Troubleshoot an encrypted Ceph OSD¶

During the deployment of a Ceph cluster, an encrypted OSD may fail to be prepared or activated and thus fail to join the Ceph cluster. In such case, remove all the disk partitions as described below.

Workaround:

1. From the Ceph OSD node where the failed encrypted OSD disk resides, erase its partition table:

   dd if=/dev/zero of=/dev/<<ADD>> bs=512 count=1 conv=notrunc
   

2. Reboot the server:

   reboot
   

3. Run the following command twice to create a partition table for the disk and to remove the disk data:

   ceph-disk zap /dev/<<ADD>>;
   

4. Remove all disk signatures using wipefs:

   wipefs --all --force /dev/<<ADD>>*;