Ceph failed OSD disk replace

[Hai]

Hello,

We have a failed OSD disk to replace on a production server. We have to hot replace it using remote hands in data center.

Are theese steps correct?

1.Set global OSD flags: noout/norebalance/nobackfill
2.Out and destroy failed disk
3.Wipe failed disk under Disks /dev/sdx
4.Physically remove failed disk.
5.Insert new disk
6.Create OSD with new disk
7. Remove global OSD flags.

Please advise, thank you!

 

[aaron]

Sounds okay.

3.Wipe failed disk under Disks /dev/sdx

Depending on how thorough you need to wipe the disk, the "Cleanup" option migh be enough. But IIRC it will only wipe the partition and LVM settings and won't zero out the disk. If it is an SSD, you want to issue a "secure erase" command anyway instead of zeroing it out manually.

 

[Hai]

Sounds okay.


Depending on how thorough you need to wipe the disk, the "Cleanup" option migh be enough. But IIRC it will only wipe the partition and LVM settings and won't zero out the disk. If it is an SSD, you want to issue a "secure erase" command anyway instead of zeroing it out manually.

Thank you, aaron!

 

[Hai]

Everything went smoothly. Not needing to perform step 3. Thanks again!

 

[aaron]

Good to hear