Problem with LVM on iSCSI after reboot

    • Problem with LVM on iSCSI after reboot

      Hi @ all,

      i played around with the iSCSI and a xenserver (6.2). On OMV there is a raid5 which is added to iscsi-plugin. Xenserver could perfectly mount it and created it's storage on it (LVM). After reboot the iSCSI storage is not longer accessible - local lvm is breaking iSCSI and blocks access. Is there any way to get back access to the raid5 via iSCSI ? A temp solution is ok ! There are no needed data on the raid5 but it would be nice to know to get them back if this happens...

      Thx
    • Re: Problem with LVM on iSCSI after reboot

      Hi all

      I have exactly the same problem :

      installed OMV 0.5.49 and iscsi plugin, configured a target with a disk in block mode, mounted it on Xenserver 6.2, created VMs, installed Oses, everything went fine.

      but after the reboot of my lab, the iscsi target has no lun anymore.

      the logs :

      Jun 30 00:41:36 ax2omv kernel: [ 863.930352] iSCSI Enterprise Target Software - version 1.4.20.2
      Jun 30 00:41:36 ax2omv kernel: [ 863.930449] iscsi_trgt: Registered io type fileio
      Jun 30 00:41:36 ax2omv kernel: [ 863.930451] iscsi_trgt: Registered io type blockio
      Jun 30 00:41:36 ax2omv kernel: [ 863.930453] iscsi_trgt: Registered io type nullio
      Jun 30 00:41:36 ax2omv kernel: [ 863.934512] iscsi_trgt: blockio_open_path(168) Can't open device /dev/sdb, error -16
      Jun 30 00:41:36 ax2omv kernel: [ 863.935002] iscsi_trgt: blockio_attach(295) Error attaching Lun 0 to Target iqn.2014-06.local.ax2omv:iqn.2014-06.org.openmediavault.tgt1
      Jun 30 00:41:36 ax2omv ietd: unable to create logical unit 0 in target 1: 16

      sdb disk is ok on the physical disks page, everything's seems ok but iscsi_trgt is unable to use the disk.

      thanks for your help
    • Hi,
      Had this issue today, i have sorted it out changing /etc/lvm/lvm.conf

      Look for 'filter' stanzas. i have excluded sde disk from LVM scan during bootup, mine config looks like this:

      Source Code

      1. filter = [ "r|/dev/sde|" ]
      2. # By default we accept every block device:
      3. # filter = [ "a/.*/" ]

      Better use some id, like:

      Source Code

      1. filter = [ "r|/dev/disk/by-path/pci-0000:00:11.0-scsi-3:0:0:0|" ]



      And before this change, vgscan worked like that:

      Source Code

      1. bash:# vgscan
      2. Reading all physical volumes. This may take a while...
      3. Found volume group "xen" using metadata type lvm2



      After the change:

      Source Code

      1. bash:# vgscan
      2. Reading all physical volumes. This may take a while...
      3. No volume groups found