| *** cloudnull has joined #openstack-tc | 00:02 | |
| *** tkajinam has quit IRC | 00:59 | |
| *** tkajinam has joined #openstack-tc | 00:59 | |
| adriant | Can we please try and push for a discussion around what our long term goal for Gnocchi might be? During the mess that was: https://review.opendev.org/#/c/744592 I've been doing some research on alternatives, and honestly... I can't say the idea of another time series DB fills me with hope. | 03:39 |
|---|---|---|
| *** diablo_rojo has quit IRC | 03:40 | |
| adriant | mnaser, ttx, InfluxDB, given that it needs enterprise for clustering isn't a good option... TimescaleDB is better, but ultimately PostgreSQL-ish with no easy integration. | 03:57 |
| adriant | I know that you don't really want us to double down on Gnocchi, but really... it kinda feels like the least painful option. | 03:58 |
| adriant | IF (big if), we can get 'trust' from JD and bring it back to OpenStack | 03:58 |
| adriant | I love the idea of having just an API around a DB we don't have to maintain, but it has to support all the same features Gnocchi give us without shelling out for enterprise products. | 04:01 |
| adriant | multi-tenancy we can build into the API layer, but there are features that we otherwise might not have, which Gnocchi gives us. | 04:03 |
| adriant | not to mention... writing an essentially new API/integrations may be more effort than continuing to maintain Gnocchi if we do end up moving it back officially to OpenStack. | 04:04 |
| *** evrardjp has quit IRC | 04:33 | |
| *** evrardjp has joined #openstack-tc | 04:33 | |
| *** belmoreira has joined #openstack-tc | 06:29 | |
| *** dklyle has quit IRC | 06:37 | |
| *** ralonsoh has joined #openstack-tc | 07:03 | |
| *** njohnston has quit IRC | 07:15 | |
| *** tosky has joined #openstack-tc | 07:25 | |
| *** e0ne has joined #openstack-tc | 08:00 | |
| ttx | adriant: thanks for doing the research. Last time this cropped up the state of time-series DBs was sad as well, which is why we punted the decision to later | 08:29 |
| ttx | I guess one alternative is to somehow plug into prometheus and say it's not our problem | 08:31 |
| ttx | adriant: Was Prometheus/TSDB an option you looked into? | 08:33 |
| ttx | (personally I'd rather spend time on hooking our wagon to the Prometheus train, than to another open-core-ish database that may or may not break us in the future) | 08:38 |
| *** slaweq_ has joined #openstack-tc | 11:13 | |
| *** slaweq has quit IRC | 11:15 | |
| *** redrobot has quit IRC | 12:10 | |
| *** zbitter is now known as zaneb | 12:10 | |
| zaneb | ttx: if you read the Prometheus docs it's pretty clear about it not being the right tool for several of the use cases we have for Telemetry | 12:11 |
| zaneb | e.g. it specifically says you should not use it for billing, because its goal is not to give 100% accurate data on usage - stuff can be lost | 12:12 |
| *** lpetrut has joined #openstack-tc | 13:21 | |
| ttx | zaneb: ah, fun. How does billing work in a pure-K8s world, then? | 13:24 |
| fungi | "it will be solved by the time we go to production" ;) | 13:45 |
| fungi | but in truth, i worked at companies who billed on lossy data; as long as you had fine granularity, a tiny percentage of lost counts didn't really eat into revenue, and then if customers complained about inaccuracies you could show that you were under-billing them | 13:48 |
| ttx | yeah sometimes simple beats accurate | 13:54 |
| fungi | for example, i built an elaborate bandwidth utilization billing system around netflow reporting from routers, because sales wanted to be able to tell customers that traffic between their systems was free and they'd only be charged for traffic to/from the internet, but all the traffic crossed the same router ports so we couldn't just bill on port utilization | 13:56 |
| fungi | netflow reports were via udp, so were not guaranteed to arrive at the collector | 13:57 |
| fungi | but they mostly did, and that was good enough to bill on | 13:57 |
| *** dklyle has joined #openstack-tc | 14:37 | |
| smcginnis | I found my cinderclient/osc gap ethercalc, if anyone is interested: https://ethercalc.openstack.org/cinderclient-osc-gap | 15:00 |
| smcginnis | I haven't touched it for quite awhile, but I don't think much has changed since then. | 15:00 |
| *** slaweq_ is now known as slaweq | 15:02 | |
| *** belmoreira has quit IRC | 15:03 | |
| *** lpetrut has quit IRC | 15:06 | |
| *** rpittau is now known as rpittau|afk | 15:18 | |
| *** tdasilva has joined #openstack-tc | 15:30 | |
| *** abhishekk is now known as abhishekk|afk | 15:42 | |
| *** abhishekk|afk is now known as abhishekk | 16:10 | |
| *** bnemec is now known as beekneemech | 16:18 | |
| *** tdasilva_ has joined #openstack-tc | 16:33 | |
| *** tdasilva has quit IRC | 16:35 | |
| *** tdasilva_ has quit IRC | 17:14 | |
| *** tdasilva_ has joined #openstack-tc | 17:14 | |
| *** lxkong has quit IRC | 17:33 | |
| *** ralonsoh has quit IRC | 17:48 | |
| *** tdasilva_ has quit IRC | 20:12 | |
| *** thiago__ has joined #openstack-tc | 20:12 | |
| *** thiago__ has quit IRC | 20:36 | |
| *** thiago__ has joined #openstack-tc | 20:37 | |
| *** slaweq has quit IRC | 20:47 | |
| *** tdasilva_ has joined #openstack-tc | 20:47 | |
| *** thiago__ has quit IRC | 20:50 | |
| *** slaweq has joined #openstack-tc | 20:55 | |
| *** slaweq has quit IRC | 21:00 | |
| *** tdasilva_ has quit IRC | 21:06 | |
| *** tdasilva_ has joined #openstack-tc | 21:06 | |
| openstackgerrit | Merged openstack/governance master: Drop all exceptions for legacy validation https://review.opendev.org/745403 | 21:21 |
| *** tosky has quit IRC | 23:07 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!