*** zhurong has quit IRC | 00:03 | |
*** thorst_ has joined #openstack-telemetry | 00:04 | |
*** thorst_ has quit IRC | 00:11 | |
*** vint_bra has joined #openstack-telemetry | 00:13 | |
*** vint_bra has quit IRC | 00:30 | |
*** vint_bra has joined #openstack-telemetry | 00:31 | |
*** vint_bra has quit IRC | 00:31 | |
*** dave-mccowan has joined #openstack-telemetry | 00:43 | |
*** cheneydc has joined #openstack-telemetry | 00:59 | |
*** thorst_ has joined #openstack-telemetry | 01:09 | |
*** zhurong has joined #openstack-telemetry | 01:12 | |
*** thorst_ has quit IRC | 01:12 | |
*** thorst_ has joined #openstack-telemetry | 01:14 | |
*** fguillot is now known as fguillot_afk | 01:19 | |
*** fguillot_afk has quit IRC | 01:20 | |
*** yassou has quit IRC | 01:22 | |
*** fguillot has joined #openstack-telemetry | 01:22 | |
*** paw has quit IRC | 01:25 | |
*** cheneydc has quit IRC | 01:27 | |
*** fguillot is now known as fguillot_afk | 01:43 | |
*** fguillot_afk is now known as fguillot | 01:43 | |
*** thorst_ has quit IRC | 01:52 | |
*** thorst_ has joined #openstack-telemetry | 01:53 | |
*** thorst_ has quit IRC | 02:02 | |
*** liuhanxi has joined #openstack-telemetry | 02:02 | |
*** fguillot has quit IRC | 02:03 | |
*** adriant has quit IRC | 02:05 | |
*** donghao has joined #openstack-telemetry | 02:07 | |
*** thorst_ has joined #openstack-telemetry | 02:20 | |
*** thorst_ has quit IRC | 02:21 | |
*** adriant has joined #openstack-telemetry | 02:29 | |
*** paw has joined #openstack-telemetry | 02:35 | |
*** thorst_ has joined #openstack-telemetry | 02:36 | |
*** dave-mccowan has quit IRC | 02:38 | |
*** zhangguoqing has joined #openstack-telemetry | 02:43 | |
*** donghao has quit IRC | 03:28 | |
*** donghao has joined #openstack-telemetry | 03:28 | |
*** donghao has quit IRC | 03:34 | |
*** khushbu_ has joined #openstack-telemetry | 03:41 | |
*** thorst_ has quit IRC | 03:50 | |
*** thorst_ has joined #openstack-telemetry | 04:02 | |
*** vishwanathj is now known as vishwanathj_zzz | 04:09 | |
*** thorst_ has quit IRC | 04:20 | |
*** thorst_ has joined #openstack-telemetry | 04:28 | |
*** thorst_ has quit IRC | 04:33 | |
*** thorst_ has joined #openstack-telemetry | 04:52 | |
*** donghao has joined #openstack-telemetry | 04:53 | |
openstackgerrit | Shengping Zhang proposed openstack/gnocchi: This patch is used to delete batch of resources https://review.openstack.org/351043 | 04:55 |
---|---|---|
*** thorst_ has quit IRC | 04:56 | |
*** khushbu_ has quit IRC | 04:57 | |
*** khushbu_ has joined #openstack-telemetry | 04:59 | |
*** thorst_ has joined #openstack-telemetry | 05:03 | |
*** paw_ has quit IRC | 05:05 | |
*** thorst_ has quit IRC | 05:08 | |
*** paw_ has joined #openstack-telemetry | 05:08 | |
*** thorst_ has joined #openstack-telemetry | 05:15 | |
*** thorst_ has quit IRC | 05:20 | |
*** yprokule has joined #openstack-telemetry | 05:32 | |
*** rcernin has joined #openstack-telemetry | 05:33 | |
*** khushbu_ has quit IRC | 05:37 | |
*** thorst_ has joined #openstack-telemetry | 05:37 | |
openstackgerrit | Shengping Zhang proposed openstack/gnocchi: This patch is used to delete batch of resources https://review.openstack.org/351043 | 05:56 |
*** adriant has quit IRC | 06:08 | |
*** pcaruana has joined #openstack-telemetry | 06:32 | |
openstackgerrit | ChangBo Guo(gcb) proposed openstack/panko: Base.Model not define __ne__() built-in function https://review.openstack.org/359293 | 06:34 |
*** tesseract- has joined #openstack-telemetry | 06:57 | |
*** shardy has joined #openstack-telemetry | 07:06 | |
*** donghao has quit IRC | 07:10 | |
*** thorst_ has quit IRC | 07:12 | |
*** shardy is now known as shardy_mtg | 07:16 | |
*** donghao has joined #openstack-telemetry | 07:30 | |
*** amoralej|off is now known as amoralej | 07:39 | |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: Initialize correctly collector https://review.openstack.org/367372 | 07:40 |
*** thorst_ has joined #openstack-telemetry | 07:56 | |
*** thorst_ has quit IRC | 07:59 | |
*** thorst_ has joined #openstack-telemetry | 08:01 | |
*** permalac has joined #openstack-telemetry | 08:03 | |
*** khushbu_ has joined #openstack-telemetry | 08:12 | |
openstackgerrit | fengchaoyang proposed openstack/ceilometer: Change fnmatch.match method to fnmatch.fnmatch https://review.openstack.org/369207 | 08:14 |
*** thorst_ has quit IRC | 08:16 | |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: Don't require gnocchiclient https://review.openstack.org/369895 | 08:21 |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: Use deps extra for optional requirements https://review.openstack.org/369896 | 08:21 |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: tox: Remove useless env https://review.openstack.org/369897 | 08:21 |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: tox: refactor target https://review.openstack.org/369898 | 08:21 |
*** thorst_ has joined #openstack-telemetry | 08:21 | |
*** yassine has joined #openstack-telemetry | 08:22 | |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: tox: refactor targets https://review.openstack.org/369898 | 08:22 |
*** yassine is now known as Guest79543 | 08:22 | |
*** zhurong has quit IRC | 08:27 | |
*** andreykurilin has left #openstack-telemetry | 08:35 | |
*** thorst_ has quit IRC | 08:38 | |
*** thorst_ has joined #openstack-telemetry | 08:39 | |
*** khushbu_ has quit IRC | 08:43 | |
*** thorst_ has quit IRC | 08:44 | |
*** thorst_ has joined #openstack-telemetry | 08:46 | |
*** thorst_ has quit IRC | 08:48 | |
*** thorst_ has joined #openstack-telemetry | 08:54 | |
*** thorst_ has quit IRC | 08:59 | |
*** thorst_ has joined #openstack-telemetry | 09:09 | |
*** thorst__ has joined #openstack-telemetry | 09:12 | |
*** thorst_ has quit IRC | 09:13 | |
*** thorst__ has quit IRC | 09:20 | |
openstackgerrit | Fei Long Wang proposed openstack/aodh: Add pre-signed support for zaqar notifier https://review.openstack.org/335289 | 09:20 |
*** efoley has joined #openstack-telemetry | 09:21 | |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: tox: refactor targets https://review.openstack.org/369898 | 09:21 |
*** efoley_ has joined #openstack-telemetry | 09:22 | |
*** efoley has quit IRC | 09:25 | |
*** cdent has joined #openstack-telemetry | 09:26 | |
*** thorst_ has joined #openstack-telemetry | 09:28 | |
*** efoley_ is now known as efoley | 09:34 | |
*** larainema has quit IRC | 09:34 | |
*** thorst_ has quit IRC | 09:35 | |
*** permalac has quit IRC | 09:38 | |
sileht | jd__, is that normal https://pypi.python.org/pypi/ceilometer ? | 09:38 |
sileht | jd__, Only version 0.2 ?? | 09:38 |
*** donghao has quit IRC | 09:50 | |
openstackgerrit | Ha Van Tu proposed openstack/ceilometer: Refactor Ceilometer alarm API https://review.openstack.org/369985 | 09:58 |
*** thorst_ has joined #openstack-telemetry | 10:02 | |
*** liuhanxi has quit IRC | 10:03 | |
*** efoley has quit IRC | 10:04 | |
jd__ | sileht: yes | 10:08 |
jd__ | sileht: we never released on PyPI :( because nobody does that in OpenStack | 10:08 |
jd__ | sileht: but we could fix that I guess | 10:09 |
sileht | oh, I have though that everything was published on pypi | 10:11 |
*** thorst_ has quit IRC | 10:11 | |
*** efoley has joined #openstack-telemetry | 10:12 | |
*** thorst_ has joined #openstack-telemetry | 10:12 | |
openstackgerrit | Shengping Zhang proposed openstack/gnocchi: This patch is used to delete batch of resources https://review.openstack.org/351043 | 10:13 |
*** donghao has joined #openstack-telemetry | 10:14 | |
*** thorst_ has quit IRC | 10:20 | |
*** liuhanxi has joined #openstack-telemetry | 10:29 | |
*** thorst_ has joined #openstack-telemetry | 10:39 | |
*** khushbu_ has joined #openstack-telemetry | 10:49 | |
*** thorst_ has quit IRC | 10:51 | |
*** shardy_mtg is now known as shardy | 10:51 | |
*** thorst_ has joined #openstack-telemetry | 10:59 | |
openstackgerrit | Merged openstack/ceilometer: Change fnmatch.match method to fnmatch.fnmatch https://review.openstack.org/369207 | 11:00 |
*** larainema has joined #openstack-telemetry | 11:06 | |
*** shardy is now known as shardy_mtg | 11:07 | |
*** thorst_ has quit IRC | 11:11 | |
*** liuhanxi has quit IRC | 11:14 | |
*** liuhanxi has joined #openstack-telemetry | 11:14 | |
*** thorst_ has joined #openstack-telemetry | 11:16 | |
*** amoralej is now known as amoralej|lunch | 11:18 | |
*** khushbu_ has quit IRC | 11:18 | |
*** thorst_ has quit IRC | 11:21 | |
*** openstackgerrit has quit IRC | 11:34 | |
*** openstackgerrit has joined #openstack-telemetry | 11:35 | |
*** efoley_ has joined #openstack-telemetry | 11:39 | |
*** efoley has quit IRC | 11:39 | |
*** yassine has joined #openstack-telemetry | 11:41 | |
*** yassine is now known as Guest8550 | 11:42 | |
*** Guest79543 has quit IRC | 11:45 | |
*** dave-mccowan has joined #openstack-telemetry | 11:47 | |
*** efoley__ has joined #openstack-telemetry | 12:03 | |
*** efoley_ has quit IRC | 12:05 | |
*** Kevin_Zheng has quit IRC | 12:06 | |
*** thorst_ has joined #openstack-telemetry | 12:10 | |
*** efoley__ has quit IRC | 12:32 | |
Matias | hey guys, i'm getting a hard to debug issue here where ceilometer-agent-notification is getting stuck without reconnecting to rabbitmq | 12:34 |
Matias | stracing the processes, it is stuck in the following syscall | 12:34 |
Matias | select(0, NULL, NULL, NULL, {999992151, 178807} | 12:34 |
Matias | look at the huge timeout | 12:34 |
Matias | and the fd set is empty | 12:34 |
Matias | it is effectively a sleep | 12:34 |
Matias | do you know what could cause this? | 12:35 |
*** gordc has joined #openstack-telemetry | 12:35 | |
sileht | Matias, your eventlet version is too old | 12:35 |
sileht | Matias, I'm guessing you use syslog ? | 12:36 |
Matias | it is logging to /var/log/ceilometer/ceilometer-agent-notification.log | 12:36 |
Matias | I don't even have eventlet installed in pip | 12:37 |
Matias | ah, just a sec, doing pip in the wrong terminal | 12:37 |
Matias | eventlet (0.19.0) | 12:37 |
Matias | sileht: it is the latest version | 12:38 |
sileht | Matias, what ceilometer versionb ? | 12:38 |
Matias | just a sec, let me check which commit I built | 12:39 |
*** shardy_mtg is now known as shardy | 12:39 | |
Matias | 9709608bb1bab999658475889637994cccf8bbfc | 12:39 |
Matias | master from august 3 | 12:39 |
sileht | ok we don't use eventlet anymore on master | 12:40 |
Matias | ops, from september 2 | 12:40 |
Matias | august 3 is the date of the commit, the merge is more recent | 12:40 |
Matias | sileht: http://paste.openstack.org/show/576118/ | 12:41 |
Matias | my log showed this when it got stuck | 12:41 |
Matias | it doesn't seem to be a networking issue because I have the central agent running in the same container and it is running ok | 12:42 |
Matias | and one of the rabbit servers is on the same physical host as ceilometer | 12:42 |
sileht | Matias, rabbit closes the connection, you should see why in rabbit lgo | 12:42 |
sileht | log | 12:43 |
Matias | sileht: http://paste.openstack.org/show/576136/ | 12:46 |
Matias | apparently some SSL issue | 12:46 |
sileht | indead | 12:46 |
Matias | but regardless of the reason for the connection being close, it is strange to have a nearly infinite retry timeout | 12:46 |
Matias | being closed* | 12:46 |
openstackgerrit | Shengping Zhang proposed openstack/gnocchi: This patch is used to delete batch of resources https://review.openstack.org/351043 | 12:46 |
sileht | Matias, perhaps not | 12:47 |
sileht | Matias, if you strace on the master PID, this can be normal | 12:47 |
Matias | i'm stracing on the child pids | 12:47 |
sileht | Matias, since children processes connect to rabbit, but not the master process | 12:47 |
sileht | Matias, :) | 12:47 |
Matias | the master process is constantly calling wait4 to monitor the childs | 12:48 |
sileht | select wih all FDs NULL is weird, that can be a bug in py-amqp | 12:49 |
sileht | We don't do select on oslo.messaging or ceilometer | 12:49 |
Matias | so almost surely it is an upstream issue, right? | 12:50 |
sileht | This is my guessing py-amqp doesn't behavior correclty when the ssl issue corrs | 12:50 |
sileht | *occurs* | 12:51 |
Matias | I have seen 1 or 2 references to this being caused by python ssl code not being reentrant or something like this | 12:51 |
Matias | but perhaps the notification agent does not use multiple threads, right? | 12:51 |
sileht | Matias, what your py-amqp version ? | 12:52 |
Matias | sileht: 1.4.9 | 12:53 |
Matias | and kombu is 3.0.35 | 12:54 |
*** amoralej|lunch is now known as amoralej | 12:56 | |
Matias | sileht: amqp is not the latest, but there is an upper constraint here: https://github.com/openstack/requirements/blob/master/upper-constraints.txt#L64 | 12:58 |
*** 17WAAMUGO has joined #openstack-telemetry | 13:00 | |
*** Kevin_Zheng has joined #openstack-telemetry | 13:00 | |
*** pradk has quit IRC | 13:03 | |
*** efoley__ has joined #openstack-telemetry | 13:04 | |
sileht | I missed that one :( | 13:04 |
*** 17WAAMUGO has quit IRC | 13:05 | |
*** zhangguoqing has quit IRC | 13:10 | |
Matias | sileht: the master process only spawns a number of workers, which set up their own connections to the messaging queue, right? i mean, no preliminar setup is done in the master | 13:11 |
sileht | Matias, no master is a dump process manager | 13:11 |
Matias | sileht: https://github.com/openstack/ceilometer/blob/79dff805bbc73760cbcbb76f7f2f1c0a4a0753b4/ceilometer/cmd/agent_notification.py | 13:11 |
Matias | this is the master, right? | 13:11 |
Matias | just calls cotyledon and spawns N processes | 13:11 |
sileht | Matias, yes the class is instanciated inside each child process | 13:12 |
*** efoley__ has quit IRC | 13:12 | |
*** efoley__ has joined #openstack-telemetry | 13:14 | |
*** shardy has quit IRC | 13:23 | |
*** rbak has joined #openstack-telemetry | 13:23 | |
*** fguillot has joined #openstack-telemetry | 13:26 | |
*** vint_bra has joined #openstack-telemetry | 13:42 | |
*** donghao has quit IRC | 13:49 | |
*** donghao has joined #openstack-telemetry | 13:49 | |
jd__ | gordc: please re-review https://review.openstack.org/#/c/368923 :) | 13:50 |
jd__ | gordc: and rofl about your bug "holy shit this is going to fast now" | 13:50 |
*** donghao has quit IRC | 13:51 | |
*** donghao has joined #openstack-telemetry | 13:52 | |
gordc | jd__: well argued. | 13:53 |
jd__ | ha :p | 13:53 |
jd__ | sileht: ^ | 13:53 |
gordc | jd__: yeah, it sucks that i have to redo my graphs now but i guess it's good that's it's faster | 13:54 |
jd__ | I've a team here testing perfs too finally fwiw gordc | 13:54 |
gordc | jd__: basically i think if you have only one granularity or if you are batching very little, that scheduling bug will come up. | 13:54 |
jd__ | I need to have them move to a more recent version of master though | 13:55 |
*** shardy has joined #openstack-telemetry | 13:55 | |
jd__ | gordc: yeah I need to take a deeper look at what you wrote, had no time yet | 13:55 |
sileht | jd__, what bug ? | 13:55 |
jd__ | gordc: though changing cotyledon is likely easy | 13:55 |
jd__ | sileht: sorry I wanted you to check https://review.openstack.org/#/c/368923 | 13:55 |
gordc | jd__: good to hear. it's not my fault if my numbers are a lot different from theirs... i'm not cheating numbers ;) | 13:55 |
jd__ | gordc: so far the numbers I got are not tremendous, but as I said they are testing master from ~1 month ago on file (no v3 compression etc) | 13:56 |
jd__ | and no groupped_series | 13:57 |
gordc | yeah, you probably won't see much on file backend. unless you test larger series. | 13:57 |
gordc | jd__: for me, the offset+write vs read+write stabalizes performance on larger series for me. | 13:58 |
gordc | and the grouped_series stuff improve performance overall by 25% | 13:59 |
jd__ | kewl | 14:02 |
*** pradk has joined #openstack-telemetry | 14:05 | |
*** pradk- has joined #openstack-telemetry | 14:05 | |
*** pradk- has quit IRC | 14:05 | |
*** efoley__ has quit IRC | 14:11 | |
*** paw has quit IRC | 14:21 | |
*** Peter___ has joined #openstack-telemetry | 14:32 | |
*** Peter___ is now known as Shengping | 14:33 | |
openstackgerrit | Merged openstack/ceilometer: Remove sql-expire-samples-only as a CLI option https://review.openstack.org/368923 | 14:35 |
openstackgerrit | Darren Hague proposed openstack/ceilometermiddleware: Background sending to a Kubernetes-based RabbitMQ https://review.openstack.org/366639 | 14:35 |
openstackgerrit | Merged openstack/ceilometer: Remove store_events options https://review.openstack.org/367982 | 14:35 |
*** paw has joined #openstack-telemetry | 14:45 | |
openstackgerrit | Merged openstack/ceilometer: Initialize correctly collector https://review.openstack.org/367372 | 14:47 |
openstackgerrit | Julien Danjou proposed openstack/gnocchi: gendoc: allow DELETE to have body https://review.openstack.org/370179 | 14:47 |
openstackgerrit | Julien Danjou proposed openstack/ceilometer: Use pbr wsgi_scripts feature to build ceilometer-api https://review.openstack.org/370185 | 14:55 |
openstackgerrit | Julien Danjou proposed openstack/ceilometer: gnocchi: enable event_dispatcher in devstack and doc https://review.openstack.org/370188 | 14:58 |
openstackgerrit | Julien Danjou proposed openstack/gnocchi: reno: fix formatting in resource type patch https://review.openstack.org/370190 | 15:02 |
openstackgerrit | Julien Danjou proposed openstack/gnocchi: doc: reverse release note order https://review.openstack.org/370192 | 15:03 |
stevelle | jd__: planning to merge https://review.openstack.org/370185 this cycle? | 15:14 |
jd__ | stevelle: likely | 15:14 |
stevelle | ty, will have to prepare a patch for openstack-ansible then | 15:14 |
jd__ | stevelle: hm it's not using mod_wsgi or the like to deploy? | 15:15 |
stevelle | sadly, not yet. | 15:15 |
jd__ | brrr | 15:15 |
stevelle | not enough time for me to convert every service yet | 15:16 |
*** Kevin_Zheng has quit IRC | 15:26 | |
sileht | jd__, gordc I don't understand was if the 'period' in cotyledon here: https://bugs.launchpad.net/gnocchi/+bug/1623263 | 15:29 |
openstack | Launchpad bug 1623263 in Gnocchi "block size too small for processing period" [Undecided,New] | 15:29 |
sileht | was if/what is | 15:29 |
openstackgerrit | Hanxi Liu proposed openstack/ceilometer: standardize release note page ordering https://review.openstack.org/367747 | 15:32 |
*** rcernin has quit IRC | 15:34 | |
*** pcaruana has quit IRC | 15:40 | |
*** shardy has quit IRC | 15:43 | |
gordc | sileht: hmm... now that i look at it, it's nothing to do with cotyledon | 15:44 |
gordc | i was thinking cotyledon controlled the delay between jobs but it looks like it's not related: https://github.com/openstack/gnocchi/blob/master/gnocchi/cli.py#L153 | 15:45 |
sileht | gordc, is the multiprocess.Manager().Queue() that sucks ? Queue is often empty ? | 15:48 |
*** donghao has quit IRC | 15:48 | |
gordc | sileht: sort of. the logic right now is to grab 16 metrics/worker every scheduling cycle. | 15:48 |
gordc | or actually it grabs 16/worker and then the worker grabs 4 of the 16 each of it's own cycles | 15:50 |
gordc | currently we hardcode the processing worker cycle to 1s https://github.com/openstack/gnocchi/blob/master/gnocchi/cli.py#L272 | 15:50 |
gordc | but when processing small amount of measures, it can handle the 4 metrics in much less than 1 second. | 15:51 |
gordc | so there's a disconnect between scheduling interval and the hardcoded processing interval | 15:51 |
sileht | gordc, the implementation of multiprocess.Manager().Queue() does a ton of object copy (objects are pickle, moved to the Manager process, unpickle, repickle, moved to worker process) | 15:52 |
sileht | gordc, all this copy can make the cpu crazy | 15:52 |
gordc | i don't think we're passing objects. it should just be the id string | 15:53 |
gordc | gnocchi is very cpu intensive though. https://review.openstack.org/#/c/368226/ patch drops it a bit | 15:54 |
gordc | one issue i had was my metricd workers are on same host as ceph osds and it basically max out cpu | 15:55 |
jd__ | not a good deployment plan | 15:58 |
openstackgerrit | Hanxi Liu proposed openstack/aodh: standardize release note page ordering https://review.openstack.org/367748 | 15:58 |
jd__ | gordc: yeah that's what I got too, CPU intensive, maybe some profiling is needed | 15:58 |
jd__ | or changing the default archive policy so that less methods are computed | 15:58 |
jd__ | it seems that nobody wants to configure anything so, let's change the default to be even less minimal | 15:59 |
stevelle | it might be that nobody understands what it means to configure the policies | 16:00 |
gordc | yeah. well the only response i got was that the folks in Australia collect at per minute and per hour... not per day | 16:01 |
gordc | stevelle: i feel it's partly that and partly 50% of deployments just use defaults (see: the openstack i deployed.lol) | 16:02 |
stevelle | some of that too, of course | 16:03 |
stevelle | Matias has tweaked his policies I know, and some of the other folks we talk to in openstack-ansible will but not all. | 16:03 |
gordc | jd__: with the grouped_serie patch, it will use maybe 70% of my system. which is 10 osd, 24metricd | 16:03 |
gordc | jd__: i have one host with 10osd, 24metricd, 32api... that one is pretty heavy.lol | 16:04 |
stevelle | btw, was meaning to ask why cors was removed from the default gnocchi paste | 16:04 |
*** efoley has joined #openstack-telemetry | 16:04 | |
gordc | stevelle: it should be there still? i think we just changed order. sileht would know. | 16:05 |
sileht | stevelle, gordc it have been moved in the code | 16:06 |
stevelle | ah, nm found it | 16:06 |
stevelle | Matias: ^ | 16:06 |
stevelle | Change-Id: Iac4bc42102b02e996159e65df223ae09a2ca10d5 | 16:06 |
jd__ | gordc: depends on many measures you send per second I guess? | 16:06 |
sileht | bbl | 16:06 |
*** rwsu has quit IRC | 16:09 | |
gordc | jd__: yeah, i'm always batching in my tests, either 60measures/metric, 720/metric or more. | 16:10 |
jd__ | gordc: I've implemented a new serialization for boundtimeseries | 16:11 |
gordc | jd__: this is what happens if you change number of granularities: https://docs.google.com/presentation/d/11B56LfrD-VKNJJW9Wm7PAk77fKWG-DHP6WfYfPiOJZk/edit#slide=id.g16ee461f52_0_19 | 16:11 |
jd__ | gordc: I'll send a patch if you want to test for perf | 16:12 |
gordc | i didn't test changing aggregates | 16:12 |
gordc | jd__: sure. i'll give it a try | 16:12 |
jd__ | gordc: I love that spike! | 16:12 |
gordc | lol, i'm pretty sure it's the compression | 16:12 |
jd__ | it's like I wrote an algorithm that works | 16:12 |
jd__ | gordc: nah I just added time.sleep(random) actually every 3600 points | 16:13 |
jd__ | gordc: is this with groupped-optimization? | 16:13 |
gordc | this is with master+grouped_series patch | 16:13 |
gordc | so yeah. | 16:14 |
gordc | there's a sleepin you code? | 16:14 |
jd__ | i'm kidding | 16:17 |
jd__ | dummie | 16:17 |
jd__ | but yeah that just shows that we may just remove some aggregatino methods by default | 16:17 |
jd__ | so people will stop complaining | 16:17 |
jd__ | let's just set "mean" everybody only uses that I guess lol | 16:18 |
gordc | oh. you could probably try it. i am half sleeping when i review sometimes. :P | 16:18 |
jd__ | lol | 16:18 |
gordc | let's use median... just to be unique. | 16:19 |
*** amoralej is now known as amoralej|off | 16:21 | |
jd__ | 95 percentile! | 16:25 |
jd__ | TSDB for ISP only | 16:25 |
gordc | lol. if we do shrink the default policy i think we definitely need to figure out the processing too fast for poll interval issue | 16:27 |
*** cdent has quit IRC | 16:32 | |
*** rcernin has joined #openstack-telemetry | 16:34 | |
*** tesseract- has quit IRC | 16:35 | |
*** Matias has quit IRC | 16:41 | |
*** Shengping has left #openstack-telemetry | 16:44 | |
*** diogogmt has joined #openstack-telemetry | 16:49 | |
*** yprokule has quit IRC | 16:50 | |
*** liuhanxi has quit IRC | 17:10 | |
*** Alexey_Abashkin has joined #openstack-telemetry | 17:38 | |
openstackgerrit | Merged openstack/aodh: standardize release note page ordering https://review.openstack.org/367748 | 17:39 |
openstackgerrit | Merged openstack/gnocchi: reno: fix formatting in resource type patch https://review.openstack.org/370190 | 17:41 |
*** AlexeyAbashkin has quit IRC | 17:41 | |
openstackgerrit | Merged openstack/gnocchi: doc: reverse release note order https://review.openstack.org/370192 | 17:41 |
*** thorst_ has quit IRC | 17:45 | |
*** thorst_ has joined #openstack-telemetry | 17:46 | |
openstackgerrit | Merged openstack/ceilometermiddleware: Bump the oslo.messaging version https://review.openstack.org/365446 | 17:51 |
*** thorst_ has quit IRC | 17:54 | |
*** vishwanathj_zzz is now known as vishwanathj | 18:01 | |
*** Guest8550 has quit IRC | 18:27 | |
*** thorst_ has joined #openstack-telemetry | 18:53 | |
*** thorst_ has quit IRC | 19:00 | |
*** efoley has quit IRC | 19:02 | |
openstackgerrit | Merged openstack/gnocchi: gendoc: allow DELETE to have body https://review.openstack.org/370179 | 19:04 |
*** pcaruana has joined #openstack-telemetry | 19:07 | |
*** rwsu has joined #openstack-telemetry | 19:11 | |
*** r-mibu has quit IRC | 19:16 | |
*** r-mibu has joined #openstack-telemetry | 19:16 | |
*** cdent has joined #openstack-telemetry | 19:25 | |
openstackgerrit | Merged openstack/ceilometer: gnocchi: enable event_dispatcher in devstack and doc https://review.openstack.org/370188 | 19:28 |
*** rwsu has quit IRC | 19:41 | |
*** rwsu has joined #openstack-telemetry | 19:43 | |
*** hoonetorg has quit IRC | 19:53 | |
*** thorst_ has joined #openstack-telemetry | 19:58 | |
*** pcaruana has quit IRC | 20:00 | |
*** thorst_ has quit IRC | 20:04 | |
*** hoonetorg has joined #openstack-telemetry | 20:08 | |
openstackgerrit | Merged openstack/ceilometer: Don't require gnocchiclient https://review.openstack.org/369895 | 20:08 |
*** cdent has quit IRC | 20:16 | |
*** paw has quit IRC | 20:28 | |
openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/aodh: sync tooz requirement https://review.openstack.org/370416 | 20:32 |
*** rcernin has quit IRC | 20:36 | |
*** rcernin has joined #openstack-telemetry | 20:39 | |
*** rcernin has quit IRC | 20:41 | |
*** rcernin has joined #openstack-telemetry | 20:41 | |
*** rcernin has quit IRC | 20:42 | |
*** rcernin has joined #openstack-telemetry | 20:42 | |
gordc | hmmm. seems like threading actually works in gnocchi now. yay! | 20:49 |
stevelle | threading, in my python? | 20:51 |
gordc | yeah. i'm still kind of confused myself... the GIL should stop all parallel fun. | 20:52 |
gordc | i just tried swtiching processes with threads... usually it's slower or nothing changes but it's faster it seems | 20:53 |
openstackgerrit | Merged openstack/ceilometer: Use pbr wsgi_scripts feature to build ceilometer-api https://review.openstack.org/370185 | 20:58 |
*** thorst_ has joined #openstack-telemetry | 21:03 | |
*** thorst_ has quit IRC | 21:09 | |
*** donghao has joined #openstack-telemetry | 21:24 | |
*** fguillot_ has joined #openstack-telemetry | 21:36 | |
*** paw has joined #openstack-telemetry | 21:36 | |
*** dave-mccowan has quit IRC | 21:48 | |
openstackgerrit | Fei Long Wang proposed openstack/aodh: Add pre-signed support for zaqar notifier https://review.openstack.org/335289 | 21:54 |
*** rbak has quit IRC | 22:03 | |
zigo | gordc: Any idea why sphinx is mad at gnocchiclient here? http://logs.openstack.org/24/370224/2/check/gate-deb-python-gnocchiclient-dpkg-debian-jessie/87957e4/console.html | 22:03 |
zigo | Even: http://logs.openstack.org/24/370224/2/check/gate-deb-python-gnocchiclient-dpkg-debian-jessie/87957e4/console.html#_2016-09-14_16_07_20_259304 | 22:04 |
zigo | The logic of doc/source/conf.py in gnocchiclient seems completely wrong for the import of themes... | 22:06 |
*** thorst has joined #openstack-telemetry | 22:08 | |
stevelle | grr, testing https://review.openstack.org/370185 I'm getting a NotInitializedError | 22:09 |
*** fguillot has quit IRC | 22:10 | |
stevelle | specifically when loading the paste file, which is in the place you would expect | 22:11 |
*** thorst has quit IRC | 22:15 | |
gordc | zigo: did this always fail? | 22:16 |
gordc | i know gnocchiclient gate is acting up right now but i'm not entirely sure it's related to gnocchiclient. something with server+client interaction | 22:16 |
zigo | gordc: I'm not sure. I've set the param to use rtd theme, we'll see what's going on. | 22:16 |
zigo | gordc: I need to spend time to upgrade it to latest release though. | 22:17 |
zigo | Currently, I'm just finishing bootstrapping all packages in upstream infra. | 22:17 |
zigo | I begin to see good results ! :) | 22:17 |
zigo | ie: cinder, designate, nova & keystone are building... | 22:17 |
zigo | Just on time for the release, I'd say. | 22:17 |
zigo | :P | 22:17 |
gordc | kk, let us know i guess. i probably have to drop in a few minutes | 22:18 |
zigo | gordc: The good thing is that jd__ and sileht will be able to contribute to the packaging there. | 22:18 |
gordc | yeah. i'll leave them to it. i don't know anything about that stuff. ignorance is bliss | 22:18 |
zigo | Oh, it worked ! :) | 22:18 |
zigo | FANTASTIC ! :) | 22:18 |
zigo | https://review.openstack.org/#/c/370224/ | 22:18 |
gordc | awesome. my work is done ;) | 22:19 |
gordc | heading offline. ttyl. | 22:20 |
*** gordc has quit IRC | 22:20 | |
*** fguillot_ is now known as fguillot_afk | 22:21 | |
*** fguillot_afk has quit IRC | 22:22 | |
stevelle | cloudnull: stuck on https://review.openstack.org/370352 -- not sure why I'm getting a NotInitializedError from oslo conf but I always have trouble making heads or tails of that | 22:22 |
stevelle | any clues welcome | 22:22 |
stevelle | it's generating this in the logs http://paste.openstack.org/show/576564/ | 22:23 |
*** rcernin has quit IRC | 22:43 | |
*** pradk has quit IRC | 22:47 | |
*** diogogmt has quit IRC | 22:47 | |
*** thorst has joined #openstack-telemetry | 23:12 | |
*** thorst has quit IRC | 23:20 | |
*** adriant has joined #openstack-telemetry | 23:28 | |
*** rwsu has quit IRC | 23:32 | |
*** fguillot has joined #openstack-telemetry | 23:49 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!