Storage

Ceph on Proxmox: device classes and CRUSH rules to keep SAS, SSD and NVMe apart

rm-device-class + set-device-class on the OSD, crush rule create-replicated default host , and the pool onto that rule. No ‘class create’: the class is born with the first OSD that uses it.

LVM: the volume group will not activate after a power failure

First pvs/vgs/lvs -a: is a PV missing or did the thin pool fail? vgchange -ay –activationmode degraded for LVM RAID; partial only to copy data read-only; lvconvert –repair for thin pools; vgcfgrestore if the metadata is broken.

mdadm: add a disk and grow a RAID 5 (reshape) without losing data

–add brings the disk in as a spare; –grow –raid-devices=N starts the reshape (hours). Backup file off the array, touch nothing until /proc/mdstat finishes, then resize2fs or xfs_growfs.

Seafile: reclaim disk space with seaf-gc (the garbage collector)

Empty the library trash in the admin panel, stop Seafile (mandatory on CE), seaf-gc.sh –dry-run and then seaf-gc.sh. With the official Docker image, docker exec seafile /scripts/gc.sh does it all.

Move k3s data to another disk

Root disk is full. Stop k3s, move the directories, leave a symlink, start again.