| kelvinliu | wallyworld: do u remember why we have the uniqID in statefulset annotation for naming storage? | 04:26 |
|---|---|---|
| kelvinliu | is it because PVs are global? | 04:27 |
| wallyworld | kelvinliu: so that if a pod gets destroyed and a new one comes up, we know which pvc to reattach. i think, that's from memory | 04:27 |
| kelvinliu | wallyworld: for statefulset, we only mange pvc template but not pvc, so storage config is always in statefulset level rather than pod level | 04:39 |
| wallyworld | yeah, there's some slightly different reason. it may be to do with allowing juju storage model to correctly tie the juju bit to the k8s bit. i'll look at the code and see if i can remember | 04:40 |
| kelvinliu | wallyworld: the application package does have those special storage handling logic, I am trying to understand if all those logic are necessary for the new stuff. | 04:43 |
| kelvinliu | wallyworld: I will probably just start from the simple version and land what we have, then rotate and enhance later when we need? | 04:44 |
| wallyworld | sure, we can iterate | 04:46 |
| icey | Morning manadart | 07:10 |
| manadart | Morning icey. | 07:29 |
| stickupkid | manadart, HO? | 10:48 |
| Chipaca | status-set is immediate, yes? | 12:51 |
| Chipaca | or is it only applied when a hook finishes successfully? | 12:52 |
| stickupkid | manadart, before you leave, HO? | 14:18 |
| jam | Chipaca: just to look like we're responsive... :) status-set is immediate | 14:18 |
| Chipaca | jam: :) thanks | 14:18 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!