We're actually pursuing a similar configuration where it's easily conceivable that we would have 230+ block devices that we want to mount on a server. We are moving to a configuration where each user in our cluster has a distinct ceph block device for their storage. We're mapping them on our nas server and serving them to other nodes via NFS. (BTW, we are growing their storage using LVM abstractions, so each volume may have multiple RBD devices as physical volumes.) While 230 is not a big number in this scenario, it's not clear to me if there are other hard-coded or performance limitations which would make reaching even 230 block devices unreasonable on a single server. ~jpr On 10/08/2013 02:04 PM, Wido den Hollander wrote: > On 10/08/2013 07:58 PM, Gaylord Holder wrote: >> Always nice to see I've hit a real problem, and not just my being dumb. >> > > May I ask why you are even trying to map so many RBD devices? Do you > need access to >230 all at the same time on each host? > > Can't you map them when you need them and unmap them when they are no > longer required? _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |