| *** spsurya has joined #openstack-freezer | 01:01 | |
| openstackgerrit | Merged openstack/freezer-api master: Add python3.7 unit test job in tox.ini https://review.opendev.org/663264 | 01:32 |
|---|---|---|
| *** goldyfruit has joined #openstack-freezer | 01:50 | |
| *** whoami-rajat has joined #openstack-freezer | 02:07 | |
| *** ivve has joined #openstack-freezer | 10:27 | |
| goldyfruit | ivve, all good ? | 12:08 |
| ivve | goldyfruit: yeah thanks, got your message last time | 13:14 |
| ivve | just havn't had time to look at it yet | 13:14 |
| ivve | reno that is | 13:14 |
| goldyfruit | no problem | 13:14 |
| ivve | looking a lot on monasca and how things are supposed to work | 13:15 |
| goldyfruit | https://docs.openstack.org/reno/latest/ | 13:15 |
| ivve | which is the last service i will look into for a while, need a break :P | 13:15 |
| goldyfruit | Yeah Monasca is a big piece | 13:16 |
| ivve | not more complex than magnum i hope | 13:16 |
| ivve | that was a big piece | 13:16 |
| goldyfruit | Magnum is pretty simple | 13:16 |
| ivve | well yea i guess the end results are different | 13:17 |
| ivve | end result from magnum is very complex while monasca end result is very simple | 13:17 |
| ivve | like.. metrics? :) | 13:17 |
| ivve | logs | 13:17 |
| goldyfruit | yeah | 13:17 |
| ivve | on the other hand a working k8s cluster | 13:17 |
| ivve | is a beast in itself | 13:17 |
| ivve | well with octavia, cinder and keystone implemented of course | 13:18 |
| goldyfruit | But magnum doesn't really have to deal with the cluster after the initial deployment | 13:18 |
| goldyfruit | scale up/down | 13:18 |
| ivve | scalig | 13:18 |
| ivve | scaling* | 13:18 |
| ivve | and healthchecks i guess | 13:18 |
| goldyfruit | We are using it since 1 year, no big issue | 13:19 |
| ivve | just trying to figure out how to use monasca from a user perspective | 13:19 |
| goldyfruit | Stein cycle helped a lot for the time spawning | 13:19 |
| ivve | it did | 13:19 |
| ivve | deployed with kolla? | 13:19 |
| goldyfruit | yeah | 13:19 |
| ivve | i had an idea | 13:19 |
| ivve | which i thought was great, and that is to have all fragments in a mount | 13:19 |
| goldyfruit | We did some investigation on Monasca | 13:20 |
| goldyfruit | deploy it | 13:20 |
| ivve | with the possibility to edit them and try new stuff easier | 13:20 |
| goldyfruit | but to complex to what it gave to us | 13:20 |
| ivve | really | 13:20 |
| ivve | what is your take on promethus? | 13:21 |
| goldyfruit | Magnum deploys it with Kubernetes | 13:21 |
| goldyfruit | And we are using it to monitor our platforms | 13:21 |
| ivve | yea i mean like for openstack | 13:21 |
| ivve | and if possible to gather metrics on instances | 13:21 |
| ivve | ceilometer and gnocchi is just very slow | 13:21 |
| goldyfruit | ivve, we are using ceilometer gnocchi | 13:22 |
| ivve | anything using influx or promethus is fast enough | 13:22 |
| ivve | i guess you could backend gnocchi to influx | 13:22 |
| goldyfruit | Nop, Ceph | 13:22 |
| ivve | currently i have tested ceph | 13:22 |
| goldyfruit | we had trouble because of the amount of data (iops) | 13:22 |
| ivve | and that was insanely slow (to get graphs to grafana) | 13:23 |
| goldyfruit | SSD pool ^^ | 13:23 |
| ivve | i guess ssd/nvme is a must? | 13:23 |
| goldyfruit | First it was on harddrive then Gnocchi was too slow to process the metrics | 13:23 |
| ivve | also gnocchi queries are very poorly documented | 13:23 |
| goldyfruit | SSD solved the issue | 13:23 |
| ivve | i have no clue how to query something in grafana | 13:23 |
| ivve | gnocchi.xyz explains everything extremly vaguely | 13:24 |
| ivve | imho | 13:24 |
| goldyfruit | Yeah | 13:24 |
| ivve | i had some older queries that worked in rocky | 13:24 |
| ivve | upgraded | 13:24 |
| ivve | only memory worked for instances | 13:24 |
| ivve | so i lost all metrics for cpu, network and storage | 13:25 |
| ivve | that sucked | 13:25 |
| goldyfruit | arf | 13:25 |
| ivve | and haven't figured it out yet | 13:25 |
| ivve | due to docs.. :( | 13:25 |
| goldyfruit | Ask to Julien | 13:25 |
| goldyfruit | Julien Danjour | 13:25 |
| goldyfruit | Danjou* | 13:25 |
| goldyfruit | Pretty sure there is an IRC chanel | 13:25 |
| goldyfruit | I've to work :D | 13:27 |
| goldyfruit | I'm still there if you need help with Reno | 13:27 |
| ivve | but the openstack node exporters don't gather metrics from compute workload? | 13:27 |
| ivve | ah alright, thanks for the talk tho :) | 13:27 |
| goldyfruit | You will need a bunch of exporter | 13:28 |
| goldyfruit | exporters* | 13:28 |
| goldyfruit | across controller, compute, storage, etc... | 13:28 |
| ivve | there was this openstack exporter from linaro? | 13:28 |
| goldyfruit | This collects only the resources like instances, images, volumes, etc... | 13:29 |
| ivve | ah | 13:29 |
| ivve | not utilization | 13:29 |
| ivve | which is what ceilometer gets | 13:29 |
| goldyfruit | It's very useful for capacity planing | 13:29 |
| goldyfruit | If you really needs metrics then I guess Monasca will be for you | 13:29 |
| goldyfruit | Ceilometer/Gnocchi gives the basics (cpu, ram, disk, network, iops) | 13:30 |
| ivve | yes | 13:30 |
| ivve | i wanna give gnocchi a chance | 13:30 |
| goldyfruit | Since our Gnocchi is on SSD pool in our Ceph we don't have any issue with it! | 13:31 |
| *** whoami-rajat has quit IRC | 13:47 | |
| *** goldyfruit has quit IRC | 13:51 | |
| *** goldyfruit has joined #openstack-freezer | 13:51 | |
| *** whoami-rajat has joined #openstack-freezer | 13:57 | |
| *** goldyfruit_ has joined #openstack-freezer | 14:42 | |
| *** goldyfruit has quit IRC | 14:42 | |
| *** goldyfruit_ has quit IRC | 14:50 | |
| *** goldyfruit has joined #openstack-freezer | 14:50 | |
| *** whoami-rajat has quit IRC | 16:07 | |
| *** whoami-rajat has joined #openstack-freezer | 16:12 | |
| *** spsurya has quit IRC | 17:40 | |
| *** goldyfruit has quit IRC | 21:17 | |
| *** goldyfruit has joined #openstack-freezer | 21:18 | |
| ivve | goldyfruit: seems the problem is ceilometer and not gnocchi.. doesn't seem to be able to poll cpu_util | 21:22 |
| ivve | enabled all pollsters and full pipeline but tons of stuff is missing and getting weird errors about volumes missing | 21:23 |
| ivve | wierd | 21:23 |
| goldyfruit | Your gnocchi policies are ok | 21:24 |
| *** goldyfruit has quit IRC | 21:26 | |
| *** goldyfruit has joined #openstack-freezer | 21:27 | |
| *** whoami-rajat has quit IRC | 21:27 | |
| *** goldyfruit_ has joined #openstack-freezer | 21:29 | |
| *** goldyfruit has quit IRC | 21:29 | |
| goldyfruit_ | ivve, what about your ceilometer pipeline ? | 21:30 |
| *** goldyfruit_ has quit IRC | 21:52 | |
| *** goldyfruit has joined #openstack-freezer | 21:58 | |
| *** goldyfruit has quit IRC | 22:19 | |
| *** goldyfruit has joined #openstack-freezer | 22:38 | |
| *** goldyfruit has quit IRC | 23:02 | |
| *** goldyfruit has joined #openstack-freezer | 23:34 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!