Wednesday, 2021-04-07

*** gvincent has joined #ara07:09
*** _KaszpiR_ has quit IRC07:42
*** _KaszpiR_ has joined #ara07:49
*** mgariepy has quit IRC11:06
*** dpawlik4 has joined #ara11:40
*** dpawlik4 is now known as dpawlik11:42
*** mgariepy has joined #ara12:14
*** bwatson has left #ara13:33
dmsimardkodiak_f: re: #ansible, did you know that you could send slack notifications about playbook runs from ara with ansible ? I did this a while ago as a PoC but doesn't mean it's a good idea :p https://github.com/ansible-community/ara/blob/master/contrib/playbooks/slack-notifications.yaml14:48
dmsimardyou could use a similar pattern with irc or whatever14:48
kodiak_fha that's wild!14:48
kodiak_fI've got ARA current up and running and I'm mostly leaving it alone for now until the new version drops14:48
kodiak_fthen we're going all-in14:48
dmsimard\o/14:49
dmsimardI can't commit but I would like to say two weeks should be enough time to ship14:50
kodiak_fI have docker-compose working for bootstrapping a full mariadb-based ARA server, but need to pick a reverse proxy and build that in and test on new version before it'll be worth contributing14:50
kodiak_fI've diverted attention at this point to Ansible-CMDB but plan to circle back to ARA after the drop14:51
dmsimardnice! re: reverse proxy, if it can help there is some "prior art" in the ansible roles: https://github.com/ansible-community/ara-collection/blob/master/roles/ara_frontend_nginx/templates/ara-api.conf.j214:52
kodiak_fI'd love to eventually see all 3 projects hit some sort of state of symbiosis between Ansible-ARA, Ansible Semaphore, and Ansible-CMDB.  Right now I think I need to run all 3 for my team, AND settle on an execution scheduler (cron, systemd etc)14:52
kodiak_fI was torn between snagging the ARA reverse proxy frontend using NGINX or using this as a good excuse to finally run Traefik + Swarm14:53
dmsimardfwiw ara has the data that ansible-cmdb presents but has no "optimized" view for it like cmdb does14:54
dmsimardmaybe one day we'll have something similar :)14:54
kodiak_fyep - I see it in the facts.  One place where Ansible-CMDB is limited is the level of complexity with introduction of additional columns or custom facts14:54
dmsimardI haven't looked at semaphore in forever and didn't realize it was still a thing after awx was open sourced14:55
kodiak_fI'd love to be able to essentially feed ARA a YAML file that details what to keep in it's CMDB14:55
dmsimardbut it should probably work with ara given it's just a regular callback14:55
kodiak_fFor Semaphore there's been a huge recent push to revive the project14:55
dmsimardkodiak_f: can you define "what to keep" ? ara keeps host facts but that's it I guess14:56
kodiak_fI'd probably skip Semaphore and run AWX but I see two huge limitations to doing so:  1.  Release cadence+complexity of upgrade, and 2.  The much bigger problem of AWX moving to a Kubernetes-only production model14:56
kodiak_fsorry s/keep/present14:56
kodiak_fso we can't ahve a 1000 column CMDB in ARA, but being able to pick and choose from facts to be columns would be perfect14:57
dmsimardyeah worth thinking about it when we get there14:57
kodiak_fright now to introduce additional columns and facts in CMDB, it's complicated, though I'm grateful for the ability to do so - I've not tried yet but it's on my backlog14:58
dmsimardthere is a challenge in the database model/API in getting unique hosts but someone is working on it: https://github.com/ansible-community/ara/pull/24714:58
dmsimardI think after that lands (after 1.5.6) we might have a foundation to build on for displaying the latest data we have about a host which would bring it closer to ansible-cmdb14:59
kodiak_fThat would be very cool14:59
kodiak_fI dislike Ansible-CMDB slightly less than a week ago now that I've had time to focus on it, but I still think it's a stale project15:00
dmsimardthe author has opinions about YAML :) https://github.com/fboender/ansible-cmdb/blob/master/src/ansiblecmdb/ihateyaml.py15:04
kodiak_fI've noticed a trend.  Most Ops/SRE (like me) love YAML, then most real devs hate it15:05
kodiak_fI'm no coder, so I enjoy YAML because I don't hit it's limitations15:05
kodiak_fdmsimard - this is your project right?15:05
kodiak_fansible-ara15:05
dmsimardit's not "my" project but I'm the author yes :p15:08
*** mgariepy has quit IRC15:09
kodiak_fah I get ya.  Well, thanks for this project.  I didn't see a coffee link15:10
kodiak_fdef owe coffee15:10
*** gvincent has quit IRC15:13
dmsimardI don't seek donations or sponsorships but thanks for the offer :D15:17
kodiak_fOK well it's highly appreciated as someone who can't afford Tower15:18
*** JqckB has quit IRC15:22
*** JqckB has joined #ara15:24
*** mgariepy has joined #ara15:29
*** eliv has quit IRC15:39
*** eliv has joined #ara15:52
*** eliv has quit IRC15:59
*** eliv has joined #ara16:00
*** sshnaidm is now known as sshnaidm|afk16:06
*** ccamacho has quit IRC16:25
dmsimardok well I appreciate that you appreciate it \o/16:31
kodiak_fOn drawback to the non-DB nature of Ansible-CMDB that I could see working around in a reimagining of ARA-CMDB would be to have some history of how facts change on a host over time17:47
kodiak_fEvery time I cut a new CMDB contents, unless I do something fancy with versioning and symlinking, the prior CMDB contents go into the bit bucket17:47
dmsimardI guess we have the data already since host facts are stored uniquely per playbook, the challenge would be in how to display it17:49
dmsimardthere is also some amount of overlap between tools and functionality -- as in, ara might not be the right tool for this kind of stuff when there are monitoring tools or time series databases17:50
kodiak_fComing from the Splunk world having each fact be a drill-down that takes you to that record over time would work17:51
kodiak_fI don't see anyone using a TSDB for fact evolution but that would be swell to see - we use influx17:51
dmsimardonce upon a time I saw an implementation that dumped data from ara and sent it to influx so you could have stuff like task times in grafana and stuff17:52
dmsimardit was neat but it was also developed against 0.x and would likely need some work to make it work with 1.x17:53
kodiak_fthat's our stack - Telegraf --> InfluxDB1x --> Grafana.  I'm terrible at Grafana so I've not done much beyond basic graphs17:53
dmsimard0.x also didn't have an API and 1.x does so the implementation might end up being different17:54
dmsimardin case you're curious: https://opendev.org/openstack/ansible-role-collect-logs/src/branch/master/plugins/modules/ara_influxdb.py17:54
dmsimard^ sshnaidm|afk: out of curiosity do you know if that is still used ?17:55
kodiak_finteresting - will check it out.  I deployed ARA w/ a "real" database from the outset because I'm planning on shipping callbacks 24/717:56
kodiak_fas such I don't really mind growing the DB constantly with facts17:56
kodiak_fCool that all of this stuff lives within OpenStack repos.  Always thought that was the coolest project.  Deploying a 10 node Icehouse cluster the hard way was about the most fun project I ever got at work17:58
dmsimardara actually grew up in the openstack community and I have an openstack cloud in my basement :)17:58
dmsimardalso btw you can keep the database size in check by pruning older playbooks or depending on other criterias: https://ara.readthedocs.io/en/latest/cli.html#ara-playbook-prune17:59
kodiak_fI may have to prune but hope I won't.  I'd like to keep years of plays18:00
sshnaidm|afkdmsimard, not used afaik18:07
dmsimardsshnaidm|afk: ok no worries maybe we can revisit it one day :)18:11
dmsimardI am generally interested in mechanisms to export data from ara to other things like elasticsearch, influxdb and such18:12

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!