IBM HACMP and Progress DB

[balebob]

Can Progress installation directory (DLC), all production db (.db files), parameter file (.pf), db start and stop scripts and all database extents installed on the shared drives in HACMP cluster?

This is my dilemma. We have two HACMP nodes and all our production dbs including Progress Installation dir (DLC) are installed on the SAN SSA drives. We have an HACMP start and stop scripts that can bring up and shutdown all the dbs as long as there is a connection between the server and the SAN box. I am wondering what will happen to all the db if the SAN box cannot see the server at all. Will all the db corrupted? If this is the case, is it helps to move the DLC including all production db files back to the server, and just leave the extents on the SAN drives?

Thanks,
Bob.

 

[ddavis]

I don't think this is what you want to do. In what case will the SAN not see the server? When the primary node has failed over?

When the server becomes disconnected from the SAN sub system, your db has crashed. When the failover node mounts the disk system and attempts to start the progress services, Progress will use crash recovery and the bi notes to recover the db right up to the point the first server and the db became disconnected.

If you move the main prod db files back to the primary node, then when you attempt to start up from the secondary node, you will not have a valid db I think.

I have only a little experience with this, so if I misunderstood the intention, someone will correct me I'm sure.

Darrell Davis

 

[balebob]

Thanks Darrell! Once I thought more about it, you are absolutely right.

When the db failover to a standby node, it will mount the file systems on the shared drives...then start the db. Since the db is crashed on the previous node, the standby system will do an automatic crash recovery...Therefore moving db files to local server will not be a good idea.

Any additional comments are appreciated.
Bob.