Top realtime monitor, memory faulting, gs160

[Thomas Wieckowski]

Hello, Our Top realtime monitor has started memory faulting on us. The only thing we've done recently, is repartition our gs160 from 2 partions back into 1. It's now a 4 qbb,32g. memory single partition. It is also in a cluster with 4 other nodes, 3 es45's and another gs160. TOP does'nt fault on these other systems, just the "re-partitioned" gs160. Top is currently at 3.5beta12 (which I think is the latest for Tru64). Any thoughts, Any response would greatly be appreciated. Thank you. Ps. It will run for a period of time, before faulting.

[Dave Bechtold]

Hi Thomas,

I'm not sure why this would be faulting on only the one system in the cluster. Assuming that it's the same executable for all members - same path to the top program.

I know top uses symbols built in to the kernel to dig out some of the data, was the kernel re-built after re-configuring the partitions? If so, did it build OK?

Is /vmunix on that member still a CDSL?
The file /vmunix should be a soft link to /cluster/members/{memb}/boot_partition/vmunix .

Dave Bechtold

[Thomas Wieckowski]

Thanks Dave for the reply. The /vmunix has the correct path to the cluster member.
I also, just for kicks re-compiled top and this compile also faults. And I did re-verify that it runs fine on the other members. We did'nt have to re-compile the kernel whenever we combined the 2 partitions back to 1. just did the hardware reconfiguration setting hp-count etc. Appreciate any thoughts you may have on this.
regards
Tom