Error Message

No errors appear, it is the time between the logging entries showing the delay and the scenario for this issue of a service that has a cluster agent starting before the cluster sends the online command.

1) SQL server onlines

03/21/2013         03:14:46 AM       INFORMATION     7036(0x40001b7c)        Service Control Manager                   
The SQL Server () service entered the running state.

 2) The Full Text search service will start outside cluster control.

 03/21/2013         03:14:46 AM       INFORMATION     7036(0x40001b7c)        Service Control Manager                   
The SQL Full-text Filter Daemon Launcher () service entered the running state.

3) Other Generic Services will start by the cluster

2013/03/21 03:14:46 VCS NOTICE V-16-1-10301 Initiating Online of Resource FD00__SQLServerAgent (Owner: , Group: _sg_fd) on System

4) The services can be seen to start after some delay

03/21/2013         03:21:03 AM       INFORMATION     7036(0x40001b7c)        Service Control Manager               
   The SQL Server Agent () service entered the running state.

and will be logged online by the cluster

2013/03/21 03:21:03 VCS INFO V-16-1-10298 Resource FD00__SQLServerAgent (Owner: , Group: _sg_fd) is online on (VCS initiated)

5) Dependent resources start to online, including Full Text serach that started after SQL server started, as indicated in step (2)

2013/03/21 03:21:03 VCS NOTICE V-16-1-10301 Initiating Online of Resource FD00__FulltextSearch (Owner: , Group: _sg_fd) on System

6) Full text Search resource completes online in the cluster after a delay even though the service has been long started:

2013/03/21 03:27:12 VCS INFO V-16-1-10299 Resource FD00__FulltextSearch (Owner: , Group: _sg_fd) is online on (Not initiated by VCS)

With the FireDrill attribute set to true, the monitoring functionality is changed and the cluster will ignore resources that come online outside of cluster control. If this is the case and the resource is configured for monitoring by the Intelligent Monitoring Framework (IMF) then events will be raised due to the service (in this case Full Text Search) changing state when it came online.  The IMF will invoke a monitor which will then be ignored due to the FireDrill attribute being set.  As the resource will not change state due to monitor being ignored, IMF monitoring will again respond to the state change and initiate another monitor. This cycle continues and delays the online of the resource.

This issue has been identified and the issue will be addressed in SFW-HA 6.1.

As a workaround to this issue in previous versions, set the IMF monitoring to 2 for GenericService resources that have the FireDrill attribute set to true. The IMF mode "2" indicating IMF monitoring for online resources only. Follow the steps provided below to set IMF mode as 2.

• Locate the Fire Drill service group
• Right click on the Generic Service agent, select properites
• look for "IMF", expand the selection and set Mode =2

Applies To

• Veritas Storage Foundation High Availability for Windows 6.0 (SFW-HA) with SRDF agent.
• Fire Drill service group online