[ClusterLabs] glusterfs 2 active primitives after reboot

Vladislav Bogdanov bubble at hoster-ok.com
Sun May 3 11:56:05 EDT 2015

03.05.2015 15:06:40 GMT+03:00, lukas <lukas.kostyan at gmail.com> wrote:
>i am testing a Debian-Wheezy cluster with corosync 1.4.2 and pacamaker 
>1.1.7. On the cluster glusterfs is used
>as storage (replicated) which is mounted to the webservers root 
>directory. To test the down time ping is used.
>Lets assume node1 fails. There is no problem and virtually no 
>interruption during change of one cl_IP instance
>on node2 (both run on node2). If node1 fails node2 restarts node1 with 
>stonith. But as soon as node1 is back
>online again all apache and IP instances stop and start again. This is 
>where the interruption happens which is
>around 5seconds. Is this behaviour to be expected? I thought the 
>instances running on node2 shouldnt stop at
>all when node1 gets back online again.
>I guess it has something to do with the primitive active on both nodes 
>(see below)? Why is the
>primitive p_apache started on the failed node?
>The CIB looks as follows:
>node vm-1 \
>         attributes standby="off"
>node vm-2 \
>         attributes standby="off"
>primitive IP ocf:heartbeat:IPaddr2 \
>         params ip="" nic="eth0" 
>clusterip_hash="sourceip-sourceport" \
>         op monitor interval="10s"
>primitive p_apache ocf:heartbeat:apache \
>         params configfile="/etc/apache2/apache2.conf" 
>statusurl="http://localhost/server-status" \
>         op monitor interval="60" timeout="20" \
>         op start interval="0" timeout="40s" start-delay="0" \
>         meta is-managed="true"
>primitive p_fence_N1 stonith:external/libvirt \
>         params hostlist="vm-1:N1" 
>pcmk_reboot_action="reboot" \
>         op monitor interval="60"
>primitive p_fence_N2 stonith:external/libvirt \
>         params hostlist="vm-2:N2" 
>pcmk_reboot_action="reboot" \
>         op monitor interval="60"
>primitive p_gluster_mnt ocf:heartbeat:Filesystem \
>         params device="localhost:/gvolrep" directory="/var/www/html" 
>fstype="glusterfs" \
>         op monitor interval="10"
>primitive p_glusterd ocf:glusterfs:glusterd \
>         op monitor interval="30"
>primitive p_volume_gluster ocf:glusterfs:volume \
>         params volname="gvolrep" \
>         op monitor interval="10"
>clone cl_IP IP \
>         meta globally-unique="true" clone-max="2" clone-node-max="2" 
>interleave="true" \
>         params resource-stickiness="0"
>clone cl_apache p_apache \
>         meta globally-unique="true" target-role="Started"
>clone cl_gluster_mnt p_gluster_mnt \
>         meta globally-unique="true" interleave="true" 
>clone cl_glusterd p_glusterd \
>         meta interleave="true"
>clone cl_glustervol p_volume_gluster \
>         meta interleave="true"
>location l_fence_N1 p_fence_N1 -inf: vm-1
>location l_fence_N2 p_fence_N2 -inf: vm-2
>colocation apache-with-ip inf: cl_apache cl_IP
>colocation c_apache_gluster inf: cl_apache cl_gluster_mnt
>colocation c_gluster_mnt_with_apache inf: cl_gluster_mnt cl_apache
>colocation c_glusterd_with_glustervol inf: cl_glusterd cl_glustervol
>colocation c_glustervol_with_gluster_mnt inf: cl_glustervol
>order o_apacher_after_mnt inf: cl_gluster_mnt cl_apache
>order o_gluster inf: cl_glusterd cl_glustervol cl_gluster_mnt cl_apache
>property $id="cib-bootstrap-options" \
>         dc-version="1.1.7-ee0730e13d124c3d58f00016c3376a1de5323cff" \
>         cluster-infrastructure="openais" \
>         expected-quorum-votes="2" \
>         stonith-enabled="true" \
>         no-quorum-policy="ignore" \
>         last-lrm-refresh="1430571170" \
>         default-resource-stickiness="100"
>op_defaults $id="op-options" \
>         timeout="240s"
>root at vm-2:~# tail -f /var/log/corosync.log | grep ERROR
>May 03 13:38:14 vm-2 pengine: [3285]: ERROR: native_create_actions: 
>Resource p_apache:1 (ocf::apache) is active on 2 nodes attempting
>May 03 13:38:14 vm-2 pengine: [3285]: ERROR: process_pe_message: 
>Transition 30: ERRORs found during PE processing. PEngine Input stored 
>in: /var/lib/pengine/pe-error-0.bz2
>Users mailing list: Users at clusterlabs.org
>Project Home: http://www.clusterlabs.org
>Getting started:
>Bugs: http://bugs.clusterlabs.org

Do you have apache started by init scripts?

More information about the Users mailing list