| 6 May 2025 |
 Cobalt | It's nice for my homelab but that is a downside I didn't see discussed yet. | 20:11:20 |
 hexa | we run debian hosts with unattended-upgrades and needrestart for basic sanity | 20:17:49 |
| 9 May 2025 |
| hexa |
Failed instance creation: Invalid instance name: Name can only contain alphanumeric and
hyphen characters
| 13:15:47 |
| hexa | what kind of silly limitation is this 😕 | 13:15:52 |
| hexa | my preferred naming scheme for instance … their FQDN | 13:16:17 |
| hexa | * my preferred naming scheme for instances is unsurprisingly … their FQDN | 13:16:26 |
| hexa | apparently goes back to 0.1.0 | 13:18:32 |
 adamcstephens | seems a bit odd | 14:31:41 |
| adamcstephens | surprised you hadn't hit that before today :) | 14:31:48 |
| hexa | yeah, me too 😄 | 14:31:56 |
| hexa | ran into another issue with terraform 😢 | 14:32:10 |
| hexa | with a zfs backed cluster, could you imagine that the instance does not get created on the machine where it creates the volume? | 14:32:45 |
| hexa | * with a zfs backed cluster, could you imagine that the instance does not get created on the machine where it created the volume? | 14:32:54 |
| hexa |
Failed instance creation: Failed creating instance record: Failed initializing instance: Failed add validation for device "gitlab-runner2-docker": Failed loading custom volume: Storage volume not found
| 14:33:25 |
| adamcstephens | with a separate volume resource? | 14:34:23 |
| hexa | yes | 14:34:38 |
| adamcstephens | i guess that makes sense as a possibility | 14:35:25 |
| hexa | resource "incus_storage_volume" "gitlab-runner2-docker" {
name = "gitlab-runner2-docker"
pool = "tank"
content_type = "block"
config = {
size = "50GiB"
}
}
resource "incus_instance" "gitlab-runner2" {
name = "gitlab-runner2"
device {
type = "disk"
name = "gitlab-runner2-docker"
properties = {
source = "tank/incus/custom/default_gitlab-runner2-docker"
pool = incus_storage_volume.gitlab-runner2-docker.pool
}
}
}
| 14:35:31 |
| hexa | removed some sections for brevity | 14:35:41 |
| adamcstephens | seems like a bug | 14:35:56 |
| hexa | yeah, can you imagine? | 14:36:20 |
| adamcstephens | does a volume migrate with its instance across nodes? | 14:36:46 |
| hexa | well, it does for the root volume | 14:37:11 |
| hexa | let me test a custom volume | 14:37:16 |
| adamcstephens | yeah, but secondary volumes are a bit odd in incus | 14:37:23 |
| hexa |
Shared filesystem are incompatible with migration.stateful=true
| 14:38:36 |
| hexa | ok, so block devices | 14:38:42 |
| hexa |
Only additional disks coming from a shared storage pool are supported with migration.stateful=true
| 14:39:29 |
| hexa | yeah, ok | 14:39:38 |
| hexa | fucked | 14:39:41 |
