| *** marcin1234 has joined #openstack-telemetry | 00:01 | |
| marcin1234 | how can I increase Gnocchi batching for "measures" ? | 00:01 |
|---|---|---|
| stevelle | marcin1234: I believe it is the batch_size option in the [collector] section of ceilometer.conf | 00:07 |
| marcin1234 | I will try that | 00:13 |
| *** g3ek has quit IRC | 00:16 | |
| *** diogogmt has quit IRC | 00:17 | |
| *** g3ek has joined #openstack-telemetry | 00:17 | |
| *** julim has joined #openstack-telemetry | 00:26 | |
| marcin1234 | guys any idea why with Gnocchi 2.1.3 I would be getting resource type instance does not exist err in Ceilometer collector log? I checked indeed resource type does not exist in the mysql table only generic | 00:32 |
| *** rwsu has joined #openstack-telemetry | 00:39 | |
| stevelle | marcin1234: you need to run the migrations with --create-legacy-resource-types | 00:40 |
| stevelle | https://github.com/openstack/gnocchi/commit/b1b7c1730bff0875089399dc66469d05d1ce5ba6 | 00:40 |
| stevelle | that's any 2.1+ | 00:41 |
| marcin1234 | ah | 00:46 |
| marcin1234 | thx stevelle | 00:46 |
| *** jwcroppe has quit IRC | 00:48 | |
| marcin1234 | you are talking about running gnocchi-upgrade --create-legacy-resource-types ? | 00:49 |
| marcin1234 | yes you are | 00:51 |
| stevelle | gl, gotta run | 00:52 |
| *** julim has quit IRC | 00:52 | |
| stevelle | looks like you got it | 00:52 |
| *** thorst_ has joined #openstack-telemetry | 00:52 | |
| *** cheneydc has joined #openstack-telemetry | 00:55 | |
| *** jwcroppe has joined #openstack-telemetry | 00:57 | |
| *** thorst_ has quit IRC | 00:57 | |
| *** thorst_ has joined #openstack-telemetry | 00:58 | |
| *** thorst_ has quit IRC | 01:06 | |
| *** rcernin has quit IRC | 01:11 | |
| *** khushbu has joined #openstack-telemetry | 01:14 | |
| *** thorst_ has joined #openstack-telemetry | 01:39 | |
| *** khushbu has quit IRC | 01:40 | |
| *** thorst_ has quit IRC | 01:40 | |
| *** thorst_ has joined #openstack-telemetry | 01:41 | |
| *** thorst_ has quit IRC | 01:50 | |
| *** thorst_ has joined #openstack-telemetry | 02:19 | |
| *** thorst_ has quit IRC | 02:21 | |
| *** thorst_ has joined #openstack-telemetry | 02:22 | |
| *** khushbu has joined #openstack-telemetry | 02:22 | |
| *** thorst_ has quit IRC | 02:30 | |
| *** chlong has quit IRC | 02:37 | |
| *** g3ek has quit IRC | 02:37 | |
| *** g3ek has joined #openstack-telemetry | 02:38 | |
| *** khushbu has quit IRC | 02:45 | |
| *** chlong has joined #openstack-telemetry | 02:49 | |
| *** sheel has joined #openstack-telemetry | 03:07 | |
| *** khushbu has joined #openstack-telemetry | 03:07 | |
| *** jwcroppe has quit IRC | 03:22 | |
| *** thorst_ has joined #openstack-telemetry | 03:28 | |
| openstackgerrit | liusheng proposed openstack/ceilometer: Add a tool to clean the legacy alarm tables https://review.openstack.org/312399 | 03:32 |
| *** thorst_ has quit IRC | 03:35 | |
| *** links has joined #openstack-telemetry | 03:50 | |
| flwang | liusheng: hi | 04:05 |
| flwang | last time you told me there is no project id info in alarm, so as the receiver of the notifier, how can i know which the project the alarm belongs to? | 04:06 |
| *** zqfan has joined #openstack-telemetry | 04:33 | |
| *** thorst_ has joined #openstack-telemetry | 04:33 | |
| *** thorst_ has quit IRC | 04:41 | |
| *** rcernin has joined #openstack-telemetry | 04:47 | |
| *** g3ek has quit IRC | 04:56 | |
| *** g3ek has joined #openstack-telemetry | 05:03 | |
| *** _nadya_ has joined #openstack-telemetry | 05:08 | |
| *** M00nr41n has quit IRC | 05:13 | |
| *** _nadya_ has quit IRC | 05:32 | |
| *** rcernin has quit IRC | 05:37 | |
| *** thorst_ has joined #openstack-telemetry | 05:38 | |
| *** tomoiaga has joined #openstack-telemetry | 05:40 | |
| *** g3ek has quit IRC | 05:45 | |
| *** thorst_ has quit IRC | 05:47 | |
| *** g3ek- has joined #openstack-telemetry | 05:47 | |
| *** g3ek- is now known as g3ek | 05:47 | |
| liusheng | flwang: there is no project id in the alarm notification message, but if you need the alarm's project id, you can show the alarm detals by alarm id. | 06:03 |
| *** M00nr41n has joined #openstack-telemetry | 06:07 | |
| *** rcernin has joined #openstack-telemetry | 06:10 | |
| *** _nadya_ has joined #openstack-telemetry | 06:16 | |
| *** khushbu__ has joined #openstack-telemetry | 06:28 | |
| *** khushbu has quit IRC | 06:28 | |
| *** davidlenwell has quit IRC | 06:30 | |
| *** p01nt3r75 has quit IRC | 06:31 | |
| *** davidlenwell has joined #openstack-telemetry | 06:40 | |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/gnocchi: Add grenade plugin https://review.openstack.org/306329 | 06:42 |
| *** thorst_ has joined #openstack-telemetry | 06:44 | |
| *** pcaruana has joined #openstack-telemetry | 06:46 | |
| *** khushbu has joined #openstack-telemetry | 06:49 | |
| *** thorst_ has quit IRC | 06:51 | |
| *** khushbu__ has quit IRC | 06:52 | |
| *** cheneydc has quit IRC | 06:53 | |
| *** yprokule has joined #openstack-telemetry | 06:57 | |
| *** _nadya_ has quit IRC | 06:58 | |
| *** tesseract- has joined #openstack-telemetry | 06:59 | |
| *** g3ek has quit IRC | 07:14 | |
| *** amoralej|off is now known as amoralej | 07:15 | |
| *** khushbu has quit IRC | 07:18 | |
| *** g3ek has joined #openstack-telemetry | 07:23 | |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/gnocchi: DNM: gate check https://review.openstack.org/334809 | 07:25 |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/gnocchi: Add grenade plugin https://review.openstack.org/306329 | 07:31 |
| *** belmoreira has joined #openstack-telemetry | 07:35 | |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: move out oslo.service https://review.openstack.org/311934 | 07:36 |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/ceilometer: pollsters: Remove eventlet timers https://review.openstack.org/309331 | 07:36 |
| openstackgerrit | ChangBo Guo(gcb) proposed openstack/ceilometer: Config: no need to set default=None https://review.openstack.org/334815 | 07:46 |
| *** thorst_ has joined #openstack-telemetry | 07:48 | |
| *** diogogmt has joined #openstack-telemetry | 07:53 | |
| *** eglynn has joined #openstack-telemetry | 07:55 | |
| *** shardy_afk is now known as shardy | 07:55 | |
| *** thorst_ has quit IRC | 07:56 | |
| *** g3ek has quit IRC | 08:04 | |
| *** khushbu_ has joined #openstack-telemetry | 08:06 | |
| *** khushbu_ has quit IRC | 08:10 | |
| *** g3ek has joined #openstack-telemetry | 08:11 | |
| openstackgerrit | Darren Hague proposed openstack/ceilometermiddleware: Add ability to hand off event sending to a queue processed by a background thread so as not to block swift proxy https://review.openstack.org/334571 | 08:24 |
| *** khushbu has joined #openstack-telemetry | 08:27 | |
| *** _nadya_ has joined #openstack-telemetry | 08:31 | |
| *** _nadya_ has quit IRC | 08:38 | |
| *** khushbu has quit IRC | 08:42 | |
| *** yassine has joined #openstack-telemetry | 08:44 | |
| openstackgerrit | Aayush Rajoria proposed openstack/aodh: Don't manually invoke pip. Use the devstack function instead. https://review.openstack.org/329834 | 08:45 |
| *** g3ek has quit IRC | 08:53 | |
| *** thorst_ has joined #openstack-telemetry | 08:53 | |
| *** g3ek has joined #openstack-telemetry | 08:55 | |
| *** ljxiash has joined #openstack-telemetry | 09:00 | |
| *** liamji has joined #openstack-telemetry | 09:00 | |
| *** thorst_ has quit IRC | 09:01 | |
| *** nicolasbock has joined #openstack-telemetry | 09:08 | |
| *** nicolasbock_ has joined #openstack-telemetry | 09:09 | |
| *** nicolasbock has quit IRC | 09:12 | |
| openstackgerrit | ChangBo Guo(gcb) proposed openstack/ceilometer: test: Fix wrong override value of config option interface https://review.openstack.org/334884 | 09:46 |
| *** ljxiash has quit IRC | 09:57 | |
| *** ljxiash has joined #openstack-telemetry | 09:57 | |
| *** thorst_ has joined #openstack-telemetry | 10:00 | |
| *** ljxiash_ has joined #openstack-telemetry | 10:01 | |
| *** ljxiash has quit IRC | 10:02 | |
| *** ljxiash_ has quit IRC | 10:05 | |
| *** thorst_ has quit IRC | 10:06 | |
| *** _nadya_ has joined #openstack-telemetry | 10:24 | |
| *** p01nt3r75 has joined #openstack-telemetry | 10:27 | |
| *** _nadya__ has joined #openstack-telemetry | 10:33 | |
| *** _nadya_ has quit IRC | 10:33 | |
| *** yassine has quit IRC | 10:57 | |
| *** ljxiash has joined #openstack-telemetry | 11:12 | |
| *** ljxiash has quit IRC | 11:14 | |
| *** ljxiash has joined #openstack-telemetry | 11:14 | |
| *** peristeri has joined #openstack-telemetry | 11:17 | |
| *** g3ek has quit IRC | 11:19 | |
| *** g3ek has joined #openstack-telemetry | 11:25 | |
| *** thorst_ has joined #openstack-telemetry | 11:26 | |
| *** gordc has joined #openstack-telemetry | 11:30 | |
| jd__ | is our gate broken? | 11:41 |
| jd__ | pretty sure I059a12b1b920f703f28aca0e2f352714118dee97 broke our gate | 11:44 |
| jd__ | coolio | 11:44 |
| *** gordc has quit IRC | 11:45 | |
| *** khushbu_ has joined #openstack-telemetry | 11:45 | |
| *** yassine has joined #openstack-telemetry | 11:55 | |
| *** khushbu_ has quit IRC | 11:58 | |
| sileht | jd__, oh I was looking for this thx! | 12:02 |
| *** g3ek has quit IRC | 12:07 | |
| *** amoralej is now known as amoralej|lunch | 12:09 | |
| jd__ | sileht: yeah I don't understand curl enough | 12:11 |
| jd__ | or bash | 12:11 |
| *** gordc has joined #openstack-telemetry | 12:12 | |
| *** g3ek has joined #openstack-telemetry | 12:13 | |
| openstackgerrit | Ildiko Vancsa proposed openstack/aodh: Add install-guide for aodh https://review.openstack.org/330048 | 12:13 |
| *** khushbu has joined #openstack-telemetry | 12:14 | |
| *** diogogmt has quit IRC | 12:21 | |
| *** liamji has quit IRC | 12:22 | |
| *** diogogmt has joined #openstack-telemetry | 12:22 | |
| sileht | jd__, something in stable branches have broken my grenade works, before my vacation, it was failling at stopping metricd after tempest tests, now it fails to start old keystone version due to requirement issues... | 12:26 |
| sileht | jd__, http://logs.openstack.org/29/306329/29/check/gate-gnocchi-grenade-dsvm-functional-file-mysql-1.x/9f14fa7/logs/old/screen-key-a.txt.gz#_2016-06-28_07_56_12_794 | 12:26 |
| jd__ | sileht: I've spent the last months fixing stable/mitaka and stable/liberty branches | 12:26 |
| jd__ | -s | 12:26 |
| jd__ | I spend my time fixing things | 12:26 |
| jd__ | what requires oslo.utils >=3.11? | 12:27 |
| jd__ | probably a upper cap missing from global-req | 12:27 |
| sileht | jd__, Got it, Gnocchi 2.1.X is installed instead of 1.3 ... | 12:30 |
| jd__ | ah… | 12:31 |
| *** diogogmt has quit IRC | 12:34 | |
| openstackgerrit | Ildiko Vancsa proposed openstack/aodh: Add install-guide for aodh https://review.openstack.org/330048 | 12:35 |
| *** liusheng has quit IRC | 12:35 | |
| *** liusheng has joined #openstack-telemetry | 12:36 | |
| openstackgerrit | Julien Danjou proposed openstack/ceilometer: devstack: do not wait using the full URL https://review.openstack.org/334994 | 12:38 |
| *** links has quit IRC | 12:40 | |
| *** amoralej|lunch is now known as amoralej | 12:47 | |
| *** links has joined #openstack-telemetry | 12:53 | |
| *** khushbu has quit IRC | 12:55 | |
| sileht | jd__, hum that looks like a pip bug ... | 12:56 |
| sileht | jd__, : | 12:56 |
| sileht | 2016-06-28 07:48:03.305 | Requirement already satisfied (use --upgrade to upgrade): oslo.utils>=3.3.0 in /usr/local/lib/python2.7/dist-packages (from gnocchi==2.1.1.dev160) | 12:56 |
| sileht | 2016-06-28 07:48:03.323 | Collecting oslo.middleware>=3.11.0 (from gnocchi==2.1.1.dev160) | 12:56 |
| sileht | 2016-06-28 07:48:03.383 | Downloading http://mirror.dfw.rax.openstack.org/pypi/packages/e9/17/e8193e20735f94859774c6af8b444a3ab127af32eb7a9900d522920eb82d/oslo.middleware-3.13.0-py2.py3-none-any.whl (52kB) | 12:56 |
| sileht | jd__, gnocchi requires oslo.utils>=3.3.0 and oslo.middleware-3.13 oàslo.utils>=3.11.0 | 12:56 |
| sileht | jd__, but pip thinks 3.3 is fine for both ... | 12:56 |
| sileht | *3.8.0* | 12:57 |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/gnocchi: Add grenade plugin https://review.openstack.org/306329 | 12:58 |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/gnocchi: metricd: use Cotyledon lib https://review.openstack.org/311744 | 12:58 |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/gnocchi: devstack: Fix requirement typo https://review.openstack.org/335007 | 12:58 |
| sileht | jd__, I just reproduce with my local pip .... | 13:00 |
| sileht | jd__, virtualenv venv; venv/bin/pip install oslo.utils===3.8.0; venv/bin/pip install -e gnocchi[test,file,mysql,keystonmiddleware] | 13:01 |
| sileht | jd__, and oslo.utils is not upgraded ... | 13:01 |
| *** M00nr41n has quit IRC | 13:05 | |
| *** yprokule_ has joined #openstack-telemetry | 13:12 | |
| *** yprokule has quit IRC | 13:13 | |
| *** diogogmt has joined #openstack-telemetry | 13:15 | |
| gordc | what do you think of shrinking the split size from 14400 to 1440? or something smaller than 14400. | 13:27 |
| *** liamji has joined #openstack-telemetry | 13:30 | |
| *** Tamayo has joined #openstack-telemetry | 13:32 | |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/gnocchi: Add grenade plugin https://review.openstack.org/306329 | 13:34 |
| *** rbak has joined #openstack-telemetry | 13:41 | |
| jd__ | sileht: isn't there a upper cap requirements with -r passed to pip? | 13:47 |
| sileht | jd__, nop | 13:47 |
| *** r-mibu has quit IRC | 13:48 | |
| sileht | jd__, you can reproduce with: virtualenv venv; venv/bin/pip install oslo.utils===3.8.0; venv/bin/pip install -e gnocchi; venv/bin/pip freeze | grep -e oslo.utils -e oslo.middleware | 13:48 |
| *** r-mibu has joined #openstack-telemetry | 13:48 | |
| sileht | jd__, oslo.utils is not upgraded even oslo.middleware-3.14 requires oslo.utils>=3.11.0 | 13:49 |
| *** pradk has joined #openstack-telemetry | 13:50 | |
| sileht | jd__, that why some peoples use the constraint file, no upgrade issue, but their always use obsolete libs | 13:51 |
| jd__ | sileht: really looks like a pip bug… | 13:53 |
| *** KrishR has joined #openstack-telemetry | 13:54 | |
| jd__ | sileht: if you pass -U to pip that seems to solve it | 13:54 |
| jd__ | but that might just be by luck | 13:54 |
| sileht | jd__, yeah I have used that as workaround | 13:55 |
| *** ametts has joined #openstack-telemetry | 13:57 | |
| sileht | jd__, I wonder why pypi doesn't provide a file with all versions availables and all requirements like any other packages systems | 13:59 |
| jd__ | sileht: or let's just use dpkg? :P | 14:00 |
| sileht | jd__, ahaha | 14:00 |
| openstackgerrit | Julien Danjou proposed openstack/ceilometer: devstack: do not wait for service to start https://review.openstack.org/334994 | 14:17 |
| *** M00nr41n has joined #openstack-telemetry | 14:27 | |
| *** g3ek has quit IRC | 14:28 | |
| *** g3ek has joined #openstack-telemetry | 14:31 | |
| *** diogogmt has quit IRC | 14:32 | |
| *** diogogmt has joined #openstack-telemetry | 14:32 | |
| *** tomoiaga has quit IRC | 14:36 | |
| *** links has quit IRC | 14:38 | |
| *** john5223 has quit IRC | 14:43 | |
| *** john5223 has joined #openstack-telemetry | 14:43 | |
| openstackgerrit | Julien Danjou proposed openstack/ceilometer: collector: use an intermediate proxy class for event dispatcher https://review.openstack.org/317431 | 14:46 |
| *** yprokule1 has joined #openstack-telemetry | 14:46 | |
| *** krotscheck_dcm is now known as krotscheck | 14:48 | |
| *** yprokule_ has quit IRC | 14:48 | |
| *** thumpba has joined #openstack-telemetry | 14:49 | |
| jd__ | sileht: don't forget to add panko to your list of projects to review ;) | 14:52 |
| jd__ | gordc: too | 14:52 |
| sileht | yeah I need to update my dashboard | 14:53 |
| gordc | we need more devs/reviewers :( | 14:53 |
| jd__ | I know gordc | 14:54 |
| gordc | jd__: sileht: i'm playing around with decrease how many points we store per split... it seems performance drops quite a bit when the obj gets larger. | 14:54 |
| gordc | based on the 14400 value righ tnow | 14:55 |
| jd__ | 14400 too big? | 14:55 |
| jd__ | even with lz4? we need delta compression again | 14:55 |
| gordc | yeah, let me show you quick graph on normal 1 worker test | 14:57 |
| gordc | just need to find it | 14:57 |
| *** Tamayo has quit IRC | 14:57 | |
| sileht | jd__, we need to create a box first | 14:57 |
| gordc | i've been doing maths on delta compression or something similar... it's quite a bit slower to compute delta but not sure if it's better or worse tradeoff. | 14:58 |
| sileht | jd__, something like this: http://cdn1.tnwcdn.com/wp-content/blogs.dir/1/files/2016/05/pied-piper-box.png | 15:00 |
| sileht | jd__, it's perfect to store a delta comrpession algorythm | 15:00 |
| jd__ | rofl | 15:00 |
| gordc | lol | 15:00 |
| jd__ | gordc: I think it's worth the data storage gain | 15:04 |
| gordc | jd__: https://docs.google.com/presentation/d/1eHZDcgN9QJk8uEo4V-tGhthpq7aOp_m1kncBFWLwwQI/edit?usp=sharing slide22 | 15:07 |
| jd__ | https://review.openstack.org/#/c/334994/ this should fix the Telemetry gate | 15:07 |
| gordc | it's actually worse in real life (the degradation). | 15:07 |
| jd__ | gordc: rewrite slower than write in Ceph? lol? | 15:08 |
| jd__ | gordc: but that has no relation with delta compression right? | 15:08 |
| *** M00nr41n has quit IRC | 15:08 | |
| gordc | most of the other stuff is just how i (tried to) stabilise ceph | 15:08 |
| gordc | jd__: rewrite means reading existing object, and then writing full (what happens now) | 15:09 |
| jd__ | ha | 15:09 |
| jd__ | 1 second for 100 points which should be 2 KB with no compression? | 15:10 |
| jd__ | or is it 100*100 points? | 15:10 |
| *** M00nr41n has joined #openstack-telemetry | 15:10 | |
| gordc | whatever number in axes is a group of 100 | 15:10 |
| jd__ | hm | 15:11 |
| gordc | so right now the compression with lz4 really only puts us down to 10B/ point | 15:11 |
| jd__ | ok so that's 1 second for 10K points so 200 Kb with no compression | 15:11 |
| gordc | something like that | 15:12 |
| jd__ | delta + lz4 goes down 9 B on average | 15:12 |
| jd__ | down to 1 B on best case scenario IIRC | 15:12 |
| gordc | that test is a bit dumber, the test is realy just reading obj, adding (timestamp, value) tuple, and writing | 15:12 |
| jd__ | 0.05 byte on best case sorry | 15:13 |
| jd__ | 9 bytes is worst case, not average | 15:13 |
| gordc | that's the delta case right? | 15:13 |
| jd__ | delta+lz4 | 15:13 |
| jd__ | that's why I think it's really worth it | 15:14 |
| gordc | right. currently the format is {values: {time:value, time:value, etc...}} | 15:14 |
| *** M00nr41n has quit IRC | 15:14 | |
| jd__ | even if we need to micro-optimize it later to be faster (think: C) | 15:14 |
| gordc | that's about 10B with lz4, i think that graph has it closer to 18B | 15:15 |
| jd__ | yes, no compression is 16 B / point + 20% overhead | 15:15 |
| jd__ | so 19 B vs 0.05 B | 15:15 |
| jd__ | :-D | 15:15 |
| gordc | if i run it in real life with with current master, you'll notice it start to drop pretty hard with 20k metrics... ie starts around 1-2seconds/metric and is around 8s/metric near the larger end. | 15:17 |
| gordc | although i'm not entirely sure where the split is happening... | 15:17 |
| *** g3ek has quit IRC | 15:18 | |
| gordc | 0.05B is if you have all zeros i'm guessing | 15:18 |
| jd__ | gordc: likely | 15:19 |
| jd__ | but it goes down very low if you have discret value that are very close | 15:19 |
| jd__ | worst case is totally random :) | 15:19 |
| gordc | i was thinking we don't need to even store delta, just the value and let the position in file dictate the time. | 15:20 |
| *** g3ek has joined #openstack-telemetry | 15:20 | |
| gordc | it's roughly 5bytes float+delimiter. | 15:20 |
| *** M00nr41n has joined #openstack-telemetry | 15:20 | |
| jd__ | so just put 0 to fill the file? | 15:21 |
| gordc | yeah. | 15:21 |
| jd__ | unless you want to use sparse files, but that would not be possible for most backends I think | 15:21 |
| gordc | right. it may be worse for swift when obj size is small. | 15:22 |
| jd__ | this is what the delta compression does gordc, it only applies to timestamps | 15:22 |
| gordc | we'd probably need to move serialize methods into each of the backends. | 15:22 |
| jd__ | the values are compressed only by lz4 | 15:22 |
| gordc | right. the delta compression does that but doesn't allow us to append. the padded format allows append but is probably more inefficient in serialising. | 15:24 |
| *** M00nr41n has quit IRC | 15:24 | |
| gordc | i think the biggest two things right now are number of iops needed and degradation based on object size. | 15:25 |
| jd__ | gordc: so open question | 15:26 |
| jd__ | because I know you are going to work on that optimization etc | 15:26 |
| jd__ | and I'm glad you do | 15:27 |
| jd__ | I'm not sure that having backported https://review.openstack.org/#/c/333367/ is going to help you in the future and is really clean | 15:27 |
| jd__ | WDYT? | 15:27 |
| jd__ | we're in the middle of things right now, and I would really like to clear things out | 15:28 |
| gordc | was thinking same thing :) was debating whether to revert the compression and let us decide what we want to do next. | 15:28 |
| gordc | i don't really have a strong idea which path to go right now. i'm open to options. | 15:28 |
| *** M00nr41n has joined #openstack-telemetry | 15:29 | |
| gordc | i'm currently just etsting how long it would take to serialise/deserialise into float+delimiter | 15:29 |
| jd__ | i'm open to letting you find the best option | 15:29 |
| jd__ | let's use that failure as a good opportunity to do better? | 15:30 |
| gordc | agreed | 15:30 |
| jd__ | so I can revert the patch in stable/2.1 and we can experiment | 15:30 |
| gordc | sounds good to me. | 15:30 |
| gordc | hold off on a release off master as well then? to avoid releasing the compression there. | 15:31 |
| *** belmoreira has quit IRC | 15:31 | |
| jd__ | gordc: I'm gonna revert there too | 15:31 |
| gordc | i'm guessing you folks don't have time / opporunity to look at this? | 15:32 |
| jd__ | so we can have 2.2 | 15:32 |
| gordc | jd__: kk | 15:32 |
| jd__ | and then we'll do 3.0 with your new fancy stuff whatever that is | 15:32 |
| openstackgerrit | Julien Danjou proposed openstack/gnocchi: Revert "carbonara: compress all TimeSerie classes using LZ4" https://review.openstack.org/335092 | 15:33 |
| jd__ | gordc: ^^^ | 15:33 |
| jd__ | gordc: now you have the future in your hans ^^ | 15:33 |
| jd__ | hands | 15:33 |
| jd__ | and I keep fixing the gates lol | 15:33 |
| gordc | you're making a huge mistake ;) | 15:34 |
| jd__ | that won't be the last | 15:37 |
| *** _nadya__ has quit IRC | 15:39 | |
| *** jmccarthy has joined #openstack-telemetry | 15:44 | |
| *** jmccarthy has left #openstack-telemetry | 15:46 | |
| jd__ | gordc: FYI https://bugs.launchpad.net/ceilometer/+bug/1596988 | 15:47 |
| openstack | Launchpad bug 1596988 in Ceilometer "ElasticSearch functional tests fail" [Critical,Triaged] | 15:47 |
| jd__ | not sure you still care though | 15:47 |
| openstackgerrit | Julien Danjou proposed openstack/panko: devstack: do not wait for service https://review.openstack.org/335107 | 15:55 |
| openstackgerrit | Julien Danjou proposed openstack/panko: tox: rename elastic search targets https://review.openstack.org/335108 | 15:55 |
| *** tesseract- has quit IRC | 15:56 | |
| openstackgerrit | Mehdi Abaakouk (sileht) proposed openstack/gnocchi: Add grenade plugin https://review.openstack.org/306329 | 15:58 |
| *** vishwanathj has joined #openstack-telemetry | 16:02 | |
| *** g3ek has quit IRC | 16:04 | |
| *** ljxiash has quit IRC | 16:05 | |
| *** g3ek has joined #openstack-telemetry | 16:07 | |
| *** yassine has quit IRC | 16:27 | |
| openstackgerrit | Merged openstack/ceilometer: devstack: do not wait for service to start https://review.openstack.org/334994 | 16:31 |
| *** khushbu has joined #openstack-telemetry | 16:32 | |
| *** zqfan has quit IRC | 16:33 | |
| *** p01nt3r75 has quit IRC | 16:34 | |
| *** _nadya_ has joined #openstack-telemetry | 16:40 | |
| *** permalac has joined #openstack-telemetry | 16:43 | |
| *** _nadya_ has quit IRC | 16:45 | |
| *** permalac_ has quit IRC | 16:45 | |
| gordc | jd__: not really. :( i'm hoping someone else takes care of it... preferably someone that uses it. we use events but not elasticserach | 16:46 |
| *** pcaruana has quit IRC | 16:49 | |
| *** rcernin has quit IRC | 16:51 | |
| *** g3ek has quit IRC | 16:52 | |
| *** g3ek has joined #openstack-telemetry | 16:55 | |
| *** vishwanathj has quit IRC | 17:02 | |
| *** yprokule1 has quit IRC | 17:12 | |
| *** hoonetorg has quit IRC | 17:19 | |
| *** julim has joined #openstack-telemetry | 17:30 | |
| jd__ | gordc: ok, then we'll see | 17:30 |
| *** flwang1 has joined #openstack-telemetry | 17:35 | |
| flwang1 | gordc: ping re aodh notifier | 17:35 |
| *** rcernin has joined #openstack-telemetry | 17:45 | |
| *** rbak_ has joined #openstack-telemetry | 17:57 | |
| *** diogogmt has quit IRC | 17:58 | |
| *** rbak has quit IRC | 17:59 | |
| *** diogogmt has joined #openstack-telemetry | 18:00 | |
| *** liamji has quit IRC | 18:02 | |
| gordc | flwang1: whatsup? | 18:06 |
| flwang1 | gordc: i'm working on an enhancement for the aodh zaqar driver | 18:07 |
| flwang1 | so i'm trying to figure out how to setup the dev env :) | 18:08 |
| flwang1 | so where is the config option to set aodh's notifier as zaqar? in ceilometer's pipeline.yaml? | 18:08 |
| *** _nadya_ has joined #openstack-telemetry | 18:08 | |
| gordc | flwang1: it would be an alarm_action... i don't know exactly how the zaqar one works but for normal webhook you specify http(s)://url to trigger when an alarm happens | 18:10 |
| flwang1 | gordc: ah, got it | 18:10 |
| gordc | i imagine it'd be zaqar://something | 18:11 |
| gordc | pradk: ^ | 18:11 |
| gordc | since you implemented it :) | 18:11 |
| pradk | gordc, looking | 18:12 |
| pradk | flwang1, yea you just specific zaqar:// with subscriber and topics | 18:13 |
| pradk | flwang1, something like: | 18:13 |
| pradk | 'zaqar://?topic=critical&subscriber=mailto:foo@example.com&ttl=7200' | 18:14 |
| flwang1 | pradk: cool, i'm trying to add a support for zaqar's pre-signed queue | 18:14 |
| pradk | cool | 18:14 |
| flwang1 | so for that way, aodh can send notification to tenant user's queue instead of the queue created by aodh's service user | 18:14 |
| flwang1 | with that way, user know the queue name | 18:14 |
| flwang1 | and the benefit is user can create subscription in zaqar as well | 18:15 |
| flwang1 | if the subscription need some complicated data | 18:15 |
| pradk | flwang1, cool, yea i remember we discussed this while back | 18:15 |
| *** diogogmt has quit IRC | 18:16 | |
| flwang1 | pradk: yep, the idea is when user create alarm, he may need input something like this 'zaqar://?topic=critical&subscriber=mailto:foo@example.com&ttl=7200&queue_name=myqueue&project_id=xxx&signature=xxx&paths=xxx&expires=xxx&methods=xxx' | 18:17 |
| flwang1 | and with those addition info(signature, expires...) aodh can create a zaqar client with a different auth way(signed-url), and then with that client, user can post messages to the queue user inputed | 18:19 |
| flwang1 | does that make sense? | 18:19 |
| pradk | flwang1, understood, i'm not sure if we should limit by queue .. but yea auth part makes sense | 18:20 |
| pradk | flwang1, so this would be optional i assume? as in default would still work as usual | 18:21 |
| pradk | for backward compatibility | 18:22 |
| *** liusheng has quit IRC | 18:24 | |
| *** liusheng has joined #openstack-telemetry | 18:24 | |
| *** pcaruana has joined #openstack-telemetry | 18:25 | |
| *** diogogmt has joined #openstack-telemetry | 18:28 | |
| flwang1 | pradk: yes, for sure | 18:29 |
| flwang1 | it's optional | 18:29 |
| flwang1 | the default way still work | 18:29 |
| *** PsionTheory has joined #openstack-telemetry | 18:29 | |
| flwang1 | the idea is | 18:30 |
| flwang1 | we would like to support it to trigger mistral | 18:30 |
| flwang1 | recently we added a mistral notification driver and which need a complicated options for subscription | 18:30 |
| flwang1 | which aodh can't do that for now | 18:30 |
| flwang1 | so we're trying to provide another way so that user can define subscriptions in zaqar but still get alarms from aodh | 18:31 |
| flwang1 | btw, are you guys aware of any production deployment of aodh for auto scaling? | 18:32 |
| flwang1 | what's the general purpose of aodh deployment for now? | 18:32 |
| pradk | i dont have data around prod deployments for aodh .. may gordc has some info | 18:34 |
| flwang1 | pradk: ok, no problem | 18:36 |
| flwang1 | pradk: btw, what's the simplest metric i should try to test the notifier? | 18:36 |
| flwang1 | i mean how to set an alarm to make the test easy? | 18:36 |
| gordc | we use aodh internally (well ceilometer alarms since it's based on older code) i don't know exact use cases | 18:37 |
| *** PsionTheory has quit IRC | 18:39 | |
| flwang1 | gordc: got, thanks. | 18:40 |
| flwang1 | i'm trying to propose a breakout session about aodh+zaqar+mistral+heat for autoscaling | 18:40 |
| flwang1 | or just aodh+ zaqar + heat or senlin | 18:40 |
| gordc | sounds cool. although you can already go straight from aodh+heat. never used any of the other projects tbh | 18:41 |
| stevelle | flwang1: you have my interest, I want to get good autoscaling | 18:41 |
| *** jwcroppe has joined #openstack-telemetry | 18:42 | |
| *** davidlenwell has quit IRC | 18:42 | |
| *** amoralej is now known as amoralej|lunch | 18:42 | |
| *** amoralej|lunch is now known as amoralej|off | 18:43 | |
| *** khushbu has quit IRC | 18:43 | |
| stevelle | flwang1: do you have any user stories / use cases / write up of what autoscaling capability you want to deliver? | 18:46 |
| *** davidlenwell has joined #openstack-telemetry | 18:52 | |
| flwang1 | gordc: yep, i know. aodh+ heat works | 18:55 |
| flwang1 | but seems not too much people use it | 18:55 |
| *** sheel has quit IRC | 18:55 | |
| flwang1 | and after google, you won't see too much info about openstack autoscaling | 18:56 |
| flwang1 | that's a pity | 18:56 |
| flwang1 | stevelle: not yet, i'm working on that | 18:57 |
| flwang1 | stevelle: we can talk more about that | 18:57 |
| flwang1 | stevelle: the quick way is joining my proposal so that we can work out a solution together | 18:57 |
| flwang1 | i'm sure there should be a queue between aodh and heat or senlin | 18:58 |
| flwang1 | especially given autoscaling won't be heat's interest | 18:58 |
| flwang1 | stevelle: the basic idea is i want to leverage the workflow of mistral | 19:02 |
| stevelle | flwang1: that scares me a bit if the use case is to create a need for mistral, or appears that way | 19:08 |
| flwang1 | not 'create a need' :) | 19:09 |
| flwang1 | like i mentioned above, we can even don't use mistral at here | 19:09 |
| flwang1 | but i still believe a queue between aodh and heat(senlin) makes sense | 19:10 |
| flwang1 | for example, pls correct me if i'm wrong | 19:10 |
| flwang1 | now can aodh trigger many different targets with one alarm? | 19:11 |
| stevelle | I don't know an answer for that right now, since the compound/composite alarms changes :) | 19:11 |
| flwang1 | for example, when there is an alarm like cpu usage >90%, i want to notify admin by email and trigger a heat stack update and a mistral workflow | 19:12 |
| flwang1 | or something like that | 19:12 |
| flwang1 | that's my init thought | 19:13 |
| flwang1 | stevelle: welcome to debate it to make it clear :) | 19:13 |
| flwang1 | stevelle: what do you mean 'compound/composite alarms changes'? | 19:14 |
| stevelle | flwang1: I have time tonight to look at this and think about it more but I believe that is a good use case by itself. It would be valuable to build up a solution and maybe a summit talk for this. | 19:16 |
| flwang1 | stevelle: awesome, if you are interested in, you can ping me at zaqar channel and we can talk more | 19:17 |
| flwang1 | now i'm doing something test/demo based on this idea | 19:17 |
| flwang1 | and again, the basic idea is using a message broker between those services/components to get a better auto scaling | 19:18 |
| stevelle | flwang1: at the root, multiple actions on alarm trigger is a valuable use case | 19:18 |
| flwang1 | stevelle: yes for sure | 19:18 |
| flwang1 | you know, some times, the auto scaling is not a simple pattern like if A do B | 19:19 |
| flwang1 | it's most like if A then Check B or check C then do D or do E or ..... | 19:19 |
| flwang1 | heat is not good at that | 19:20 |
| flwang1 | and we(zaqar) is planning to add the retry for notification | 19:21 |
| flwang1 | which is another gap for aodh+heat way | 19:21 |
| flwang1 | so those are some very rough ideas | 19:21 |
| flwang1 | if we can write them down and figure out some detailed user cases, it would be great | 19:22 |
| flwang1 | i always believe auto scaling is one of critical user case of cloud | 19:22 |
| flwang1 | unfortunately, we didn't see enough effort in this domain | 19:23 |
| flwang1 | my 0.02 | 19:23 |
| openstackgerrit | Merged openstack/gnocchi: Revert "carbonara: compress all TimeSerie classes using LZ4" https://review.openstack.org/335092 | 19:28 |
| *** ametts has quit IRC | 19:40 | |
| *** flwang1 has quit IRC | 19:49 | |
| *** hoonetorg has joined #openstack-telemetry | 19:54 | |
| *** ametts has joined #openstack-telemetry | 19:56 | |
| *** peristeri has quit IRC | 19:57 | |
| *** g3ek has quit IRC | 20:02 | |
| *** g3ek has joined #openstack-telemetry | 20:03 | |
| *** _nadya_ has quit IRC | 20:43 | |
| *** thumpba has quit IRC | 20:45 | |
| *** g3ek has quit IRC | 20:48 | |
| *** ametts has quit IRC | 20:49 | |
| *** g3ek has joined #openstack-telemetry | 20:50 | |
| *** harlowja has quit IRC | 21:19 | |
| *** julian1_ has joined #openstack-telemetry | 21:28 | |
| *** g3ek has quit IRC | 21:36 | |
| *** thorst_ has quit IRC | 21:37 | |
| *** julian1_ has quit IRC | 21:37 | |
| *** g3ek has joined #openstack-telemetry | 21:37 | |
| *** julian1_ has joined #openstack-telemetry | 21:37 | |
| *** thorst_ has joined #openstack-telemetry | 21:37 | |
| *** thorst__ has joined #openstack-telemetry | 21:40 | |
| *** thorst_ has quit IRC | 21:42 | |
| *** thorst__ has quit IRC | 21:44 | |
| *** flwang1 has joined #openstack-telemetry | 21:59 | |
| *** thorst_ has joined #openstack-telemetry | 22:00 | |
| *** davidlenwell has quit IRC | 22:02 | |
| *** rbak_ has quit IRC | 22:04 | |
| *** davidlenwell has joined #openstack-telemetry | 22:11 | |
| *** harlowja has joined #openstack-telemetry | 22:16 | |
| *** harlowja has quit IRC | 22:23 | |
| *** g3ek has quit IRC | 22:24 | |
| *** g3ek has joined #openstack-telemetry | 22:29 | |
| *** gordc has quit IRC | 22:38 | |
| *** KrishR has quit IRC | 22:47 | |
| *** pradk has quit IRC | 22:53 | |
| openstackgerrit | gordon chung proposed openstack/gnocchi: simplify model loading https://review.openstack.org/335248 | 22:58 |
| marcin1234 | anyone familiar with WARNING requests.packages.urllib3.connectionpool [-] Connection pool is full, discarding connection in metricd logs? | 23:00 |
| openstackgerrit | gordon chung proposed openstack/gnocchi: simplify model loading https://review.openstack.org/335248 | 23:00 |
| *** rcernin has quit IRC | 23:03 | |
| *** thorst_ has quit IRC | 23:06 | |
| *** g3ek has quit IRC | 23:06 | |
| *** thorst has joined #openstack-telemetry | 23:07 | |
| *** g3ek has joined #openstack-telemetry | 23:11 | |
| *** harlowja has joined #openstack-telemetry | 23:12 | |
| *** thorst has quit IRC | 23:15 | |
| *** khushbu_ has joined #openstack-telemetry | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
