[Pacemaker] DRBD into standalone mode when failover

[Andrew Beekhof]

On 04/06/2013, at 11:35 PM, Weihua JIANG <weihua.jiang at gmail.com> wrote:

> Hi all,
> 
> I want a typical active/passive mode HA solution.
> 
> My Pacemaker configuration as below:
> 3 Nodes:
> node Lezbxh0jl
> node Ljn74rici
> node L472nxxdy (standby)
> The 3rd node L472nxxdy is only used for quorum election. So, I forced it to enter standby mode to avoid resource migrated to it.
> 
> The resource configuration is:
> primitive drbd_proj ocf:heartbeat:drbd \
>         params drbd_resource="r0" \
>         op monitor interval="15s"
> primitive fs_proj ocf:heartbeat:Filesystem \
>         op monitor interval="15s" \
>         params device="/dev/drbd0" directory="/proj/drbd" fstype="ext3" \
>         meta target-role="Started"
> primitive ip_proj ocf:heartbeat:IPaddr2 \
>         params ip="192.168.201.201" cidr_netmask="32" \
>         op monitor interval="15s" \
>         meta target-role="Started"
> primitive myserv lsb:proj-service \
>         op monitor interval="15s" \
>         meta target-role="Started" is-managed="true"
> ms ms_drbd_proj drbd_proj \
>         meta master-max="1" master-node-max="1" clone-max="2" clone-node-max="1" notify="true" target-role="Started"
> location master_prefer_on_active ms_drbd_proj 50: Ljn74rici
> colocation fs_ip_on_same_node inf: ip_proj fs_proj
> colocation fs_on_drbd inf: fs_proj ms_drbd_proj:Master
> colocation myserv_on_ip inf: myserv ip_proj
> order proj_after_drbd inf: ms_drbd_proj:promote fs_proj:start
> order ip_after_fs inf: fs_proj:start ip_proj:start
> order myserv_after_ip inf: ip_proj:start myserv:start
> property $id="cib-bootstrap-options" \
>         dc-version="1.1.6-3.el6-a02c0f19a00c1eb2527ad38f146ebc0834814558" \
>         cluster-infrastructure="openais" \
>         expected-quorum-votes="2" \
>         stonith-enabled="false"
> rsc_defaults $id="rsc-options" \
>         resource-stickiness="100"
> 
> Currently, I have following problems:
> 1. After failover, the DRBD partition enters into standalone mode. This means the failover can only happen once.
> 2. The failover is slow. I checked /var/log/messages and found there are many attempts to migrate resource to standby node.

That would be unusual.  Logs + full config?

> After these failed attempts, resource migrates to backup node. So, how to restrict certain resources to only certain node sets instead of using standby node solution?
> 3. In some situations, the failover will not happen. For example, if I block the active node's network via network congestion leads the service unavailable (up to 1 min), but HA doesn't switch.
> 
> Can you help to review my configuration and tell which parts shall be fixed?
> 
> Thanks
> Weihua
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org