E.7. Common Behaviors: Cluster Service Daemons

Service status check fails

Common Causes: User script reports error, clurmtabd not running on NFS service, smbd and nmbd not running for a service with a Samba share.

Test Case: Create a service with an init script that returns a status output of 1.

Expected Behavior: Service restarts on the current owner.

Verification: Consult system logs for a service restart event. The restarts field of the service's status information should be incremented.

A member fails to start services

Common Causes: User script returns error due to file system errors.

Test Case: Create a service with a user script which returns 1 for the start phase only on one member. Attempt to enable the service on this member.

Expected Behavior: Service is stopped and started on another member, provided the services stops successfully.

Service start fails on all members

Common Causes: User script returns error, file system errors.

Test Case: Create a service with a user script which returns 1 for the start phase on all members.

Expected Behavior: Service is placed into the disabled state.

Verification: Run clustat and verify that the service is in the disabled state.

Service stop fails on a member

Common Causes: User script returns error; file system can not be unmounted.

Test Case: Create a service script which returns 1 for the stop phase.

Expected Behavior: Service is placed into the failed state. At this point, the administrator must intervene to determine the cause of the failure and the appropriate course of action. The service must then be disabled before it can be enabled.

Verification: Run clustat and verify that the service has been placed in to the failed state.