Using shared SAS disk array for Openstack ?

asked 2020-03-23 04:16:50 -0500

lbndev gravatar image

Hello,

I have a few recycled servers I'd like to use to deploy Openstack. Among them, there are two HP DL360 G7, both connected to the same HP MSA P2000 SAS disk array.

I've been reading about Cinder, Ceph, CRUSH rules, etc... and I still can't figure out how I should use this hardware. I'd like the cluster to be able to use disk space from the "internal" disks and from the disk array.

How can I configure (OSDs / CRUSH rules / ???) so that the Openstack cluster makes optimal use of the disk array ? By "optimal" I mean that it should "understand" that data stored on this array can be accessed through either one of the servers physically attached to it.

From my understanding, CRUSH rules are hierarchical, and I I couldn't find a way to express "/dev/sdc on nodeA is the same disk as /dev/sdc on nodeB".

I've also found that there is a specific Cinder driver for HP MSA arrays using FC or iSCSI, but not for SAS. Is there another way to better use this hardware for Openstack (= replace Ceph by something else) ?

Thank you

edit retag flag offensive close merge delete

Comments

You don't want Ceph to be on that array, it has to manage its own disks. Could be used as shared storage for compute nodes for live migration or for highly available control services (stonith device). Another way is without ceph, just configure openstack with your SAS backend for cinder & nova.

eblock gravatar imageeblock ( 2020-03-24 03:19:49 -0500 )edit