Hello, I implemented iscsi target support over sheepdog: git://git.kernel.org/pub/scm/linux/kernel/git/tomo/tgt.git sheepdog iSCSI target daemon (tgt) can provide qemu sheepdog image files as SCSI logical units to iSCSI initiators. You need to run tgt daemon on a host that sheep daemon runs on. Shortly, what I try to do is converting a sheepdog storage cluster into an iSCSI storage cluster (something like HP's lefthand). I'm still thinking about how to handle node failure. That is, tgt daemons needs to get the node failure information from sheep daemon and one node in the cluster need to take over the volume that the failed tgt daemon provided. Currently, cow or snapshot vdi isn't supported. So you need to create a qemu sheepdog image with the following patch: http://lists.wpkg.org/pipermail/sheepdog/2010-October/000706.html Anyone is interested? the following example is for people that already use tgt: fujita at rose:~$ ./git/sheepdog/collie/collie vm list Name |Vdi size |Allocated| Shared | Status ----------------+---------+---------+---------+------------ test14 | 512 MB| 512 MB| 0.0 MB| not running You can use sheepdog vdi name as a path like this: tgtadm --op new --mode logicalunit --tid 1 --lun 1 -b test14 --bstype sheepdog root at rose:~/git/tgt# tgtadm --op show --mode target Target 1: iqn.2001-04.com.example:storage.sr.rose.sys1.xyz System information: Driver: iscsi State: ready I_T nexus information: I_T nexus: 1 Initiator: iqn.2005-03.org.open-iscsi:e4441d4e4616 Connection: 0 IP Address: 10.76.0.30 LUN information: LUN: 0 Type: controller SCSI ID: IET 00010000 SCSI SN: beaf10 Size: 0 MB Online: Yes Removable media: No Readonly: No Backing store type: null Backing store path: None Backing store flags: LUN: 1 Type: disk SCSI ID: IET 00010001 SCSI SN: beaf11 Size: 537 MB Online: Yes Removable media: No Readonly: No Backing store type: sheepdog Backing store path: test14 Backing store flags: Account information: ACL information: ALL |