstack-ceilometer | 14:26 | |
| *** exploreshaifali has quit IRC | 14:28 | |
| *** changbl has quit IRC | 14:32 | |
| gordc | cdent: i have no idea why notification agent doesn't idle for me. | 14:38 |
| gordc | cdent: i think the first one is just the added loaded of pushing x times the number of discrete messages just causing load. | 14:39 |
| gordc | debating whether it's better to move back to batch... or at the very least default to batch and have an option to send individual samples | 14:39 |
| *** fawadkhaliq has quit IRC | 14:43 | |
| *** r-daneel has joined #openstack-ceilometer | 14:46 | |
| *** jaypipes has joined #openstack-ceilometer | 14:48 | |
| *** shaifali__ has joined #openstack-ceilometer | 14:49 | |
| *** dikonoo has joined #openstack-ceilometer | 14:50 | |
| *** dikonoor has quit IRC | 14:50 | |
| *** exploreshaifali_ has quit IRC | 14:52 | |
| *** kbyrne has quit IRC | 14:56 | |
| cdent | gordc: a raise load/raised cpu usage doesn't necessarily mean something bad | 14:57 |
| cdent | it can mean we are doing more, more quickly | 14:57 |
| cdent | and thus making more efficient use of the cpu | 14:57 |
| cdent | in theory, if there is a sufficiently sized threadpool would should be able to accomodate for i/o latency better when _not_ batching | 14:58 |
| cdent | however, theory and reality rarely map | 14:58 |
| cdent | s/would/we/ | 14:58 |
| cdent | if you're seeing raised load but _not_ raised cpu usage (user or sys) and perhaps with raised "wa" usage then that could be a sign of a problem | 15:00 |
| cdent | if load is higher and cpu (real and sys) usage is higher, but there is still idle cpu and wa is not much changed, then what we've got is faster delivery of messages (at the cost of using more cpu) | 15:01 |
| *** alexpilotti has joined #openstack-ceilometer | 15:02 | |
| *** ddaskal has joined #openstack-ceilometer | 15:03 | |
| *** exploreshaifali_ has joined #openstack-ceilometer | 15:04 | |
| openstackgerrit | Ilya Tyaptin proposed openstack/ceilometer: Initial separating unit and functional tests https://review.openstack.org/204128 | 15:04 |
| *** shaifali__ has quit IRC | 15:04 | |
| *** exploreshaifali_ has quit IRC | 15:06 | |
| gordc | cdent: what's 'wa'? | 15:09 |
| *** kbyrne has joined #openstack-ceilometer | 15:10 | |
| cdent | that's "wait": it means the percentage of process that are paused waiting on i/o (if I recall correctly) | 15:10 |
| gordc | cdent: got it | 15:10 |
| cdent | sorry it is perce of cpu time that is waiting on i/o (indepdendent of number of processes) | 15:11 |
| gordc | i guess my concern would be that the compute agent is already putting quite a bit of load onto nova api... if users notice compute agent itself consuming load on the compute node, they'll start b*tching again | 15:11 |
| *** ildikov has quit IRC | 15:12 | |
| jasonamyers | gordc: that's they're secret they never stop complaining :P | 15:13 |
| cdent | gordc: we should make CONF.shuffle_time_before_polling_task default to the same value as the polling cycle | 15:14 |
| gordc | yeah, that's all our secrets | 15:14 |
| cdent | so that all the polling tasks are spread out | 15:14 |
| cdent | right now the default is to wake up and do everything | 15:14 |
| cdent | which is a _horrible_ idea | 15:14 |
| gordc | cdent: yeah, i don't mind implementing mandatory shuffle | 15:14 |
| openstackgerrit | Pradeep Kilambi proposed openstack/ceilometer: Update install docs with gnocchi dispatcher info https://review.openstack.org/206110 | 15:16 |
| cdent | (jd__: what you are and mike trying to accomplish with this enginefacade stuff. It feels like you and he aren't talking entirely clearly about the end goal. Or at least it is non-obvious to observers | 15:16 |
| gordc | regarding the previous load stuff, we need a way to measure faster. is it fair to say throughput the time it takes for all samples in poll cycle to reach collector | 15:16 |
| gordc | as oppose to time it takes for a single sample to reach collector? | 15:17 |
| cdent | that's going to vary hugely on whether or not shuffle is being used... | 15:17 |
| cdent | so a potential measurement is coming up with a way to describe latency in any given situation | 15:18 |
| cdent | how long it takes any given measure to traverse the system under different scernarios | 15:18 |
| cdent | but even then you need to know what matters | 15:19 |
| cdent | you'd tune differently for whether you are trying to maximize for any single measure | 15:19 |
| cdent | or all the measures | 15:19 |
| gordc | cdent: agreed. just want to know criteria for faster/better/etc | 15:19 |
| gordc | rough criteria | 15:19 |
| jasonamyers | with shuffle = one, without shuffle = all | 15:20 |
| cdent | I don't know well enough what a normal scenario/environment is for the pollster to give a target | 15:20 |
| cdent | without it feels a bit like making stuff up | 15:21 |
| gordc | we'll define one for our talk. | 15:21 |
| cdent | yeah, we'll need to, but it would be good to make sure it is somewhere associated with the real world | 15:22 |
| gordc | yeah, right now my criteria is high CPU load bad, low CPU load good... which as you've mention can be flawed. | 15:22 |
| gordc | ain't no real world. | 15:22 |
| cdent | do we have any data on what people expect or can tolerate? that is are people okay if a datapoint traverses the system in >60s (to make an extreme example)? | 15:23 |
| cdent | If not, why not? | 15:23 |
| gordc | cdent: we use to do anomoly/threat detection with some of the audit events... i'd assume that case you'd want "as quick as possible" | 15:24 |
| cdent | that then would be an argument in favor of not batching | 15:25 |
| gordc | yep. | 15:26 |
| cdent | (at the cost of increased load) | 15:26 |
| * gordc is leaning more to configurable option | 15:28 | |
| jasonamyers | gordc: +1 config | 15:28 |
| cdent | makes sense | 15:28 |
| jasonamyers | default to lower load | 15:28 |
| cdent | should be pretty easy to add | 15:28 |
| cdent | shall I go ahead? | 15:29 |
| gordc | i hate options but it is what it is. | 15:29 |
| cdent | hah | 15:29 |
| *** dobson has quit IRC | 15:29 | |
| gordc | cdent: yeah, if you have time. | 15:29 |
| gordc | agree with jasonamyers that we should default to 'batch'. | 15:29 |
| *** dikonoo has quit IRC | 15:30 | |
| cdent | yeah, I think so | 15:30 |
| cdent | It will keep me from getting distracted on much more hairbrained schemes | 15:30 |
| gordc | my attempt to reign in the craziness. | 15:31 |
| cdent | (I was thinking about ways to spread the polling smoothly over the entire cycle) | 15:32 |
| cdent | (not randomly) | 15:32 |
| gordc | ... i will let you go down that rabbit hole alone. | 15:32 |
| cdent | your lack of support is noted for future reprisals | 15:33 |
| gordc | ha | 15:33 |
| *** dobson has joined #openstack-ceilometer | 15:34 | |
| gordc | i actually have zero advice on your plans. i have no idea how to do that. | 15:34 |
| gordc | ityaptin: hey, were you able to sync up with pauldix on what we need to get influxdb working in gnocchi? | 15:35 |
| openstackgerrit | Merged openstack/ceilometer: Improves send_test_data tools https://review.openstack.org/203366 | 15:35 |
| *** afazekas has quit IRC | 15:36 | |
| openstackgerrit | Merged openstack/ceilometer: Close and dispose test database setup connections https://review.openstack.org/205138 | 15:40 |
| *** ildikov has joined #openstack-ceilometer | 15:48 | |
| openstackgerrit | gordon chung proposed openstack/ceilometer: add flexible grouping key https://review.openstack.org/204140 | 15:56 |
| *** thorst1 is now known as thorst1_away | 15:56 | |
| *** r-mibu has quit IRC | 15:58 | |
| *** mgarza_ has joined #openstack-ceilometer | 16:00 | |
| *** vishwanathj has joined #openstack-ceilometer | 16:01 | |
| *** dikonoor has joined #openstack-ceilometer | 16:03 | |
| *** fawadkhaliq has joined #openstack-ceilometer | 16:05 | |
| *** _nadya_ has quit IRC | 16:08 | |
| *** _nadya_ has joined #openstack-ceilometer | 16:08 | |
| *** lexholden has quit IRC | 16:12 | |
| *** _nadya_ has quit IRC | 16:13 | |
| *** Ala has quit IRC | 16:20 | |
| *** fawadkhaliq has quit IRC | 16:21 | |
| *** afazekas has joined #openstack-ceilometer | 16:36 | |
| *** _cjones_ has joined #openstack-ceilometer | 16:38 | |
| ityaptin | gordc: Hi! At Friday I checked most issues which I met during the gnocchi driver and these issues have been already fixed. Most problems were in GROUP BY requests and it have been changed to better side. | 16:39 |
| ityaptin | gordc: in influxdb 0.9.2 | 16:39 |
| *** lexholden has joined #openstack-ceilometer | 16:49 | |
| *** jwcroppe has joined #openstack-ceilometer | 16:49 | |
| *** afazekas has quit IRC | 16:58 | |
| openstackgerrit | Chris Dent proposed openstack/ceilometer: Add a batch_polled_samples configuration item https://review.openstack.org/206148 | 16:59 |
| cdent | gordc: that ^ took me 5 minutes to write the code and hours to fail to write a useful test | 16:59 |
| *** afazekas has joined #openstack-ceilometer | 17:00 | |
| openstackgerrit | Chris Dent proposed openstack/ceilometer: Add a batch_polled_samples configuration item https://review.openstack.org/206148 | 17:04 |
| *** yprokule has quit IRC | 17:05 | |
| *** thorst1_away is now known as thorst1 | 17:08 | |
| *** ankita_wagh has joined #openstack-ceilometer | 17:17 | |
| *** prad has quit IRC | 17:20 | |
| *** afazekas has quit IRC | 17:22 | |
| openstackgerrit | Chris Dent proposed openstack/ceilometer: Add a batch_polled_samples configuration item https://review.openstack.org/206148 | 17:24 |
| cdent | managed a resonable test ^ | 17:24 |
| *** vishwanathj has quit IRC | 17:32 | |
| *** eglynn has quit IRC | 17:34 | |
| *** prad has joined #openstack-ceilometer | 17:37 | |
| *** boris-42 has quit IRC | 17:40 | |
| *** pauldix_ has joined #openstack-ceilometer | 17:44 | |
| *** pauldix has quit IRC | 17:44 | |
| *** mgarza_ has quit IRC | 17:44 | |
| *** pauldix_ has quit IRC | 17:44 | |
| gordc | ityaptin: that's good to hear. | 17:48 |
| openstackgerrit | Ilya Tyaptin proposed openstack/ceilometer: Initial separating unit and functional tests https://review.openstack.org/204128 | 17:49 |
| ityaptin | gordc: Also I send to review a separating of unittests and functional tests | 17:49 |
| ityaptin | gordc: ^^^ Please, take a look, if you have a minute | 17:49 |
| *** afazekas has joined #openstack-ceilometer | 17:49 | |
| *** dikonoor has quit IRC | 17:51 | |
| gordc | ityaptin: is there a way to enable functional tests in gate right now? | 17:51 |
| gordc | so we can keep testing both? | 17:51 |
| *** pauldix has joined #openstack-ceilometer | 17:52 | |
| *** harlowja has joined #openstack-ceilometer | 17:52 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements https://review.openstack.org/205650 | 17:54 |
| ityaptin | gordc: yep, in current test we should comment 'check experimental' for running functional job on these context and if all works good I send to review patch with moving functional from experimental to regular | 17:54 |
| gordc | yeah. that would be good. | 17:55 |
| cdent | prad you back yet? | 17:55 |
| prad | cdent, in 1:1 | 17:56 |
| cdent | roger | 17:56 |
| ityaptin | cdent, jd__: Hi! I fixed notes and some issues in aggregation pipeline patch. Could you take a look https://review.openstack.org/#/c/172340/ ??? | 17:58 |
| * cdent looks | 17:58 | |
| *** vishwanathj has joined #openstack-ceilometer | 17:58 | |
| ityaptin | gordc: I will glad if you take a look too :) Thanks! | 17:58 |
| gordc | ityaptin: sure. will give it a try today | 18:01 |
| prad | cdent, i'm here now | 18:04 |
| cdent | i just wanted to see if gordc had his ears on enough now to talk abou the grenade thing, cuz I'll be gone shortly | 18:05 |
| prad | cdent, dint have time to explore what we spoke yet.. | 18:05 |
| prad | cool yea | 18:05 |
| prad | gordc, so here is the situation .. | 18:05 |
| cdent | yeah, I just wanted to bounce it off him in case he had something to say (no point trying the idea if he or someone else hates it) | 18:05 |
| prad | gordc, the grenade change i made to copy the meter.yaml wont work because, as per grenade the code should work with old existing config | 18:06 |
| prad | though in our case yaml is not really a config | 18:06 |
| * cdent kicks gordc | 18:07 | |
| prad | so one of the suggestions was to use pkg_resources and load the yaml from the code and override via config | 18:08 |
| prad | hehe | 18:08 |
| prad | he's probably ignoring us | 18:08 |
| cdent | well I know he hates me so that's cool, but it's rude to ignore you | 18:08 |
| prad | he hates me too, so i dont mind | 18:09 |
| cdent | ah well, then, no problem | 18:09 |
| prad | cdent, but yea, i see we do soemthing similar in sphinx | 18:10 |
| prad | cdent, to load translations | 18:10 |
| *** changbl has joined #openstack-ceilometer | 18:12 | |
| prad | cdent, if you need to leave thats fine, i can update you what he says.. meanwhile i'll give it a try this afternoon and see if it helps | 18:16 |
| cdent | not quite gone yet, I started late today | 18:16 |
| prad | cool | 18:16 |
| *** afazekas has quit IRC | 18:19 | |
| *** fawadkhaliq has joined #openstack-ceilometer | 18:25 | |
| * gordc wakes | 18:27 | |
| gordc | isn't the whole point of the grenade patch to handle the exceptional case of adding new file? | 18:28 |
| gordc | prad: ^ | 18:28 |
| cdent | gordc: the grenade patch will be rejected because it violates "the theory of upgrading" | 18:28 |
| gordc | prad: when we added the api paste file we had teh same thing. | 18:29 |
| prad | gordc, evidently not, the point is to abide by https://github.com/openstack-dev/grenade#theory-of-upgrade | 18:29 |
| cdent | but there are additional reasons to consider using pkg_resources in this case | 18:29 |
| prad | gordc, atleast thats what sdague told us | 18:29 |
| cdent | (and probably in a few others): We should put config in /etc/ceilometer, not defaults | 18:30 |
| cdent | defaults should be more permanent | 18:30 |
| gordc | cdent: so i'll be perfectly honest, i don't know difference between /etc/ceilometer and defaults. | 18:30 |
| cdent | exactly! | 18:30 |
| cdent | that's a significant problem | 18:30 |
| cdent | there are set of the things that the system must have to run | 18:31 |
| cdent | at the moment those things are required in /etc/ceilometer but ideally everything could run (but maybe do nothing) without any /etc/ceilometer | 18:32 |
| cdent | I was going to make this point long ago, but there was no point disrupting the way things are | 18:32 |
| cdent | but now this issue with grenade opens the door | 18:32 |
| gordc | can't we run now without /etc/ceilometer?... err except for ceilometer.conf | 18:32 |
| cdent | we maybe can | 18:33 |
| cdent | but once meters.yaml exists, we'll have some issues | 18:33 |
| gordc | the issue being tempest expects the meters. | 18:34 |
| gordc | i should clarify the above, what i meant was i don't know what "/etc/ceilometer/" is referring to and what "defaults" is referring to. | 18:36 |
| gordc | based on that, i still may or may not know the difference :) | 18:37 |
| cdent | let me try it from another angle | 18:37 |
| prad | gordc, anything that we're migrating to yaml as part of the existing code is essentially the defaults(for the issue we have in hand) | 18:37 |
| cdent | prad++ | 18:38 |
| prad | and custom meters user would want to add from liberty on will be in the usual config location as override | 18:38 |
| gordc | so we want to keep existing code and only make new meters declarative? | 18:41 |
| cdent | gordc, no | 18:41 |
| cdent | we want to treat the new yaml _as_ code | 18:41 |
| *** liusheng has quit IRC | 18:41 | |
| gordc | how would users disable it? | 18:41 |
| *** liusheng has joined #openstack-ceilometer | 18:41 | |
| *** boris-42 has joined #openstack-ceilometer | 18:42 | |
| cdent | setting parts to null, none, whatever the yaml is? | 18:42 |
| gordc | disable the 'coded' yaml i meant. | 18:42 |
| cdent | yeah: in the custom file in /etc/ceillometer (which merges with the coded yaml) they would declare something off | 18:43 |
| gordc | from what i undertand you want to hardcode existing meters in the yaml context. | 18:43 |
| cdent | (by a mechanism to be determined0 | 18:43 |
| cdent | that's what was already going to be the case: a _gigantic_ yaml file in /etc/ceilometer | 18:43 |
| gordc | how would they know what's on?.. or what the defaults are? | 18:43 |
| cdent | now it'll be a smaller one | 18:43 |
| cdent | gordc: how do they now? | 18:43 |
| gordc | cdent: they don't. which is why the yaml makes it more obvious | 18:44 |
| cdent | only if it is short | 18:44 |
| gordc | well if we ignore ildikov and we create yamls per project it would be :) | 18:44 |
| cdent | I agree with you that it is a pain in the butt | 18:44 |
| cdent | but the grenade thing is a hoop we need to get through and this is a way to do it that also has some semantic sense | 18:45 |
| gordc | for me, i'm not really sure how the meter.yaml is different from this: https://review.openstack.org/#/c/114278/ | 18:45 |
| cdent | I'm not 100% sold on the idea, its just something that I've had percolating for a while, so prad and I decided to bring it up | 18:45 |
| prad | so | 18:46 |
| cdent | gordc: I think it is just a matter of time passing | 18:46 |
| prad | when we say yaml as python code, would we ahve yaml file as is and just load it via pkg_resources | 18:46 |
| cdent | prad, that's what I think, yeah | 18:46 |
| ildikov | gordc: you mean create files for projects and go for the dricetion of storing them at those projects repo? | 18:46 |
| prad | cdent, something like pkg_resources.resource_string(__name__, "meters.yaml") and move meter.yaml under ceilo/meter or soemthing | 18:47 |
| prad | cdent, essentially move it out of etc | 18:47 |
| gordc | ildikov: lol yeah. not plannig on it now. but just to explain multiple smaller files item | 18:48 |
| gordc | prad: what would happen from user perspective? they have to edit package? | 18:48 |
| cdent | something like that yeah. I don't know the exact details; it seems likely that there's some kind of obscurationist way of using pkg_resources which is de rigueur in openstack | 18:49 |
| prad | gordc, yea from users perspective, its not very ideal.. they would expect it to be in some *usual* config location | 18:49 |
| ildikov | gordc: ok :) | 18:49 |
| cdent | no | 18:49 |
| cdent | the user would not edit the pkg_resource file | 18:49 |
| cdent | they would override it | 18:49 |
| cdent | wouldn't they? | 18:50 |
| *** ankita_wagh has quit IRC | 18:50 | |
| prad | cdent, how would you disable something then? .. so say i want to disable sahara metrics | 18:50 |
| prad | which are part of the pkg resources file | 18:50 |
| cdent | can you link me to the current file so I can have a look | 18:50 |
| prad | cdent, https://github.com/openstack/ceilometer/blob/master/etc/ceilometer/meters.yaml | 18:51 |
| * cdent hadn't realized it had already merged | 18:51 | |
| *** ankita_wagh has joined #openstack-ceilometer | 18:51 | |
| gordc | prad: yeah, that'd be very strange workflow. | 18:51 |
| cdent | prad I would think disabling comes in two forms: | 18:52 |
| *** _nadya_ has joined #openstack-ceilometer | 18:52 | |
| cdent | a) the notifications never happen in the first place, so no worries | 18:53 |
| cdent | b) the override.yaml (not its real name) would have the same structure, and the sequence on metric: would be merged with the existing one. to disable you could do something like: {name: "image.size", event_type: null} | 18:54 |
| cdent | yes, they need to look at the pkg resources file to know what the various names are, but: | 18:55 |
| prad | cdent, yea so we'll have to duplicate the meter definitions in both files? | 18:55 |
| cdent | a) that's how it works now, you have to edit the entry points | 18:55 |
| prad | or yea look at the pkg resources file | 18:55 |
| cdent | b) we can make a tool that does that | 18:55 |
| cdent | prad, no not duplicate | 18:55 |
| cdent | the override file only changes stuff | 18:56 |
| cdent | it can add meters | 18:56 |
| cdent | or disable existing meters | 18:56 |
| prad | cdent, my main worry is how would i as a user know for example say i have sahara meters defined .. without looking at pkg resources file | 18:56 |
| cdent | ceilometer-config-inspector-tool --list-meters ? | 18:56 |
| cdent | (again not its real name) | 18:56 |
| prad | if i have to look at this file.. thats a extra step we're introducing in the work flow.. which is may be ok, i'm not quite sure | 18:56 |
| gordc | this seems like we're moving the hassle of coding meters to not the user needs to do a lot more stff. | 18:57 |
| gordc | i use 'a lot' loosely | 18:57 |
| cdent | gordc: no, becuae right now the user has no control over anything | 18:57 |
| cdent | (unless they go digging in entry_points.txt) | 18:57 |
| cdent | you guys keep talking about a work flow that doesn't currently exist | 18:58 |
| prad | correct, which is why we're moving to yaml and making it easier for user to edit.. if we move the meters back to some place else arent we ending up with the same? | 18:59 |
| cdent | no, because we're provide an override/custom file | 18:59 |
| cdent | which allows them control they currently don't have: add custome metrics, disabled ones they don't like | 18:59 |
| cdent | the interface may not be perfect yet, but it is an improvement | 19:00 |
| prad | yea understand the override and i'm with you on that. I like the overall idea, just one piece concerns me which is the pkg resources file the user has to go inspect to see whats already there and come override it in another file. | 19:01 |
| gordc | cdent: different... arguably an improvement. i think it sort of worries me that we needed so many lines to explain how to override/disable stuff. | 19:01 |
| prad | is that too much distraction for the user vs just going in and changing the meter.yaml as we have now | 19:01 |
| cdent | even if the meters.yaml file were in /etc/ceiloemter i'd still argue that there should be an override file instead of edit in place | 19:01 |
| cdent | gordc, we can probably do much better than my suggestion, I came up with that in the space of this conversation | 19:02 |
| cdent | you could also have a 'disable:' key in the file that just takes a list of meters | 19:02 |
| gordc | cdent: understood. for me, the story went from: "to enable, a meter. add fields x, y, z to meters.yaml and your good" | 19:03 |
| gordc | now it's that, and "to disable something, do this. to override something do this, etc..." | 19:04 |
| cdent | I'm not sure that's a function of just this idea, I think that's a function of actually talking through the use cases more | 19:04 |
| gordc | "etc..." is to play up the hyperbole and anger you :) | 19:04 |
| cdent | you'll have to try harder than that! | 19:04 |
| cdent | Besides I wouldn't get angry with you, I'd get angry with the system known as openstack for which you have stockhold syndrome ;) | 19:05 |
| cdent | s/hold/holm/ | 19:05 |
| gordc | lol, it's so nice though. it locked me in this hellhole but pays me. | 19:05 |
| *** r-mibu has joined #openstack-ceilometer | 19:05 | |
| cdent | you get paid! | 19:06 |
| * cdent is getting a raw deal | 19:06 | |
| gordc | lol yeah, your masters don't play the confusion game. | 19:06 |
| gordc | for me, i just want to make sure we're not jumping through hoops to satify grenade. | 19:07 |
| prad | i guess there are not many options.. other than somehow convince grenade guys to include this change | 19:07 |
| cdent | I think even without grenade it is good that we talk about all three of enable, disable and modify | 19:07 |
| cdent | because we hadn't really though much beyond "add another one" | 19:08 |
| gordc | what's to stop us from just hard coding it a cycle and then removing it next cycle. | 19:08 |
| cdent | and apparently it is more complex than we had been thinking | 19:08 |
| cdent | gordc: that assume that people upgrade in a clean fashion | 19:09 |
| gordc | i think the original story is still pretty simple. you want a meter, it needs to be in this file. | 19:09 |
| cdent | that begs the question about how to bug diagnosis | 19:10 |
| gordc | in the case of api-paste file. is the idea that if there's no workaround, only then can you add grenade exception? | 19:10 |
| cdent | my system will never be like their system if they are editing the only file that has meter.yaml info in it | 19:10 |
| cdent | it would be better if I could tell them: mv your custom file out so that the core file takes over | 19:10 |
| cdent | is the bug still there? | 19:10 |
| cdent | gordc: it's not clear, it kind of seemed like a back port would be needed | 19:10 |
| prad | what i dont get is, we should have a situation where a new config file was introduced in one release and needs to be included.. if they dont allow this how would it work? | 19:11 |
| cdent | I gotta go | 19:11 |
| prad | in our case especially when existing tempest tests depend on it | 19:11 |
| gordc | cdent: kk i'll think some more i guess. | 19:11 |
| cdent | if you guys get anywhere may you could summarize to the mailing list and make it visible to the [qa] gang too so that we can get input from more than just us | 19:11 |
| cdent | we are working on a lot of assumptions | 19:12 |
| cdent | have a good eve all | 19:12 |
| *** cdent has quit IRC | 19:12 | |
| prad | i mean, i get their point that code should work with old config as you upgrade.. but in some cases, you introduce new functionality that requires new config | 19:12 |
| prad | only way to upgrade is include the new file? | 19:13 |
| gordc | i will get in trouble if i talk more. | 19:13 |
| gordc | lol | 19:13 |
| prad | lol | 19:13 |
| gordc | prad: what if we had two copies. | 19:13 |
| gordc | the pkg_resources version | 19:13 |
| gordc | and we'd only use iff we couldn't find meter.yal | 19:13 |
| gordc | and we never ever promote the internal version | 19:14 |
| gordc | and then we kill it | 19:14 |
| prad | yea we could, its just ugly to have to maintain two versions and it is bound to bite us at some point | 19:14 |
| gordc | next cycle... | 19:14 |
| gordc | like right away | 19:14 |
| gordc | or you can maintain the internal version enough to trick grenade. | 19:15 |
| gordc | just have cpu_util and glance notification | 19:15 |
| gordc | or whatever tempest tests | 19:15 |
| prad | hmm yea just to satisfy grenade.. i guess | 19:16 |
| gordc | the disable logic is not something i want to support. | 19:16 |
| gordc | i don't understand the argument really. | 19:16 |
| prad | i agree, the disable approach seems a bit backwards to me, a lot more user has to do to figure out | 19:16 |
| prad | for code pov its definitely cleaner impl | 19:16 |
| prad | but end user will have to go a extra step to do what they want | 19:17 |
| gordc | if we split services there's no way you're going to hardcode a fallback service just in case the person didn't start the new service. | 19:17 |
| prad | if i understood correctly | 19:17 |
| prad | right | 19:17 |
| prad | lemme play around with the pkg_resource approach and see if grenade is happy | 19:18 |
| prad | at least something to get this moving | 19:18 |
| gordc | we should just create dummy.yaml files so we can use them in future cycles | 19:18 |
| prad | gordc, you mean per project files? | 19:19 |
| gordc | prad: no i mean just blank files that are never referenced. | 19:24 |
| gordc | prad: but we can just use them later so grenade never complains :) | 19:24 |
| *** safchain has quit IRC | 19:26 | |
| *** ildikov has quit IRC | 19:28 | |
| *** thorst1 has quit IRC | 19:36 | |
| *** vishwanathj has quit IRC | 19:37 | |
| *** fawadkhaliq has quit IRC | 19:44 | |
| *** harlowja has quit IRC | 19:45 | |
| *** thorst1 has joined #openstack-ceilometer | 19:55 | |
| *** harlowja has joined #openstack-ceilometer | 20:01 | |
| *** _nadya_ has quit IRC | 20:15 | |
| prad | gordc, so the duplicate(backup) yaml seems to be working.. the grenade script dint fail on my vm. though i left the yaml in ceilometer/meter/ dir.. is there a better place to put it? | 20:16 |
| *** _nadya_ has joined #openstack-ceilometer | 20:16 | |
| gordc | prad: somewhere hidden. | 20:17 |
| gordc | prad: i'm just wondering if we want to start handling it as multiple files now. | 20:17 |
| prad | gordc, like just have glance ones separate and just duplicate that file? | 20:18 |
| gordc | i guess just hide it somewhere for now. i don't really have a preference where tbh | 20:18 |
| gordc | prad: nah, i thought about it a bit more. i don't think we gain anything | 20:18 |
| gordc | i think it's cool to hide the single file somewhere. | 20:19 |
| prad | k cool | 20:19 |
| gordc | prad: while i have your attention, do you know why we don't default to generic resource when we don't find matching resource. | 20:19 |
| prad | gordc, in gnocchi? | 20:20 |
| gordc | the dispatcher | 20:20 |
| *** _nadya_ has quit IRC | 20:20 | |
| gordc | right now we try to find specific resource. if we find no match, we'll drop it | 20:21 |
| gordc | makes sense to just store it as generic no? | 20:21 |
| prad | gordc, perhaps we havent gotten to adding that support.. we have generic resource in gnocchi, so should be doable | 20:21 |
| gordc | kk | 20:22 |
| gordc | seems to be dropping a good chunk of data right now | 20:22 |
| *** yassine has joined #openstack-ceilometer | 20:23 | |
| *** vishwanathj has joined #openstack-ceilometer | 20:29 | |
| *** ankita_w_ has joined #openstack-ceilometer | 20:32 | |
| *** ankita_wagh has quit IRC | 20:33 | |
| *** ankita_wagh has joined #openstack-ceilometer | 20:34 | |
| *** ankita_w_ has quit IRC | 20:34 | |
| *** kfox1111_ has joined #openstack-ceilometer | 20:37 | |
| *** julim_ has quit IRC | 20:38 | |
| kfox1111_ | project_id in the meter's doesn't seem to be an openstack uuid. what is it? | 20:38 |
| gordc | kfox1111_: we discovered this recently. | 20:40 |
| gordc | http://lists.openstack.org/pipermail/openstack-dev/2015-June/068104.html | 20:40 |
| gordc | ceilometer doesn't generate ids... we just pull ids from payloads. | 20:41 |
| gordc | it's really two cases: ceilometer is pulling the wrong field or more likely, the service is generating some made up id. | 20:42 |
| gordc | it's probably the latter... that said, which meter are you referring to? | 20:42 |
| kfox1111_ | confusing since, I'm pulling 'instance' metters by instance id and seeing stuff like: | 20:42 |
| kfox1111_ | dad318e0-a2fb-4663-ab43-18c595818faa bf07b33d835e41e8b076843e291c1111 | 20:42 |
| kfox1111_ | where the first is the instance id, adn the second the project_id. | 20:43 |
| gordc | which release are you using? i can take a look at code | 20:43 |
| kfox1111_ | juno. | 20:44 |
| kfox1111_ | looks too short to be the uuid without the dashes. | 20:44 |
| prad | gordc, so was thinking, instead of duplicating the file via pkg_resources, why cant we add a data_files in setup.cfg ? wouldnt that work? | 20:47 |
| prad | gordc, something like https://gist.github.com/pkilambi/07bfa6250327729e6381 | 20:47 |
| *** harlowja_ has joined #openstack-ceilometer | 20:47 | |
| kfox1111_ | is it a md5sum of the uuid or something? | 20:49 |
| gordc | kfox1111_: hmm. i'm not sure instance meters has anything special done to it. we seem to be just pulling a value. | 20:49 |
| kfox1111_ | how strange. | 20:49 |
| gordc | prad: possibly. tbh i'm not sure how that works. | 20:50 |
| *** harlowja has quit IRC | 20:51 | |
| prad | gordc, though i dont know if this will reflect at upgrade time, which is what grenade complains | 20:52 |
| kfox1111_ | something's mutated the value somehow... | 20:53 |
| gordc | and it appears fine in nova? | 20:53 |
| kfox1111_ | I think so. I don't have notification messages to verify. | 20:54 |
| *** tongli has quit IRC | 20:57 | |
| openstackgerrit | Pradeep Kilambi proposed openstack/ceilometer: Enable entry points for new declarative meters https://review.openstack.org/203032 | 20:58 |
| gordc | kfox1111_: for the instance meter, is the metadata extremely verbose? | 20:58 |
| gordc | trying to figure out if it's coming from notification or pollster | 20:58 |
| kfox1111_ | let me see if I can get that... | 20:58 |
| gordc | more specifically does it contain more than the attributes defined here: https://github.com/openstack/ceilometer/blob/stable/juno/ceilometer/compute/pollsters/util.py#L38-L72 | 20:59 |
| kfox1111_ | khow off hand what the field's named? | 21:01 |
| kfox1111_ | resource_metadata? | 21:03 |
| gordc | does your metadata have 'launched_at'? | 21:03 |
| kfox1111_ | keys() are: [u'node', u'state_description', u'image_meta.base_image_ref', u'vcpus', u'event_type', u'disk_gb', u'availability_zone', u'terminated_at', u'ephemeral_gb', u'instance_type_id', u'instance_flavor_id', u'image_meta.disk_format', u'image_meta.container_format', u'user_id', u'deleted_at', u'reservation_id', u'os_type', u'image_ref_url', u'access_ip_v6', u'audit_period_beginning', u'root_gb', u'display_name', u'instance_type', u'tenant_i | 21:04 |
| gordc | kfox1111_: that's coming from notification | 21:04 |
| kfox1111_ | yeah.... and inside: u'tenant_id': u'f5018fb2c0304fea9283d7f92e790db9' | 21:05 |
| kfox1111_ | so that one's short too. | 21:05 |
| gordc | it doesn't look like we modify resource_id in ceilometer. unless there's a hidden bug somewhere. | 21:06 |
| gordc | do you have resources where it doesn't have the 'project_id' appended? | 21:07 |
| kfox1111_ | not that I've found. | 21:07 |
| kfox1111_ | haven't looked too hard though. | 21:07 |
| kfox1111_ | I'm trying to do a query to show all instances associated with a project. | 21:07 |
| kfox1111_ | its not helping when the project id isnt an id. :/ | 21:07 |
| gordc | i'm inclined to say it's something in nova but you can open a bug if you'd like | 21:08 |
| gordc | kfox1111_: we tried to normalise it in ceilometer but when we tried we realised it would break other projects. | 21:08 |
| kfox1111_ | k. thanks. I've started asking in the nova channel. | 21:09 |
| gordc | i didn't get much feedback but if you look at the model of all the projects in openstack, the ids are set at varchar255 (which is way beyond uuid) | 21:09 |
| gordc | so i'm guessing it's a big fix. | 21:09 |
| kfox1111_ | I'm seeing the strange projectid's in nova.instance.project_id. so I think it is for sure an issue there. | 21:16 |
| kfox1111_ | is ceilometer going to even allow a project to fetch their own meters for a project_id that looks this strange? | 21:16 |
| gordc | kfox1111_: i'm not sure. it depends on what the role user has in keystone... i'd guess no though. | 21:19 |
| gordc | if you can track down the issue in nova i think it'd be good to bring it up on mailing list | 21:19 |
| gordc | no one seemed very concern when i brought it up but maybe if others start complaining we can start standarding the ids we have in OpenStack | 21:21 |
| kfox1111_ | yeah. the lack of a horizon interface for ceilometer for users probably means users haven't actually ran acorss this yet. :/ | 21:22 |
| *** ddaskal has quit IRC | 21:25 | |
| *** htruta_ has joined #openstack-ceilometer | 21:36 | |
| *** yassine has quit IRC | 21:42 | |
| *** ccrouch has quit IRC | 21:42 | |
| *** ankita_w_ has joined #openstack-ceilometer | 21:43 | |
| *** ankita_wagh has quit IRC | 21:45 | |
| *** julim has joined #openstack-ceilometer | 21:46 | |
| openstackgerrit | gordon chung proposed openstack/ceilometer: fix gnocchi resources yaml https://review.openstack.org/206235 | 21:53 |
| gordc | prad: what do you guys plan to do in gnocchi with bad ids? | 21:56 |
| *** ddaskal has joined #openstack-ceilometer | 22:02 | |
| kfox1111_ | odd. I see the same strange id's in keystone's db. | 22:13 |
| kfox1111_ | a different string then what keystone tenant-list shows for the tenant... | 22:14 |
| *** edmondsw has quit IRC | 22:14 | |
| gordc | kfox1111_: that's quite the rabbit hole | 22:14 |
| kfox1111_ | this one's totally new to me. I've had no idea the db's have stored a different version of the uuid all this time. | 22:17 |
| *** tongli has joined #openstack-ceilometer | 22:18 | |
| gordc | kfox1111_: heading home. sorry can't help debug. feel free to create a mailing list item or bug... laters. | 22:20 |
| *** gordc has quit IRC | 22:21 | |
| *** harlowja_ has quit IRC | 22:21 | |
| *** harlowja has joined #openstack-ceilometer | 22:21 | |
| *** prad has quit IRC | 22:28 | |
| *** alexpilotti has quit IRC | 22:54 | |
| *** r-daneel has quit IRC | 23:02 | |
| *** jwcroppe has quit IRC | 23:02 | |
| *** jwcroppe has joined #openstack-ceilometer | 23:03 | |
| *** arahal has joined #openstack-ceilometer | 23:06 | |
| *** jaosorior has quit IRC | 23:11 | |
| *** thorst1 has quit IRC | 23:12 | |
| *** jaypipes has quit IRC | 23:19 | |
| *** arahal has quit IRC | 23:37 | |
| *** arahal has joined #openstack-ceilometer | 23:37 | |
| *** arahal has quit IRC | 23:37 | |
| *** thorst has joined #openstack-ceilometer | 23:51 | |
| *** tongli has quit IRC | 23:54 | |
| *** flwang1 has joined #openstack-ceilometer | 23:58 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!