Process Timeout

[Sigi]

Hi

We made a process resource with a script behind it.
The script do nothing other than proof if a service(not a windows service) is running or not.
This service need longer than 150 seconds to start, but the resource is failiing everytime after 120 seconds.
Where is this 120 Second timeout setting?

any suggestions?

Thanks
Daniel

[g_lee · Accepted solution]

Disclaimer: Not a Windows specialist, however overall concept should still hold.

VCS 5.0 Administrator's Guide:
http://sfdoccentral.symantec.com/sf/5.0/windows/pdf/VCS_Admin.pdf

Under Resource type attributes, see:
MonitorTimeout (default 60 seconds)
MonitorInterval (default 60 seconds)

What's probably happening is the online program runs and completes succcessfully, then after 60 seconds (MonitorInterval) the first probe runs to check if it's up/running. As the process is not running yet, it times out after 60 seconds (MonitorTimeout), and since it's still not up it fails (hence the 120 second interval).

If the process takes 150 seconds to start, but once it's running you're happy to monitor every 60 seconds (default), look at the OnlineWaitLimit attribute so it will wait X monitorintervals after Online before marking it as failed.
Note: this will affect all Process resources unless you override the attribute for this particular instance (see Admin Guide for details/instructions to do this).

Hope that helps.

[Wally_Heim]

Hi Sigi,

Is there a specific reason you are using the Process agent to do this?

On SFW-HA clustering we have a resource type called GenericSerivce which is designed to control Windows Services.  It has an DelayAfterOnline attribute that can be used to tune the agent to wait a given number of seconds before trying to monitor the service. 

We also have a resource type called ServiceMonitor that will simply monitor a service state.  It does not control the service in any way. But instead it will report as online if the service is running and offline/faulted if the service is not running. 

The items that g_lee mentioned will work fine for the Process agent or any other agent having similar problems.

Thanks,
Wally

[Sigi]

Hi

Thanks for the help, the setting "Monitor Timeout" was the problem.

The reason why i using this resourcetype is simple.
We have thirdparty application and this application is monitored with a windows service.
This service start some processes and monitor them. This service is started within 10sek but all the processes behind the service need several seconds more to start up.
There is a second service which depends on the first one's processes and it should start as soon as all processes are online and not the service.
And we discovered some Problems with DelayAfterOnline attribute, so we decided to make a timeout-process resource.

[g_lee]

Glad that you solved your problem - however it's nice to see you marked your own post as the solution despite requiring help/explanation provided by others - good to keep in mind if help is requested in future.

[Sigi]

well this was a misstake, sorry for that.