No failover relocation took place in the cluster configured with "spare" mode relocation type:
# shaman -c $CLUSTERNAME get-config |grep RESOURCE_RELOCATION_MODE RESOURCE_RELOCATION_MODE=spare
Shaman.log indicates that no nodes suitable for the 'spare' mode were found:
26-12-14 02:55:44.197 shaman-monitor(5119): Detected node md.3ff818985dfc4037 CRASH 26-12-14 02:55:44.265 shaman-monitor(5119): Executed script /usr/share/shaman/notify 26-12-14 02:55:44.267 shaman-monitor(5119): execute 'pstorage -c pcs revoke -I 219193' 26-12-14 02:55:44.632 shaman-monitor(5119): 'pstorage -c pcs revoke -I 219193' completed 26-12-14 02:55:44.633 shaman-monitor(5119): Trying to evacuate the node 'md.3ff818985dfc4037' using the 'spare' mode... 26-12-14 02:55:46.031 shaman-monitor(5119): No nodes suitable for the 'spare' mode were found in '.shaman' 26-12-14 02:55:46.031 shaman-monitor(5119): Failed to evacuate the node 'md.3ff818985dfc4037' using the 'spare' mode: 2
A spare node is a hardware node, which is registered in the cluster and has no resources stored on it. Thus, relocation will not work if the node is not empty or containing broken or duplicated resources.
Make sure that spare node is empty and doesn't have any broken resources for it:
Determine host id of "spare" node:
# cat /etc/pstorage/host_id
Check resources for the node by shaman:
# ls -la /pstorage/<$CLUSTER_NAME>/.shaman/md.<$HOST_ID>/resources/
Directory should be empty if there are no VEs present on this host. If it is not empty while server doesn't have VEs registered - remove incorrect resources.
For more detailed steps refer to:
117032 How to move shaman resource out of broken state
123693 shaman: Failed to read XML model: syntax error
117944 Virtual environments did not relocate after a node crash - Shaman misconfiguration
119207 Tuning Parallels Cloud Storage High Availability Cluster configuration.
119300 How can I track the movement of resources in a high availability cluster on Parallels Cloud Storage after a node crash?