File Events, Clustered Servers & Agents

[Dave Martin]

So, we're upgrading an application environment from its simple little single Unix box to a three-headed clustered servers (still Unix) which share the application's shared file system via a SAN.

At present, I have a variety of file events watching for periodic file drops which instantiate job insertions based on those files.

How best do I go about making sure I can still see files if one of the heads of the application server dies? File events & agent lists don't go together (I'm currently running 5.3.1, I should point out)... so, do I have to set up another two file events per file? (Yuck.) Or is there another, simpler answer?

Thanks,

Dave Martin              

[Michelle Morris]

(In the windows world), a cluster responds to the same machine name no matter which head is active.  So \\cluster01\filename1 will reach the file, no matter which head is active.  Is this not the behavior in the UNIX world?

I had:

cluster01 which was made up of server01 and server02.  The Windows Cluster service would point the cluster01 name to the active node so the file events would just use the cluster01 name in the share.

Michelle