Story Crater Bot
6e97def009
Longhorn's StorageClass nodeSelector matches node tags (nodes.longhorn.io spec.tags), not k8s hostnames - "talos-cp-3" was never set as a node tag, so every PVC provision attempt failed with "specified node tag talos-cp-3 does not exist". diskSelector: paperless-media already pins placement correctly on its own.
28 lines
1.3 KiB
YAML
28 lines
1.3 KiB
YAML
# Dedicated StorageClass for paperless-ngx document media, backed by the 4TB
|
|
# USB HDD on talos-cp-3 (see longhorn-add-disks-job.yaml). Single disk, single
|
|
# node — no Longhorn replica is possible, so numberOfReplicas is 1 by
|
|
# necessity, not choice. diskSelector pins placement to the tagged disk only,
|
|
# so a volume from this class never lands on cp-3's regular (already
|
|
# DiskPressure) default pool. reclaimPolicy is Retain, not Delete: a PVC
|
|
# accident here has no replica to fall back on, so an accidental delete must
|
|
# not also take the underlying volume with it.
|
|
apiVersion: storage.k8s.io/v1
|
|
kind: StorageClass
|
|
metadata:
|
|
name: longhorn-paperless-media
|
|
provisioner: driver.longhorn.io
|
|
allowVolumeExpansion: true
|
|
reclaimPolicy: Retain
|
|
volumeBindingMode: WaitForFirstConsumer
|
|
parameters:
|
|
numberOfReplicas: "1"
|
|
# diskSelector alone is sufficient: only cp-3 has a disk tagged
|
|
# paperless-media, so placement is already pinned. A nodeSelector value
|
|
# here must be a Longhorn *node* tag (set via nodes.longhorn.io spec.tags),
|
|
# not a Kubernetes hostname - "talos-cp-3" was never a node tag, which made
|
|
# every provision attempt fail with "specified node tag talos-cp-3 does
|
|
# not exist".
|
|
diskSelector: "paperless-media"
|
|
staleReplicaTimeout: "30"
|
|
fsType: "ext4"
|