Hi folks, I would like to try out iscsi for virtual block devices. This would be very interesting especially for a migration at run time.

On Fri, Jul 09, 2010 at 02:15:58PM +0200, Harald Dunkel wrote: > Hi folks, > > I would like to try out iscsi for virtual block > devices. This would be very interesting especially > for a migration at run time. > > Problem is: It still doesn't work :-(. libvirt can > login to the iscsi pool, but using virsh vol-list I > just get an empty list, even though /dev/disk/by-path > shows the expected block devices. > > There was an old thread on this list describing the > procedure: > > http://www.mail-archive.com/libvir-list@redhat.com/msg06320.html > > Are there any news about this? > > > I am using libvirt 0.8.1, iet 1.4.20.1 and open-iscsi > 2.0.871.3, as included in Debian Testing. Hi Harri, It works fine on my Fedora boxes across several versions, so there must be some difference in where information is located in Debian and Fedora. I'm having a quick look. Are you familiar enough with C that you could have a look through the libvirt code and try to track it down?

Hi Dave, Sorry for the delay. See below On 07/12/10 20:23, Dave Allan wrote: > > Hi Harri, > > I spun up a Debian testing box with the versions of libvirt, iet& > open-iscsi that you're using, and it works fine for me. My pool xml > is: > > <pool type="iscsi"> > <name>iscsitest</name> > <source> > <host name="localhost"/> > <device path="iqn.2001-04.com.example:foobarbaz"/> > </source> > <target> > <path>/dev/disk/by-path</path> > </target> > </pool> > > Gives the 3 configured LUs as volumes: > > Name Path > ----------------------------------------- > 3.0.0.0 /dev/disk/by-path/ip-127.0.0.1:3260-iscsi-iqn.2001-04.com.example:foobarbaz-lun-0 > 3.0.0.1 /dev/disk/by-path/ip-127.0.0.1:3260-iscsi-iqn.2001-04.com.example:foobarbaz-lun-1 > 3.0.0.2 /dev/disk/by-path/ip-127.0.0.1:3260-iscsi-iqn.2001-04.com.example:foobarbaz-lun-2 > In my environment the new volume is not listed by virsh vol-list, even though /dev/disk/by-path shows the iscsi block devices: # virsh vol-list --pool iscsi.dpcl064 Name Path ----------------------------------------- # ls -al /dev/disk/by-path/ total 0 drwxr-xr-x 2 root root 320 Jul 15 10:56 . drwxr-xr-x 6 root root 120 Jul 15 2010 .. lrwxrwxrwx 1 root root 9 Jul 15 10:56 ip-172.19.97.183:3260-iscsi-iqn.2010-07.de.aixigo.ac.dpcl064:storage-lun-0 -> ../../sdd lrwxrwxrwx 1 root root 10 Jul 15 10:56 ip-172.19.97.183:3260-iscsi-iqn.2010-07.de.aixigo.ac.dpcl064:storage-lun-0-part1 -> ../../sdd1 lrwxrwxrwx 1 root root 10 Jul 15 10:56 ip-172.19.97.183:3260-iscsi-iqn.2010-07.de.aixigo.ac.dpcl064:storage-lun-0-part2 -> ../../sdd2 lrwxrwxrwx 1 root root 9 Jul 15 09:32 pci-0000:00:03.0-usb-0:2:1.0-scsi-0:0:0:0 -> ../../sdc lrwxrwxrwx 1 root root 9 Jul 15 09:33 pci-0000:01:0e.0-scsi-0:0:0:0 -> ../../sda : : The virt-manager GUI knows about the defined iscsi volume, so I would guess this is just a minor bug in virsh.

I did an upgrade to libvirt-0.8.2 today. Using this new version the problem seems to be gone. I would guess you started your test with 0.8.2, too? Many thanx for your help. Regards Harri