[Linux-HA] Integrating iSCSI service to DRBD for iSCSI failover

[Andrew Beekhof]

On Tue, Oct 27, 2009 at 11:28 AM, sajan agrawal <boyin_net at yahoo.com> wrote:
> Hi
>
> I am a naive user of pacemaker and have been trying to achieve iSCSI failover on Pacemaker. I have tried couple of things though, which logically looks fine. But, I am not able to get any progress in this. So, I wanted to share this problem with you and help me sort out the problem.
>
> This is how my CIB file looks like :
>
> node $id="1c96735b-59c8-47d9-a2c6-cd2a2731d936" drbd_desk1 \
>        attributes standby="off"
> node $id="b8c4fa16-df49-45b7-a850-cb9f65af036e" drbd_desk2 \
>        attributes standby="off"
> primitive drbd_test ocf:linbit:drbd \
>        params drbd_resource="r0" \
>        op monitor interval="15s" \
>        meta target-role="started"
> primitive ip_drbd ocf:heartbeat:IPaddr2 \
>        params ip="192.168.30.33" nic="eth0" \
>        meta target-role="started"
> primitive iscsiLU ocf:heartbeat:iSCSILogicalUnit \
>        params lun="0" path="/dev/drbd0" target_iqn="iqn.2009-10.com.systems:Testme" \
>        meta target-role="started"
> primitive iscsiTG ocf:heartbeat:iSCSITarget \
>        params iqn="iqn.2009-10.com.systems:Testme" tid="1" \
>        meta target-role="started"
> group drbd_group ip_drbd iscsiTG iscsiLU
> ms ms_DRBD drbd_test \
>        meta master-max="1" master-node-max="1" clone-max="2" clone-node-max="1" notify="true"
> property $id="cib-bootstrap-options" \
>        dc-version="1.0.5-462f1569a43740667daf7b0f6b521742e9eb8fa7" \
>        cluster-infrastructure="openais" \
>        no-quorum-policy="ignore" \
>        stonith-enabled="false" \
>        default-resource-stickiness="1000" \
>        expected-quorum-votes="2" \
>        last-lrm-refresh="1256633765"
>
>
> crm_mon shows perfectly fine when both the nodes are up and shows all services are running.
>
> I powered off the master (desk2 at present), desk1 becomes master fine, but the iscsiLU resource is seen 'Stopped' at this machine. I think this service is not migrated here.
>
> Can anyone guide me thtough this ?
>

Not without logs or the output from 'cibadmin -Q' while the cluster is
in the state you describe.