Good morning, what Ceph version is this? Apparently it's not cephadm managed? If it is, there's no need to fiddle with ceph-volume yourself, the orchestrator can handle that for you, either by using a suitable spec file or via command line. Every now and then users on this list discuss about multiple OSDs per NVMe, you don't really benefit from that anymore. So you might want to reconsider and maybe deploy one OSD per drive. If you don't want to reconsider, you should be able to gather from the ceph-volume.log what might be preventing the creation of the second OSD.
Can I re-run the ceph-volume command and it is smart enough to see the first OSD is already provisioned and then create the missing second OSD? Or should I zap and purge the OSD which was created and then re-run the ceph-volume command?
You can safely test what happens without any impact. The prepare command doesn't activate the OSDs yet, so there's no data moving around until the OSDs are up. And if you set osd_crush_initial_weight = 0, you're even safer: the newly created OSDs get a crush weight of 0, meaning there's no remapping happening until you're satisfied with the result. We've been using that for 10 years now. Regards, Eugen Zitat von Elias Carter <elias@dropbox.com>:
I provision OSDs with the following command: ``` ceph-volume --log-path /tmp/ceph-log lvm batch --prepare --bluestore --yes --dmcrypt --osds-per-device 2 /dev/nvme0n1 --crush-device-class nvme ```
What I have observed is that sometimes ceph-volume will provision the first OSD successfully, but then it will fail to provision the second OSD. When this happens ceph-volume does not attempt to clean up the first OSD which was created.
My question is what should be done in this scenario?
Can I re-run the ceph-volume command and it is smart enough to see the first OSD is already provisioned and then create the missing second OSD? Or should I zap and purge the OSD which was created and then re-run the ceph-volume command?
Elias _______________________________________________ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-leave@ceph.io