The Ceph documentation for adding drives shows the use of the prepare statement twice, with the same drive for the journal as shown below (where /dev/ssd is the journal drive):

enter image description here

Once the prepare statement has been executed once for preparing /dev/sdb, the partitions have already been made on /dev/ssd, so preparing /dev/sdc fails.


Is the prepare statement supposed to work by destroying and recreating the appropriate number of partitions on /dev/ssd (now 2 instead of just 1) or are we expected to manually configure the partitions ourselves and the documentation is wrong? If we manually create our own partitions on the shared journal drive, what command do we specify for preparing drives sdb and sdc, since if we leave out the journal part (:/dev/ssd) on each of the prepare commands, it will create two partitions on each drive instead of just one, as it expects the second partition to be used for journaling? Finally, if we manually create our own journalling partitions, should they use the XFS filesystem in production?


you may parted ssd to two partition /dev/ssd1 /dev/ssd2

parted sdb to one partition sdb1
parted sdc to one partition sdc1

and then

ceph-deploy osd prepare osdserver1:/dev/sdb1:/dev/ssd1
ceph-deploy osd prepare osdserver1:/dev/sdc1:/dev/ssd2

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.