4 replies to this topic

[Venkata Ragavan Shanmugam]

Hi,

Greetings to all !!

We have $U V5.0 running in a two node OpenVMS cluster (runs on both the ndoes).

During every maintenance, we shut down $U in one of the node and run the IU_RTS job from the other node.

Not any time we have had a complete reorganization of all FDL $U database files. ( Failing due to "File locked currently by another user" error)

Is this a normal behaviour?

The below files have not been successfully reorganized most of the times:

File name Significance
------------------------------------------------
U_FMSB50.001 Expected launches
U_FMCX50.001 Job monitor
U_FMFU50.001 Launch parentage
U_FMLP50.001 Operating events (status)
U_FMER50.001 Expected resource events

Before submitting the IU_RTS job, I delete all Uprocs that are currently running and waiting to be submitted in few minutes. Also the queues where $U jobs is submitted are stopped to avoid new submissions to run.

Is there a "best practice" for a successful reorganization of all FDL files? What is the expected behaviour out of $U Maintenance?

Experts please suggest

Cheers
Venkata

[men]

Hi Venkata,

some of the Dollar Universe data files are locked by a process.

Have you tried checking which process (using command sh def /fil)?

I assume that if you have killed the Dollar Universe batch jobs which were running, and that the engines are correctly stopped, the only option left is that some interfaces are still up (e.g.: Motif GUI or character interface). I saw this happen at some of Orsyp's customers.

Also in V500, make sure you have the latest IU_RTS, which is cluster wide as long as you have correctly set the logical which lists the members of the cluster for Dollar Universe. But even with that version of the Uproc, data files must not be locked otherwise the convert will fail.

Michel

[Venkata Ragavan Shanmugam]

Hi Michael,

Yes, I did kill all the Dollar universe jobs in the queue as well stopped the queues to prevent future launches. ( since $U is running in one of the node)

However, I am not sure if we did close the GUI. During next maintenance, we will be vigile to to close the GUI.

I am not sure about what do you mean by "latest IU_RTS"

We have updated the system with latest patch level cluster wide.

Cheers
Venkata

[men]

Hi Venkata,

if you have installed the SPRE0025 you have the latest version of the IU_RTS in the Application Area

Check the GUI and keep us posted

Michel

[men]

One of our customers (a Belgian company with outsoucing in India) had the same problem: after careful check, it appeared that some interfaces were still active (GUI, character interface) including remote instances which were locking the files of the instance on the VMS cluster.

Once all these interfaces - local and remote - were stopped, the reorganization run absolutely fine

Michel