4

I recently set up my Kubernetes cluster with three worker and one master node, as storage service I'm using rook-ceph. As I know the rook pods require a raw disk in order to run the osd-pods.

Last time I managed to create them as well within a disk partition (primary) that I labeled as 'physical volume'. The OSD-pods where running. back then. The problem now is that I cannot recreate this state again.

Does anyone know how to configure a disk partition (created with fdisk), such that it gets recognized as a raw disk on its own? I'm using Ubuntu 18.04.

If the problem lies within the image tags (for Ceph/ceph:14.XXX and rook/ceph 1.0.0), which image tags should I use?

The last time it worked for me I used the files from Rooks GitHub page: https://github.com/rook/rook/tree/master/cluster/examples/kubernetes/ceph

XH_P
  • 41

1 Answers1

0

I could recreate it and the solution is that you indeed have to clean up the disk. For this you can use fdisk (delete all partitions) and then unmount everything that is mounted to the certain disk. After that remove the unmounted directories from /etc/fstab. Then you are good to go!

XH_P
  • 41