adriant | ttx, zaneb, it's not even quite the granularity, Prometheus at least to me doesn't quite feel as suitable for a bunch of the use cases we have with Ceilometer/Gnocchi. InfluxDB and TimescaleDB are closer because we have a certain amount more flexibility over the data. I could be wrong, but in my experience the benefit (and in this case detriment) | 01:34 |
---|---|---|
adriant | of prometheus is how opinionated it is about the data it stores/uses. I feel like we wouldn't get the same level of functionality out of it for resource info/metadata and such. | 01:34 |
adriant | But then again, I don't know enough about Prometheus that closely to make a valid choice. | 01:38 |
adriant | I'd also just be worried about the level of migration needed to move to a fully Prometheus focused world. In OpenStack that feels like too far away as even a possibility. | 01:39 |
adriant | plus Prometheus doesn't appear to have native multi-tenancy, so we'd either be buying into another service that wraps around it, or building one that tags each metric with a project id, and filters by it | 01:44 |
adriant | So... another API like Ceilometer v1 or Gnocchi | 01:44 |
adriant | with a collector service that includes the project id, or support for exporters in the services themselves to provide metrics | 01:45 |
*** evrardjp has quit IRC | 04:33 | |
*** evrardjp has joined #openstack-tc | 04:33 | |
*** dklyle has quit IRC | 06:39 | |
*** tosky has joined #openstack-tc | 07:42 | |
*** e0ne has quit IRC | 07:51 | |
*** slaweq has joined #openstack-tc | 13:00 | |
*** mrunge has joined #openstack-tc | 15:52 | |
*** njohnston_ has joined #openstack-tc | 16:12 | |
*** e0ne has joined #openstack-tc | 19:16 | |
*** njohnston_ has quit IRC | 20:06 | |
*** e0ne has quit IRC | 20:32 | |
*** irclogbot_0 has quit IRC | 21:07 | |
*** irclogbot_0 has joined #openstack-tc | 21:12 | |
*** tosky has quit IRC | 22:49 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!