search

SUSE / DeepSea

A collection of Salt files for deploying, managing and automating Ceph.
GNU General Public License v3.0
160 stars 75 forks source link

Improve OSD zap documentation? #1060

Open smithfarm opened 6 years ago

smithfarm commented 6 years ago

With filestore OSDs, it was enough to do sgdisk -Z --clear but with bluestore this is no longer sufficient.

DeepSea zaps OSDs (filestore, bluestore, whatever) when one does:

salt-run disengage.safety
salt-run state.orch ceph.purge

The man page also mentions:

salt 'storage*' osd.wipe
salt 'storage*' osd.destroy

but it's not clear how to use these, for example, to zap an individual OSD. Could the documentation be improved in this respect?

swiftgist commented 6 years ago

I believe if you check https://github.com/SUSE/DeepSea/blob/master/srv/salt/_modules/osd.py#L1537, it does that and a bit more.

smithfarm commented 6 years ago

@swiftgist Has that been backported to SES5?

smithfarm commented 6 years ago

OK, from a cursory glance it looks like the destroy code is in SES5 as well.

What remains is to raise awareness of how to use it. I've had several people approach me about this - they had become accustomed to zap filestore OSDs with "sgdisk", which no longer works with bluestore, and were completely unaware (as was I) that DeepSea has this capability.

In the documentation I found: salt 'storage*' osd.destroy

Does this zap all OSD disks in the entire cluster?

smithfarm commented 6 years ago

I changed the issue title and re-wrote the description.