I have a bunch of sparc servers that have 2 disks a piece. I was wondering if there was a way to bunch them all together in one zpool? I couldn''t find any information on this... This message posted from opensolaris.org
On Wed, 3 May 2006, Donald Lee wrote:> I have a bunch of sparc servers that have 2 disks a piece. I was > wondering if there was a way to bunch them all together in one zpool? I > couldn''t find any information on this...I don''t think you can do this. My understanding is that all ZFS devices must be (or at least, appear to be) local. I guess you might be able to export the actual device for a disk, but it sounds dodgy to me... -- Rich Teer, SCNA, SCSA, OpenSolaris CAB member President, Rite Online Inc. Voice: +1 (250) 979-1638 URL: http://www.rite-group.com/rich
On May 3, 2006, at 13:33, Donald Lee wrote:> I have a bunch of sparc servers that have 2 disks a piece. I was > wondering if there was a way to bunch them all together in one zpool? > I couldn''t find any information on this...One thing we''ve chatted about to address this issue (but not tested, partly because all of the required pieces are not yet in place) is exporting the drives via iSCSI. Of course, this can''t be done until both initiator and target support for iSCSI are in Solaris, and I don''t know what the schedule for that might be. --Ed -- Ed Gould Sun Microsystems File System Architect Sun Cluster ed.gould at sun.com 17 Network Circle +1.650.786.4937 MS UMPK17-201 x84937 Menlo Park, CA 94025
It''s probably not quite what you had in mind, but this will achieve the result. I''ve done something similar for zfs testing. Have each individual machine schlep out the disk as a target via iscsi. I haven''t happened to notice whether solaris has an iscsi target, but netbsd''s target works with the opensolaris initiator. Pick another machine somewhere (or one of these) to pull in all the remote disks with its iscsi initiator, and assemble them into a pool. Give some consideration to the fault scenarios that this type of disk access could entail :) If you need to, share the filesystems back to the other nodes. This message posted from opensolaris.org