[Pacemaker] Configuring ocf:heartbeat:Filesystem monitoring

Terry Johnson terry.johnson at scribendi.com
Wed Apr 11 13:22:38 UTC 2012


Hi -

I'm working on a two node cluster on CentOS 6.2, using iscsi shared storage, and I'm having difficulty detecting a lost connection to the storage.

Maybe this is a n00b issue, but I understand that monitor depth="20" in ocf:heartbeat:Filesystem is supposed to create a test file and check whether it can write to that file, and declare the resource failed if the filesystem goes read-only. The test file does not get created, and I can't see where any errors might be being logged to.

I've tested this configuration by disabling a switch port. The second node picks up the services correctly, but the first node keeps on running them too, and does not notice that it no longer has a writeable filesystem.  If the port is reconnected, they both have the same ext3 filesystem mounted at once, which makes a fine mess.

Here's my current configuration. Have I missed some vital detail?  Should I be brining the iscsi connection into pacemaker too?

node dba
node dbb
primitive ClusterIP ocf:heartbeat:IPaddr2 \
        params ip="10.232.169.130" cidr_netmask="24" \
        op monitor interval="30s" nic="eth0" \
        meta target-role="Started"
primitive p_fs_mysql ocf:heartbeat:Filesystem \
        params device="/dev/sdb1" directory="/mnt/mysql" fstype="ext3" options="noatime,nodiratime,noexec" \
        op start interval="0" timeout="60" \
        op stop interval="0" timeout="240" \
        op monitor interval="30s" depth="20" \
        meta target-role="Started"
primitive p_mysql lsb:mysql \
        op start interval="0" timeout="60s" \
        op stop interval="0" timeout="60s" \
        op monitor interval="15s" \
        meta target-role="Started"
group g_mysql p_fs_mysql ClusterIP p_mysql \
        meta target-role="Started"
property $id="cib-bootstrap-options" \
        dc-version="1.1.6-3.el6-a02c0f19a00c1eb2527ad38f146ebc0834814558" \
        cluster-infrastructure="openais" \
        expected-quorum-votes="2" \
        stonith-enabled="false" \
        no-quorum-policy="ignore"
rsc_defaults $id="rsc-options" \
        resource-stickiness="200"

corosync (1.4.1-4.el6_2.1)
pacemaker (1.1.6-3.el6)

Any suggestions are appreciated.

Terry.

Follow us on Twitter:
http://www.twitter.com/Scribendi_Inc
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20120411/47ea66e6/attachment-0003.html>


More information about the Pacemaker mailing list