On Mon, Apr 22, 2013 at 9:26 AM, konradwro <konradwro@xxxxx> wrote: > Hello, i have cephfs mounted on each node, when i put file on node1 i can > see file on node2 and its ok. But cephfs + apache + php5 works quite slow, > with noticeable delay. I think your best option is to optimize this setup. Lets first make sure you're not hitting a bug in libcephfs. There's been some discussion of apache (web workloads) on top of cephfs already: http://www.mail-archive.com/ceph-devel@xxxxxxxxxxxxxxx/msg10991.html So, do you have anything interesting showing up in dmesg? What workloads are you testing your apache setup with? How are you measuring performance? > > I've tested RBD + apache + php5 on one of the nodes, which give satisfactory > boost in performance. > > How to create a cluster joining all the nodes, using RDB, to mount the same > filesystem on every node?. This most likely isn't the right solution for your needs. The filesystems that you would create on an rbd block device (ext4, xfs, etc.) are all single node file systems, and don't work well when shared across nodes. Doing so will corrupt your filesystem. -sam > > > _______________________________________________ > ceph-users mailing list > ceph-users@xxxxxxxxxxxxxx > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |