Dear Ceph Developers:
At present, we are deploying ceph by cephadm.
We have no idea how to deploy the scenario as follows.
Disk Information:
data_device: /dev/sdb /dev/sdc
db_device: /dev/sdm /dev/sdn
Deploy result:
osd_service1:
osd1: /dev/sdb is used for data. /dev/sdm is used for db.
osd2: /dev/sdc is used for data. /dev/sdn is used for db.
osd_services2:
osd3: /dev/sdm is used for data.
osd4: /dev/sdn is used for data.
Two options are being considered by us.
1) /dev/sdm and /dev/sdn are partitioned into two partitions respectively.
osd1: /dev/sdb /dev/sdm1
osd2: /dev/sdc /dev/sdn1
osd3: /dev/sdm2
osd4: /dev/sdn2
Partitions are not supported by 'ceph-volume batch' and cephadm service is not applicable in this case. So the option is difficult to implement even if we want to change the code of cephadm.
2) /dev/sdm and /dev/sdn create multi-lvs.
osd1: /dev/sdb /dev/sdm--ceph-lvs1(It means that /dev/sdm has a lvs named ceph-lvs1)
osd2: /dev/sdc /dev/sdn--ceph-lvs1
osd3: /dev/sdm--ceph-lvs2
osd4: /dev/sdn--ceph-lvs2
osd1 and osd2 are easy to implement by cephadm. But osd3 and osd4 are difficult to implement.
Do you have a good idea for this scenario?
Thanks very much.
_______________________________________________ Dev mailing list -- dev@xxxxxxx To unsubscribe send an email to dev-leave@xxxxxxx
