clearing query about package configuration and control file.

[khilari]

Hi people, okay so I just want to double check on my understanding of the package configuration and control file.
The package configuration has basically all the policies in it that what would happen in a state of node or package failure(failover, node_fail_fast etc etc). And package control file has the packages (ip address, volume group name, file system name, its subnet etc).
Now, in the package configuration file the
Run_script
Run_script timeout no_timeout
Halt script
Halt_script timeout no_timeout

Q)The run_script and halt script should basically be pointing towards the control file path /etc/cmcluster/pkg1/pkg1.cntl

And in the package control file, how do u get the ip address for the package, is it the same as the node ip address, but isn’t it re locatable when the package fail overs, that is what makes it seamless for the users. Q).So, how do u determine the Ip address for the package.

Furthermore, the service commands, I believe means that which service of the application needs to be up and running. So, the package control file is basically enabling the volume group, checking it and then mounting it along with putting ip address to it, when it starts and then runs the service commands which are specified in the control file. Q) Now, why do some packages have monitoring scripts also.????.

[A. Clay Stephenson]

Monitors serve to test the health of a package. For example, an Oracle package might consist of many daemons. The death of any one of these daemons might kill the Oracle instance. The package needs some means of detecting this failure. In some cases, it makes more sense to attempt to restart the package on the same node since that will be much faster than a node switch. In general, my approach to testing Oracle, for example, is rather than looking for all the processes associated with this instance, is too simply execute a small query periodically. If that SQL executes with the expected results then it's a rather safe bet that all is well.

[melvyn burnard]

I suggest you read the Managing Serviceguard manual at http://docs.hp.com/en/ha and possibly also consider attending a Serviceguard training course to really gain more insight into the operation and use of Serviceguard