queues+clustering

[Lawrence Czlapinski]

1. Multiple queue managers: We had a cluster with multiple queue manager files. One of the queue managers got confused and one of the queue manager files kept growing.
2. We find that using non-local disks for clusters works better if multiple systems have access to the system disk(s). Then if one system file disk server is brought down the cluster can still run. We often use the smaller local disks for additional pagefile and swapfile.
Lawrence

[Jan van den Ende]

Lawrence,

YES!!

And, for continuity, shadowing is your friend!

Proost.

Have one on me.

jpe

[nipun_2]

Thank you all for the reply.

This is my final analysis for our current setup.

There is a startup (not VMS_startup) file in the sys$common directory which contains generic ques (intialization and execution ques for current 2 nodes) as Joseph and many others mentioned.

So I simply edited the file to include the new node (EV68's execution ques and also setup the name in generic ques to include the execution ques of EV68).

So each node has its own execution queues to which computational tasks are assigned by the generic qeues.

The applications that run automatically gets assigned the tasks to the nodes that are free, (and/or up and running) through these queues.

As time goes I am understanding more and will discuss more but for now my work is accomplished. Please feel free to put in any comments.

Nipun

[nipun_2]

Thanks all of you.
Especially Uwe,Ian and Jan for answering all my posts in perfect details

Nipun