*** rbak has quit IRC | 00:06 | |
*** jwcroppe has quit IRC | 00:10 | |
*** gordc has quit IRC | 00:10 | |
*** ddieterly has quit IRC | 00:11 | |
*** cdent has quit IRC | 00:40 | |
*** jfluhmann has quit IRC | 00:47 | |
*** diogogmt has joined #openstack-telemetry | 00:56 | |
*** cheneydc has joined #openstack-telemetry | 01:00 | |
*** zqfan has joined #openstack-telemetry | 01:09 | |
*** ljxiash_ has quit IRC | 01:09 | |
openstackgerrit | Merged openstack/aodh: Replace LOG.warn with LOG.warning https://review.openstack.org/262444 | 01:25 |
---|---|---|
*** ljxiash has joined #openstack-telemetry | 01:36 | |
*** ljxiash has quit IRC | 01:37 | |
*** ljxiash has joined #openstack-telemetry | 01:37 | |
*** Liuqing has joined #openstack-telemetry | 01:37 | |
openstackgerrit | Merged openstack/gnocchi: Change LOG.warn to LOG.warning https://review.openstack.org/263107 | 01:38 |
*** prashantD has quit IRC | 01:40 | |
openstackgerrit | Merged openstack/ceilometer: gnocchi: fix stack resource type https://review.openstack.org/260592 | 01:43 |
openstackgerrit | Merged openstack/ceilometer: Trivial: Remove vim header from source files https://review.openstack.org/262429 | 01:43 |
*** dave-mccowan has quit IRC | 01:48 | |
*** ljxiash has quit IRC | 01:49 | |
*** thorst has joined #openstack-telemetry | 01:49 | |
*** thorst_ has joined #openstack-telemetry | 01:51 | |
*** ljxiash has joined #openstack-telemetry | 01:51 | |
*** thorst has quit IRC | 01:53 | |
*** ddieterly has joined #openstack-telemetry | 02:02 | |
*** diazjf has joined #openstack-telemetry | 02:03 | |
openstackgerrit | april proposed openstack/python-ceilometerclient: Bug fix for 1511346 https://review.openstack.org/263530 | 02:22 |
*** thorst_ has quit IRC | 02:27 | |
*** thorst has joined #openstack-telemetry | 02:28 | |
*** thorst has quit IRC | 02:29 | |
*** thorst has joined #openstack-telemetry | 02:29 | |
openstackgerrit | liusheng proposed openstack/aodh: Refactor Gnocchi and threshold evaluators https://review.openstack.org/261225 | 02:29 |
openstackgerrit | Merged openstack/aodh: Fix an minor error in test_hbase_table_utils.py https://review.openstack.org/261394 | 02:36 |
openstackgerrit | Merged openstack/aodh: Trivial: Remove vim header from source files https://review.openstack.org/262427 | 02:36 |
openstackgerrit | Merged openstack/gnocchi: indexer: do not store useless revision https://review.openstack.org/260559 | 02:36 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements https://review.openstack.org/260249 | 02:39 |
*** llu has quit IRC | 02:43 | |
*** llu has joined #openstack-telemetry | 02:45 | |
*** liamji has joined #openstack-telemetry | 02:47 | |
*** llu has quit IRC | 02:51 | |
*** llu has joined #openstack-telemetry | 02:53 | |
*** diazjf has quit IRC | 03:05 | |
openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/aodh: Replace deprecated LOG.warn with warning https://review.openstack.org/263319 | 03:07 |
*** diazjf has joined #openstack-telemetry | 03:12 | |
*** prashantD has joined #openstack-telemetry | 03:12 | |
*** diazjf has quit IRC | 03:13 | |
*** vishwanathj has joined #openstack-telemetry | 03:27 | |
*** llu has quit IRC | 03:29 | |
*** llu has joined #openstack-telemetry | 03:31 | |
*** PSS has quit IRC | 03:31 | |
*** links has joined #openstack-telemetry | 03:52 | |
openstackgerrit | april proposed openstack/python-ceilometerclient: Improve ceilometer alarm enabled input parameter validation https://review.openstack.org/263530 | 03:54 |
*** cheneydc has left #openstack-telemetry | 03:55 | |
*** llu has quit IRC | 04:05 | |
*** llu has joined #openstack-telemetry | 04:08 | |
*** ljxiash has quit IRC | 04:25 | |
*** r-mibu has joined #openstack-telemetry | 04:25 | |
*** ljxiash has joined #openstack-telemetry | 04:25 | |
*** ljxiash has quit IRC | 04:26 | |
*** ljxiash has joined #openstack-telemetry | 04:26 | |
*** thorst has quit IRC | 04:27 | |
*** thorst has joined #openstack-telemetry | 04:28 | |
*** thorst has quit IRC | 04:36 | |
*** jwcroppe has joined #openstack-telemetry | 04:54 | |
openstackgerrit | Merged openstack/aodh: Trival: Remove unused logging import https://review.openstack.org/261643 | 04:55 |
*** ddieterly has quit IRC | 05:01 | |
*** PsionTheory has quit IRC | 05:03 | |
*** Liuqing has quit IRC | 05:05 | |
*** Liuqing has joined #openstack-telemetry | 05:06 | |
*** cheneydc has joined #openstack-telemetry | 05:23 | |
*** julim has joined #openstack-telemetry | 05:24 | |
*** thorst has joined #openstack-telemetry | 05:34 | |
*** yprokule has joined #openstack-telemetry | 05:40 | |
*** jwcroppe has quit IRC | 05:42 | |
*** thorst has quit IRC | 05:42 | |
*** Liuqing has quit IRC | 05:51 | |
*** yarkot has joined #openstack-telemetry | 05:55 | |
*** prashantD has quit IRC | 06:01 | |
*** ddieterly has joined #openstack-telemetry | 06:02 | |
*** prashantD has joined #openstack-telemetry | 06:06 | |
*** ddieterly has quit IRC | 06:07 | |
*** prashantD has quit IRC | 06:19 | |
*** yarkot has quit IRC | 06:20 | |
*** ildikov_afk is now known as ildikov | 06:22 | |
*** thorst has joined #openstack-telemetry | 06:39 | |
*** Pandy has joined #openstack-telemetry | 06:42 | |
*** jwcroppe has joined #openstack-telemetry | 06:43 | |
*** r-mibu has quit IRC | 06:44 | |
*** thorst has quit IRC | 06:47 | |
*** jwcroppe has quit IRC | 06:47 | |
openstackgerrit | april proposed openstack/python-ceilometerclient: Improve ceilometer alarm enabled input parameter validation https://review.openstack.org/263530 | 06:54 |
*** Pandy has quit IRC | 06:59 | |
*** maestro1 has joined #openstack-telemetry | 07:01 | |
*** ddieterly has joined #openstack-telemetry | 07:04 | |
*** ddieterly has quit IRC | 07:09 | |
openstackgerrit | april proposed openstack/python-ceilometerclient: Improve ceilometer alarm enabled input parameter validation https://review.openstack.org/263530 | 07:27 |
openstackgerrit | april proposed openstack/python-ceilometerclient: Improve ceilometer alarm enabled input parameter validation https://review.openstack.org/263530 | 07:29 |
*** belmoreira has joined #openstack-telemetry | 07:30 | |
*** maestro1 has quit IRC | 07:48 | |
*** _nadya_ has joined #openstack-telemetry | 07:52 | |
*** jwcroppe has joined #openstack-telemetry | 07:52 | |
*** _nadya_ has quit IRC | 07:53 | |
openstackgerrit | liusheng proposed openstack/aodh: Add composite rule rule API support https://review.openstack.org/257722 | 07:56 |
*** safchain has joined #openstack-telemetry | 07:58 | |
*** ddieterly has joined #openstack-telemetry | 08:04 | |
openstackgerrit | Julien Danjou proposed openstack/gnocchi: swift: fix typo in unprocessed measures deletion https://review.openstack.org/263621 | 08:08 |
*** ddieterly has quit IRC | 08:09 | |
openstackgerrit | Zi Lian Ji proposed openstack/python-gnocchiclient: Supported resources list cli support https://review.openstack.org/263623 | 08:11 |
*** r-mibu has joined #openstack-telemetry | 08:14 | |
openstackgerrit | Chaozhe Chen proposed openstack/ceilometer: Replace LOG.warn with LOG.warning https://review.openstack.org/262442 | 08:15 |
*** jwcroppe has quit IRC | 08:22 | |
*** maestro has joined #openstack-telemetry | 08:30 | |
*** shardy has joined #openstack-telemetry | 08:33 | |
*** maestro has left #openstack-telemetry | 08:35 | |
*** maestro has joined #openstack-telemetry | 08:35 | |
openstackgerrit | liusheng proposed openstack/aodh: Add composite rule rule API support https://review.openstack.org/257722 | 08:45 |
*** idegtiarov_ has joined #openstack-telemetry | 08:48 | |
*** thorst has joined #openstack-telemetry | 08:49 | |
*** _nadya_ has joined #openstack-telemetry | 08:53 | |
*** thorst has quit IRC | 08:56 | |
*** _nadya_ has quit IRC | 08:58 | |
openstackgerrit | Merged openstack/ceilometer: [refactor] remove redundant import of options https://review.openstack.org/257785 | 08:59 |
*** ddieterly has joined #openstack-telemetry | 09:05 | |
*** _nadya_ has joined #openstack-telemetry | 09:06 | |
*** _nadya_ has quit IRC | 09:08 | |
*** ddieterly has quit IRC | 09:09 | |
*** achatterjee has joined #openstack-telemetry | 09:11 | |
openstackgerrit | Zi Lian Ji proposed openstack/python-gnocchiclient: Supported resources list cli support https://review.openstack.org/263623 | 09:13 |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: collector: Use oslo.messaging batch listener https://review.openstack.org/234831 | 09:13 |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: notification: Use oslo.messaging batch listener https://review.openstack.org/250814 | 09:13 |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: remove useless notification listener helper https://review.openstack.org/250815 | 09:13 |
*** mattyw has joined #openstack-telemetry | 09:14 | |
*** cheneydc has quit IRC | 09:17 | |
*** cheneydc has joined #openstack-telemetry | 09:18 | |
openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/gnocchi: Replace deprecated LOG.warn with LOG.warning https://review.openstack.org/263373 | 09:18 |
*** yassine__ has joined #openstack-telemetry | 09:19 | |
openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/aodh: Replace deprecated LOG.warn with LOG.warning https://review.openstack.org/263319 | 09:19 |
*** achatterjee has quit IRC | 09:19 | |
*** achatterjee has joined #openstack-telemetry | 09:19 | |
*** eglynn has joined #openstack-telemetry | 09:25 | |
*** eglynn has quit IRC | 09:25 | |
*** eglynn has joined #openstack-telemetry | 09:26 | |
*** liamji has quit IRC | 09:42 | |
*** thorst has joined #openstack-telemetry | 09:54 | |
*** cheneydc has quit IRC | 10:00 | |
*** thorst has quit IRC | 10:01 | |
*** boris-42 has quit IRC | 10:03 | |
*** ddieterly has joined #openstack-telemetry | 10:06 | |
*** _nadya_ has joined #openstack-telemetry | 10:09 | |
*** ddieterly has quit IRC | 10:10 | |
*** _nadya_ has quit IRC | 10:14 | |
*** mattyw has quit IRC | 10:22 | |
*** mattyw has joined #openstack-telemetry | 10:22 | |
openstackgerrit | OpenStack Proposal Bot proposed openstack/ceilometer: Updated from global requirements https://review.openstack.org/260249 | 10:25 |
*** thorst has joined #openstack-telemetry | 10:59 | |
*** cdent has joined #openstack-telemetry | 11:02 | |
jd__ | so we really are getting spammed by newbie commiters sending random cleanup patches | 11:05 |
jd__ | be careful when reviewing | 11:05 |
jd__ | not everything is… worth it | 11:05 |
*** thorst has quit IRC | 11:06 | |
cdent | Do you have a particular place where you draw the line? Many tiny improvements are a net positive if the revew load isn't too much of a bear | 11:06 |
*** ddieterly has joined #openstack-telemetry | 11:07 | |
*** _nadya_ has joined #openstack-telemetry | 11:10 | |
*** ddieterly has quit IRC | 11:11 | |
*** _nadya_ has quit IRC | 11:15 | |
*** ljxiash has quit IRC | 11:32 | |
*** ljxiash has joined #openstack-telemetry | 11:32 | |
*** ljxiash has quit IRC | 11:32 | |
*** ljxiash has joined #openstack-telemetry | 11:33 | |
*** ljxiash has quit IRC | 11:37 | |
cdent | I can't figure out why, upstream, here in the future, we are so concerned with testing against ancient stuff and things that are in distro defaults. Especially given the release latency. We should be testing against the future so by the time stuff is out it is aligned with the present... | 11:58 |
*** mattyw has quit IRC | 12:02 | |
*** thorst has joined #openstack-telemetry | 12:04 | |
*** ddieterly has joined #openstack-telemetry | 12:07 | |
*** thorst has quit IRC | 12:11 | |
*** ddieterly has quit IRC | 12:12 | |
*** jwcroppe has joined #openstack-telemetry | 12:19 | |
*** jwcroppe has quit IRC | 12:24 | |
*** belmoreira has quit IRC | 12:42 | |
*** ljxiash has joined #openstack-telemetry | 12:47 | |
*** thorst has joined #openstack-telemetry | 12:47 | |
*** cheneydc has joined #openstack-telemetry | 12:47 | |
*** ddieterly has joined #openstack-telemetry | 13:08 | |
*** links has quit IRC | 13:10 | |
*** _nadya_ has joined #openstack-telemetry | 13:12 | |
*** gordc has joined #openstack-telemetry | 13:12 | |
*** ddieterly has quit IRC | 13:12 | |
*** _nadya_ has quit IRC | 13:16 | |
jd__ | cdent: some of the changes are just 'wrong' and duplicates | 13:17 |
jd__ | other than that, I take any patch that improves or cleans the code | 13:17 |
jd__ | cdent: I guess the problem is that we don't do *both* :) | 13:18 |
jd__ | (re: testing the future) | 13:18 |
cdent | testing against things like LTS versions of distros should be the distros' problem | 13:19 |
*** pradk has joined #openstack-telemetry | 13:27 | |
*** pradk has quit IRC | 13:27 | |
gordc | cdent: any chance you working on nova scheduler? | 13:29 |
cdent | gordc: yes, but it is a very slow process, what's up? | 13:29 |
gordc | you have any docs on it? or know of any related bps? | 13:30 |
gordc | cdent: or can you write a report on it for me :) | 13:30 |
cdent | I sort of already did there's three of these starting at: https://anticdent.org/unpacking-the-nova-scheduler-part-1-of.html | 13:31 |
gordc | preferably the latter. i'll remove you name but cite you as a ref | 13:31 |
gordc | ... this work out better than i expected. | 13:31 |
cdent | I also recently edited: https://github.com/openstack/nova/blob/master/doc/source/filter_scheduler.rst | 13:32 |
cdent | You may owe me some kind of tasty snack now. | 13:33 |
cdent | a pending bp of interest: https://review.openstack.org/#/c/256323/ | 13:34 |
cdent | s/bp/backlog spec/ | 13:34 |
gordc | i'll get you a coffee at timmys. | 13:34 |
*** renatoarmani has joined #openstack-telemetry | 13:34 | |
gordc | tim hortons is going to benefit from openstack | 13:34 |
cdent | quite | 13:35 |
gordc | cdent: i don't know if this is mentioned anywhere in docs but did we start unifying scheduler from nova/cinder/neutron? or is that a few cycles from reality (if ever) | 13:36 |
cdent | That's an ongoing notion that has stalled out several times. | 13:37 |
cdent | The current iteration is that in order for it to happen the way resource tracking is done needs to be adjusted | 13:37 |
cdent | so that it is not entirely based on compute nodes | 13:37 |
cdent | There was a project called Gantt which was going to extract the scheduler but it died on the vine so the new plan is to make the scheduler in nova more bounded so that extraction can happen as more of a lift and shift, rather than make-something-new-over-here | 13:38 |
gordc | cool cool. so at least a year away | 13:40 |
cdent | Seems that way, as hard as that is to believe | 13:40 |
gordc | this is awesome info. i was completely guessing at what you were working on. | 13:41 |
*** ddieterly has joined #openstack-telemetry | 13:41 | |
gordc | need to make a note to bug you on everything | 13:41 |
* cdent needs to learn to keep his mouth shut | 13:42 | |
gordc | too late | 13:42 |
*** mattyw has joined #openstack-telemetry | 13:44 | |
*** shardy has quit IRC | 13:47 | |
*** shardy has joined #openstack-telemetry | 13:49 | |
gordc | jd__: you have thoughts on how to implement that 'unit' bug in gnocchi? | 13:59 |
gordc | should we make it more generic and have some sort of tag support? | 13:59 |
jd__ | gordc: I don't have any particular idea | 14:02 |
jd__ | I think have a unit make sense and can be 'structured' by the API | 14:03 |
jd__ | so I would not see any objection to add an attribute to the metric itself | 14:03 |
gordc | yeah. it's definitely going to be bigger change with generic tags. | 14:04 |
gordc | my concern is that we'll keep adding more and end up at same place as ceilometer model where 'meter' is really overloaded. | 14:05 |
gordc | i'll add some comments | 14:05 |
*** ddieterly has quit IRC | 14:07 | |
*** maestro has quit IRC | 14:19 | |
*** jwcroppe has joined #openstack-telemetry | 14:20 | |
*** maestro has joined #openstack-telemetry | 14:21 | |
*** links has joined #openstack-telemetry | 14:24 | |
*** jwcroppe has quit IRC | 14:25 | |
*** maestro has quit IRC | 14:26 | |
*** _nadya_ has joined #openstack-telemetry | 14:29 | |
*** kbyrne has quit IRC | 14:33 | |
*** kbyrne has joined #openstack-telemetry | 14:40 | |
*** ddieterly has joined #openstack-telemetry | 14:43 | |
*** diogogmt has quit IRC | 14:43 | |
*** links has quit IRC | 14:49 | |
*** jwcroppe has joined #openstack-telemetry | 14:51 | |
*** kbyrne has quit IRC | 14:52 | |
*** kbyrne has joined #openstack-telemetry | 14:53 | |
*** kbyrne has quit IRC | 14:55 | |
jd__ | gordc: we have the type (cumulative, gauge, etc) and the unit that are missing for exploiting the data | 14:55 |
jd__ | over than that, I'm not sure we need more tags stuff | 14:55 |
jd__ | we do need something like that on resources though | 14:55 |
*** rbak has joined #openstack-telemetry | 14:57 | |
*** ddieterl_ has joined #openstack-telemetry | 14:58 | |
*** kbyrne has joined #openstack-telemetry | 15:00 | |
gordc | jd__: yeah, i was debating whether type and unit could exist just as docco... | 15:00 |
gordc | but i guess we need to consider for when machines take over. | 15:01 |
*** kbyrne has quit IRC | 15:01 | |
jd__ | we should | 15:01 |
*** ddieterly has quit IRC | 15:01 | |
gordc | jd__: you mean resource tags so we can stop creating a new table per type? | 15:02 |
*** kbyrne has joined #openstack-telemetry | 15:02 | |
*** renatoarmani has quit IRC | 15:04 | |
*** sbadia has left #openstack-telemetry | 15:07 | |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: collector: Use oslo.messaging batch listener https://review.openstack.org/234831 | 15:13 |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: notification: Use oslo.messaging batch listener https://review.openstack.org/250814 | 15:13 |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: remove useless notification listener helper https://review.openstack.org/250815 | 15:13 |
jd__ | gordc: yeah, though I don't think not having a table per type is a good path | 15:14 |
jd__ | there's a set of trade-off to do, not sure right now what we should do | 15:15 |
*** ddieterl_ has quit IRC | 15:18 | |
*** renatoarmani has joined #openstack-telemetry | 15:18 | |
gordc | why do we have multiple tables? because of the metadata or for 'sharded' tables? | 15:20 |
*** changbl has quit IRC | 15:27 | |
*** cheneydc has quit IRC | 15:27 | |
*** yprokule has quit IRC | 15:31 | |
*** shardy has quit IRC | 15:35 | |
*** yprokule has joined #openstack-telemetry | 15:36 | |
*** ddaskal has joined #openstack-telemetry | 15:37 | |
*** cdent has quit IRC | 15:42 | |
*** shardy has joined #openstack-telemetry | 15:44 | |
*** yarkot has joined #openstack-telemetry | 15:49 | |
*** cdent has joined #openstack-telemetry | 15:53 | |
*** renatoarmani has quit IRC | 16:04 | |
*** cdent has quit IRC | 16:09 | |
*** shakamunyi has joined #openstack-telemetry | 16:14 | |
*** yprokule has quit IRC | 16:14 | |
*** dave-mccowan has joined #openstack-telemetry | 16:15 | |
*** barra204 has quit IRC | 16:15 | |
*** diogogmt has joined #openstack-telemetry | 16:17 | |
openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/gnocchi: Replace deprecated LOG.warn with LOG.warning https://review.openstack.org/263373 | 16:17 |
*** yprokule has joined #openstack-telemetry | 16:19 | |
*** diazjf has joined #openstack-telemetry | 16:21 | |
*** PsionTheory has joined #openstack-telemetry | 16:25 | |
*** mattyw has quit IRC | 16:26 | |
*** pradk has joined #openstack-telemetry | 16:26 | |
*** jwcroppe has quit IRC | 16:31 | |
*** jwcroppe has joined #openstack-telemetry | 16:32 | |
*** cdent has joined #openstack-telemetry | 16:32 | |
*** ddieterly has joined #openstack-telemetry | 16:33 | |
*** yprokule has quit IRC | 16:35 | |
*** idegtiarov_ has quit IRC | 16:39 | |
*** renatoarmani has joined #openstack-telemetry | 16:48 | |
*** mattyw has joined #openstack-telemetry | 16:56 | |
*** diazjf has quit IRC | 16:56 | |
dan-t | Hi, I'm trying to understand how coordination works between the notification agents. Is there a write-up of that somewhere? Thanks | 16:59 |
*** diazjf has joined #openstack-telemetry | 16:59 | |
gordc | dan-t: how it logically works or how to configure it? | 17:12 |
*** safchain has quit IRC | 17:14 | |
dan-t | gordc: I'm looking for logically | 17:20 |
gordc | dan-t: i don't know if this chart is clear but you can look at slide 2: https://docs.google.com/presentation/d/1QgjDOLRnKDboqP8P1LvV0kR5aQEv_VJsDtMlh6u7tIY/edit?usp=sharing | 17:22 |
gordc | dan-t: feel free to ask me questions (i might be delayed, going to grab lunch sometime soon) | 17:23 |
dan-t | sounds good, thanks! | 17:23 |
*** sudorandom has joined #openstack-telemetry | 17:23 | |
*** ddieterly has quit IRC | 17:25 | |
*** yarkot has quit IRC | 17:30 | |
*** shardy has quit IRC | 17:34 | |
*** shardy has joined #openstack-telemetry | 17:36 | |
openstackgerrit | LiuNanke proposed openstack/gnocchi: Test: make enforce_type=True in CONF.set_override https://review.openstack.org/263820 | 17:37 |
*** mattyw has quit IRC | 17:45 | |
cdent | gordc: I could do with that nourishing snack or coffee now | 17:51 |
*** ddieterly has joined #openstack-telemetry | 17:51 | |
gordc | cdent: it'll take a few days to ship. i'm shipping with 'cheapest and slowest damn option' | 17:56 |
gordc | i'm sure timmys can survive the trip | 17:56 |
_nadya_ | I see you're talking about coordination... I'm testing it in various configurations now and cannot process 100 000 samples. agents becomes core dumped :( | 18:00 |
*** ddaskal has quit IRC | 18:00 | |
cdent | _nadya_: 100,000 over what time period? | 18:00 |
_nadya_ | as fast as possible. I'm generating it using our test tool | 18:01 |
gordc | what do you mean by 'core dumped' | 18:01 |
_nadya_ | Segmentation fault (core dumped) | 18:02 |
_nadya_ | ceilometer-anotification failed to start | 18:02 |
_nadya_ | but actually they start, work about a minute and fail | 18:02 |
gordc | how do they work by 'failed to start'? | 18:03 |
gordc | how many agents do you have? and how many queue sets? | 18:03 |
_nadya_ | let me prepare some logs for you | 18:04 |
gordc | kk, drop them in a bug | 18:04 |
_nadya_ | gordc: this is several lines from the end http://paste.openstack.org/show/483055/ | 18:05 |
_nadya_ | gordc: queues looks good http://paste.openstack.org/show/483056/ | 18:06 |
_nadya_ | I run several agents (2,3,4) but after a minute everything fails | 18:06 |
gordc | i'd include # agents, # of queue_sets, oslo.messaging version, and distro | 18:06 |
gordc | what does 2,3,4 mean? you are constantly spawning/killing agents? | 18:07 |
* cdent is pretty sure it's sileht's fault ;) | 18:07 | |
gordc | yeah, well it's definitely his fault. i'm being a good samaritan. | 18:07 |
_nadya_ | gordc: I wanted to run 7 agents initially :) but after several of them stop, I decided to run 3. The latest run was with one but in coordination mode | 18:10 |
_nadya_ | I'm just curious actually, probably my devstack just cannot manage the load. | 18:11 |
gordc | do you have heartbeat_timeout_threshold enabled? | 18:11 |
*** changbl has joined #openstack-telemetry | 18:12 | |
_nadya_ | gordc: I see the following | 18:13 |
_nadya_ | 2016-01-05 17:22:56.761 15119 WARNING oslo.service.loopingcall [-] Function 'ceilometer.coordination.PartitionCoordinator.heartbeat' run outlasted interval by 3.23 sec | 18:13 |
gordc | i'd turn that off https://bugs.launchpad.net/ceilometer/+bug/1478135 | 18:13 |
openstack | Launchpad bug 1478135 in oslo.messaging "notification agent never idles" [Undecided,Confirmed] | 18:13 |
_nadya_ | gordc: I'll check that | 18:13 |
gordc | it defaults to 60 in liberty. i change it to 0 | 18:14 |
gordc | but yeah, i don't think you need 7 agents. i'd probably just try 3 and maybe have 6-9 pipeline queues sets | 18:15 |
gordc | also, if you're dumping 100k messages on queue and then starting agents, you are going to have a bad time. the bigger queue is, the slower the reading (just how mq work) | 18:16 |
gordc | but yeah, i'd open a bug with info regardless (i haven't tested it this cycle) | 18:17 |
_nadya_ | gordc: yep, I'm trying to see how delta transformers work. How precisely it will transform messages | 18:18 |
_nadya_ | gordc: I'm trying to create good pictures for that to show you :) | 18:19 |
_nadya_ | gordc: I was testing it with Redis also, it works rather good. We have almost the same results for transformers as with ipc queues | 18:19 |
gordc | cool cool | 18:21 |
*** yarkot has joined #openstack-telemetry | 18:21 | |
*** _nadya_ has quit IRC | 18:23 | |
*** prashantD has joined #openstack-telemetry | 18:24 | |
*** _nadya_ has joined #openstack-telemetry | 18:30 | |
*** terriyu has joined #openstack-telemetry | 18:40 | |
*** ljxiash has quit IRC | 18:48 | |
*** prashantD has quit IRC | 18:56 | |
*** ddaskal has joined #openstack-telemetry | 19:01 | |
*** _nadya_ has quit IRC | 19:02 | |
*** r-mibu has left #openstack-telemetry | 19:03 | |
*** yarkot has quit IRC | 19:05 | |
*** ddaskal has quit IRC | 19:05 | |
*** eglynn has quit IRC | 19:09 | |
*** yarkot has joined #openstack-telemetry | 19:12 | |
dan-t | gordc: when enabling multiple notification agents, should one set workload_partitioning to true? | 19:17 |
dan-t | by "enabling multiple notification agents" what I mean is setting the number of workers | 19:18 |
*** prashantD has joined #openstack-telemetry | 19:18 | |
dan-t | to something greater than 1 | 19:18 |
gordc | dan-t: you don't necessarily need to. | 19:19 |
gordc | if you have no transformers in your pipeline, you can deploy as many agents as you need without coordinations. | 19:19 |
gordc | but if you do, yes, you should have workload_partitioning enabled to ensure transformations are done correctly. | 19:20 |
*** prashantD has quit IRC | 19:22 | |
dan-t | how is the data partitioned in that case? Based on the code, it looks like possibly by resource_id? | 19:28 |
dan-t | in the case where workload_paritioning is on (sorry) | 19:28 |
*** prashantD has joined #openstack-telemetry | 19:29 | |
gordc | dan-t: that's correct, by resource_id. basically main listener greedily grab from external queues and then sorts them into buckets | 19:32 |
gordc | then one of the queue listeners will process it | 19:33 |
*** ddaskal has joined #openstack-telemetry | 19:45 | |
alextricity25 | Hello! Are there any good blog/tutorials on how to write an external system that ceilometer can publish to? | 19:52 |
gordc | alextricity25: what type of systems? there's this https://developer.ibm.com/opentech/2015/07/06/viewing-keystone-cadf-notifications-with-ceilometer-and-rabbitmq/ | 19:55 |
gordc | although that may be using ceilometer event storage. | 19:56 |
gordc | i know when i was at ibm we use http dispatcher to send data to a separate service rather than use ceilometer storage. | 19:56 |
alextricity25 | I guess now the best thing to do would be to use the UDP publisher, right? I'm looking for something that will help me get started on utilizing that feature | 19:57 |
alextricity25 | So i can write my own system that will do things with that data | 19:58 |
gordc | don't think there's a doc for it. i'd assume main usages i know are people using kafka publisher to push data to some kafka queue and have something else ingest it | 19:58 |
gordc | oh the udp publisher? um, i don't know of anyone using udp and not sending it to collector but that should be possible | 19:59 |
alextricity25 | gordc: What would be the best way to get samples/events from ceilometer and send them to an external system(one that i'm looking to write)? | 20:02 |
gordc | you could republish them to queue (to a new topic if you want to still use ceilometer storage) if you want to guarantee delivery. | 20:05 |
gordc | udp or http might be easiest if you don't have any way to consume from queue | 20:06 |
alextricity25 | Do you know of any good docs out there that might say a little more on using the httpd dispatcher? :) | 20:07 |
alextricity25 | That might be the best option for me. I just don't know how complicated it is. | 20:07 |
gordc | http://www-01.ibm.com/support/knowledgecenter/SS42VS_7.2.4/com.ibm.dsm.doc/t_dsm_guide_openstack_enable_comm.html | 20:09 |
gordc | you probably just need to do steps 8 onwards | 20:09 |
gordc | and ignore the cadf_only part | 20:09 |
gordc | hmmm... actually step 11 onwards (since it's in upstream code) | 20:11 |
openstackgerrit | gordon chung proposed openstack/python-aodhclient: add alarm interface https://review.openstack.org/263887 | 20:12 |
openstackgerrit | Rohit Jaiswal proposed openstack/ceilometer: Enhances get_meters to return unique meters https://review.openstack.org/259626 | 20:14 |
*** yassine__ has quit IRC | 20:15 | |
alextricity25 | gordc: Cool! What if I wanted to write a database dispatcher to write to my application's database directly? Would that be fairly involved? | 20:16 |
gordc | alextricity25: depends, if you ignore the ceilometer api (which you should) then you really just need to inherit the base connection class and implement the record_metering_data aspect of it. | 20:20 |
gordc | https://github.com/openstack/ceilometer/blob/master/ceilometer/storage/base.py#L110 | 20:20 |
gordc | and create a connection to your db in init function... not sure what you are planning specifically, writing your own dispatcher wouldn't be hard either. | 20:22 |
gordc | (might be a bit easier since you'll need to add your new db as an entry_point to get it recognised | 20:23 |
alextricity25 | Thanks gordc! We are looking for a way for ceilometer to send certain samples/events to our application that we are going to write. I'm still unsure whether it will be easier for us to write a dispatcher to make a connection to our database, or use the httpd dispatcher. I'm going to have to do more brain storming | 20:26 |
gordc | sounds good. play around with different publishers and dispatchers and see what works best i guess. | 20:27 |
alextricity25 | gordc: Do you know if the httpd dispatcher can send filtered samples/events? | 20:28 |
gordc | you could filter them from pipeline. not explicitly at http though | 20:29 |
alextricity25 | hmm...maybe that would be configured through the pipeline configs | 20:29 |
gordc | yep :) | 20:29 |
*** terriyu has quit IRC | 20:40 | |
*** yarkot has quit IRC | 20:41 | |
*** kelv has quit IRC | 20:46 | |
dan-t | gordc: When setting workload_partitioning to true for the notification agent, it seems to produce a number of "ceilometer-pipe" queues in rabbitmq. Then, it appears that each worker connects to each of those queues. For example, if I have 3 workers, each of those workers will connect to all of those queues, and each queue will have 3 consumers. Is that expected? | 20:47 |
gordc | dan-t: hm.. no it should have only one each. | 20:48 |
gordc | maybe try setting heartbeat_timeout_threshold=0 in oslo.messaging | 20:48 |
gordc | if that doesn't work. can you create a bug with your code verison, oslo.messaging version and maybe your partitioning settings | 20:50 |
dan-t | Yea, that didn't resolve it. I can create a bug. I think I'm running ceilometer 5.0.1 and oslo.messaging 3.0.0. Is there a recommended version of oslo.messaging to run with liberty ceilometer? | 20:53 |
gordc | not entirely sure. i'm trying to trace which versions are broken | 20:57 |
gordc | i recall one of the 3.x builds being broken | 20:57 |
gordc | i'll try testing this tomorrow. 3.0.0 seems fine based on logs | 20:59 |
*** ljxiash has joined #openstack-telemetry | 21:00 | |
dan-t | well, what I saw was all notification workers appeared to connect to all "ceilometer-pipe" queues, and that was impacting "rate_of_change" transformations -- as each worker appeared to opportunistically grab messages from each of those queues. | 21:01 |
gordc | yeah, that makes sense. it should only have 1 consumer per pipeline queue (occassionally 0 or >1 when new agents are added) | 21:03 |
gordc | if there's 3 on all the queues, something is up | 21:03 |
gordc | it also could be tooz, which handles group membership. it looks like the agents think they are the only agent alive | 21:04 |
*** ljxiash has quit IRC | 21:05 | |
dan-t | ah, kk | 21:05 |
dan-t | I think this is a misconfiguration on my part | 21:20 |
dan-t | as I have no [coordination] section in ceilometer.conf | 21:20 |
gordc | oh. yeah. you need that. | 21:23 |
gordc | strange it didn't warn you. i guess that can be a bug. i'm not sure there's a case for enabling workload_partitioning but not coordination | 21:23 |
openstackgerrit | gordon chung proposed openstack/python-aodhclient: fix test debug https://review.openstack.org/263916 | 21:25 |
openstackgerrit | gordon chung proposed openstack/python-gnocchiclient: fix test debug https://review.openstack.org/263917 | 21:27 |
*** renatoarmani has quit IRC | 21:29 | |
*** yarkot has joined #openstack-telemetry | 21:41 | |
*** harlowja has quit IRC | 22:09 | |
*** harlowja has joined #openstack-telemetry | 22:10 | |
*** boris-42 has joined #openstack-telemetry | 22:20 | |
*** thorst has quit IRC | 22:24 | |
*** thorst has joined #openstack-telemetry | 22:25 | |
*** thorst has quit IRC | 22:30 | |
*** dave-mccowan has quit IRC | 22:37 | |
*** cdent has quit IRC | 22:41 | |
*** thorst has joined #openstack-telemetry | 22:49 | |
*** thorst has quit IRC | 22:53 | |
*** zqfan has quit IRC | 23:01 | |
*** ddaskal has quit IRC | 23:08 | |
*** ddaskal has joined #openstack-telemetry | 23:09 | |
*** diazjf has quit IRC | 23:12 | |
*** ddaskal has quit IRC | 23:14 | |
*** shardy has quit IRC | 23:30 | |
*** gordc has quit IRC | 23:33 | |
*** thorst has joined #openstack-telemetry | 23:42 | |
*** diogogmt has quit IRC | 23:54 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!