Hello all, I was thinking if such scenario could be possible: 1 - Export/import a ZFS filesystem in two solaris servers. 2 - Export that filesystem (NFS). 3 - Mount that filesystem on clients in two different mount points (just to authenticate in both servers/UDP). 4a - Use some kind of "man-in-the middle" to auto-balance the connections (the same IP on servers) or 4b - Use different IP''s and balance through DNS. After "little" problems with this initial (mount) setup, the NFS conversation should work without problems, without stale file handle issues, right? Thinking in such configuration i''m assuming some "concepts", that i''m asking your corrections if i''m wrong: 1) Using ZFS send/receive i will have the SAME filesystem across machines. 2) Using ZFS, the file handles in NFS protocol are not beeing made using disk luns, major/minor numbers, etc... I''m not willing to implement such solution, but what i really want to know is IF that configuration is possible, and IF i''m not wrong about my assumptions. Thanks a lot for your time! Leal. This message posted from opensolaris.org
msl wrote:> Hello all, > I was thinking if such scenario could be possible: > 1 - Export/import a ZFS filesystem in two solaris servers. > 2 - Export that filesystem (NFS). > 3 - Mount that filesystem on clients in two different mount points (just to authenticate in both servers/UDP). > 4a - Use some kind of "man-in-the middle" to auto-balance the connections (the same IP on servers) > or > 4b - Use different IP''s and balance through DNS. > After "little" problems with this initial (mount) setup, the NFS conversation should work without problems, without stale file handle issues, right? > > Thinking in such configuration i''m assuming some "concepts", that i''m asking your corrections if i''m wrong: > 1) Using ZFS send/receive i will have the SAME filesystem across machines. > 2) Using ZFS, the file handles in NFS protocol are not beeing made using disk luns, major/minor numbers, etc... > I''m not willing to implement such solution, but what i really want to know is IF that configuration is possible, and IF i''m not wrong about my assumptions. > > Thanks a lot for your time! > > Leal. > >I''m very confused. What are you trying to do? We already have Solaris Cluster which manages ZFS mounts, exports, and logical IP addresses. -- richard
On 3/13/08, Richard Elling <Richard.Elling at sun.com> wrote:> > msl wrote: > > Hello all, > > I was thinking if such scenario could be possible: > > 1 - Export/import a ZFS filesystem in two solaris servers. > > 2 - Export that filesystem (NFS). > > 3 - Mount that filesystem on clients in two different mount points (just > to authenticate in both servers/UDP). > > 4a - Use some kind of "man-in-the middle" to auto-balance the > connections (the same IP on servers) > > or > > 4b - Use different IP''s and balance through DNS. > > After "little" problems with this initial (mount) setup, the NFS > conversation should work without problems, without stale file handle issues, > right? > > > > Thinking in such configuration i''m assuming some "concepts", that i''m > asking your corrections if i''m wrong: > > 1) Using ZFS send/receive i will have the SAME filesystem across > machines. > > 2) Using ZFS, the file handles in NFS protocol are not beeing made > using disk luns, major/minor numbers, etc... > > I''m not willing to implement such solution, but what i really want to > know is IF that configuration is possible, and IF i''m not wrong about my > assumptions. > > > > Thanks a lot for your time! > > > > Leal. > > > > > I''m very confused. What are you trying to do? > We already have Solaris Cluster which manages ZFS mounts, exports, > and logical IP addresses. > -- richard > > _______________________________________________ > zfs-discuss mailing list > zfs-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/zfs-discussHe wants to mount the ZFS filesystem (I''m assuming off of a backend SAN storage array) to two heads, then round-robin NFS connections between the heads to essentially *double* the throughput.>From what he''s described here, it isn''t going to work. ZFS isn''t a clusterfilesystem. You can''t currently mount it from two systems at the same time. That is, unless something big has changed recently... I haven''t been doing a very good job of keeping up lately, but I think it''d be hard to miss an announcement like that! --Tim -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20080313/bd9e621b/attachment.html>
Tim wrote:> > > > > He wants to mount the ZFS filesystem (I''m assuming off of a backend > SAN storage array) to two heads, then round-robin NFS connections > between the heads to essentially *double* the throughput.pNFS is the droid you are looking for.