Thursday, 2022-09-29

prometheanfirewould I be right to say that the elk/grafana roles in osa-ops are fairly unmaintained?02:43
*** ysandeep|sick is now known as ysandeep03:48
*** ysandeep is now known as ysandeep|afk04:45
jrosser_prometheanfire: really only my team seem to use it significantly - we have a quite large elk deployment built from the ops repo stuff05:55
*** ysandeep|afk is now known as ysandeep06:19
noonedeadpunkwe aimed to use it, but then with elastic license change folks decided not to. 07:26
*** ysandeep is now known as ysandeep|lunch08:05
jrosser_https://www.terraform.io/language/upgrade-guides#removal-of-deprecated-state-storage-backends09:15
jrosser_goodbye swift backend in terraform09:15
noonedeadpunkyeah, saw thread in ML to get some volunteers to maintain it in terraform09:25
noonedeadpunkbtw rocky patch has passed finally09:27
noonedeadpunkjrosser_: you wanted to drop distro jobs from it before merging?09:28
jrosser_perhaps that depends what we want to backport?09:29
noonedeadpunkI think this exact patch is the only thing we need (except gluster patch that already landed)09:30
*** ysandeep|lunch is now known as ysandeep09:44
jrosser_well, if we want distro rocky9 on yoga then we should keep it as is09:44
jrosser_otherwise we should drop the distro jobs on master then backport09:44
jrosser_hmm well they are all non voting09:45
jrosser_lets merge this then tidy up the jobs09:46
jrosser_i don't even know why really we keep the focal one, it's been broken for ges09:46
jrosser_*ages09:46
noonedeadpunkjrosser_: they are broken because we're still isntalling yoga client and using master collections that require newer client10:07
noonedeadpunkSo assumption is that once we switch repos to use Zed, they will just work10:08
noonedeadpunkand they can jsut work on yoga as well10:08
spatelwho use swift nowadays ? may be legacy deployment. 10:36
noonedeadpunkwikimedia ? :D10:45
noonedeadpunkhttps://wikitech.wikimedia.org/wiki/Swift/Ring_Management10:46
noonedeadpunkThey tried out ceph and drop it out next week and forgot it ever existed for them10:46
noonedeadpunkspatel: this is better link https://wikitech.wikimedia.org/wiki/Media_storage#Media_storage10:47
spatel:D 10:49
noonedeadpunkI was even able to find patch that was testing/implementing ceph and was smiling reading comments with results10:50
noonedeadpunkthough rgw quite evolutionized since 2013 to be fair10:51
noonedeadpunkbut mostly feature-wise rather then design wise, and iirc they had design issues with their scale10:51
*** dviroel|out is now known as dviroel11:17
*** ysandeep is now known as ysandeep|afk12:12
opendevreviewDmitriy Rabotyagov proposed openstack/openstack-ansible master: Add Rocky Linux 9 to zuul and docs  https://review.opendev.org/c/openstack/openstack-ansible/+/85719112:17
opendevreviewDmitriy Rabotyagov proposed openstack/openstack-ansible master: Replace usage of which with command  https://review.opendev.org/c/openstack/openstack-ansible/+/85955012:17
*** dviroel is now known as dviroel|doc_appt12:27
*** ysandeep|afk is now known as ysandeep12:49
*** ysandeep is now known as ysandeep|dinner14:28
*** dviroel|doc_appt is now known as dviroel|doc14:46
*** dviroel|doc is now known as dviroel14:46
prometheanfireya, forgot about the elk licencing stuff til I logged into kibana and starting seeing gated things15:48
*** ysandeep|dinner is now known as ysandeep|out15:56
jrosser_tbh for looking at logs you don't need anything than the stuff under the Elastic Licence16:02
jrosser_and if you're trying to do things that are not looking at logs, then it's probably the wrong tool16:03
jrosser_we run ELK for logs and prometheus for metrics16:03
jrosser_and imho there is now more available without cost in ELK than there used to be to stay closer to opensearch16:05
jrosser_prometheanfire: you are probably going to have a bad time with metricbeat and particularly packetbeat16:06
jrosser_i just would not bother with those at all16:06
prometheanfirewe have a log ingestion thing at work, I may just set up osa to ship via the journald shipping or rsyslog if I have to16:08
prometheanfireand ya, prometheus for metrics, I imagine you are not using the osa-ops prometheus stuff?  are you monitoring the individual containers, openstack services, hype 'fullness'?16:14
prometheanfireatm I plan on letting haproxy monitor services for me and I monitor haproxy, galera cluster status and hype status, along with systemd status (failed / restarted units)16:15
jrosser_we have committed some stuff to OSA to enable prometheus exporters on haproxy and galera, and i think there is one also in rabbitmq16:20
prometheanfireah, ya, I think I remember seeing vars for that now16:21
jrosser_then we also have the openstack exporter and also i think libvirt-exporter16:21
prometheanfireneat16:22
*** chkumar|rover is now known as raukadah17:15
lowercasejrosser_: this one? https://github.com/zhangjianweibj/prometheus-libvirt-exporter17:54
lowercaseif you sent a message just now, my vpn dropped for a sec.17:55
jrosser_lowercase: this one https://opendev.org/vexxhost/libvirtd_exporter17:55
lowercaseno readme?17:56
lowercaseand the project hasn't been touched in 2 years 17:56
jrosser_i expect contributions are welcome17:57
lowercasei found the doc17:57
mgariepyyou can bind it to a RO socket on the libvirt side.\18:12
*** dviroel is now known as dviroel|afk19:42
*** dviroel|afk is now known as dviroel20:52
*** tosky_ is now known as tosky21:24
*** ysandeep|out is now known as ysandeep23:58

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!