| 19 Jun 2026 |
 Mic92 | Before it was failing at the large transaction that makes the system go out of disk | 12:06:27 |
 hexa | nice | 12:06:43 |
| Mic92 | do we have zfs snapshots? | 12:07:24 |
| hexa | no | 12:07:38 |
| Mic92 | Okay than zfs just needs a bit to free up space | 12:07:52 |
| hexa | certainly | 12:08:04 |
| hexa | watching zfs list it's doing | 12:08:25 |
| Mic92 | Slowly making progress | 12:15:31 |
| Mic92 | zroot/nix 610G 457G 96K none
| 13:25:58 |
| Mic92 | Well, gc gives me enough time to fix the issue that was introduced by downloading nars again to compute the build product | 13:53:03 |
| Mic92 | Before this was only downloading a single NAR and no closure | 13:53:15 |
| Mic92 | Also doesn't need to be persisted | 13:53:25 |
| Mic92 | Can we not have this feature :) | 13:53:31 |
| hexa | all good, would be good to get going | 14:32:23 |
| hexa | and good to vacuum the db before we get going | 14:32:31 |
| Mic92 | GC is finished now. | 14:33:20 |
| hexa | ok, then I'll vacuum | 14:33:37 |
| Mic92 | already in progress | 14:34:11 |
| hexa | hm ok | 14:34:13 |
| hexa | looks like we're down to 6.5G | 14:39:09 |
| hexa | zroot/nix/store 77.4G 987G 77.4G /nix
| 14:39:33 |
| hexa | still a lot | 14:39:38 |
| Mic92 | hydra evals do that | 14:40:31 |
| Mic92 | They create gcroots | 14:40:37 |
| hexa | yup | 14:40:55 |
| hexa | does it make sense to keep much around now? | 14:41:05 |
| Mic92 | We could remove them to, we would make new evals | 14:41:45 |
| hexa | and is keeping of jobset results still working? | 14:41:47 |
| hexa | now that the queue-runner is not getting the outputs any more | 14:42:07 |
| Mic92 | ah you mean long term | 14:42:27 |
