topic Re: OpenVMS Alpha V8.3 MONITOR problem in cluster in Operating System - OpenVMS

OpenVMS Alpha V8.3 MONITOR problem in cluster

Jim Geier_1 — Thu, 22 Nov 2007 14:18:20 GMT

We are running OpenVMS Alpha V8.3 in a cluster of three ES45 and one DS10 systems. The ES45s all run from a common system disk. This problem also manifests itself from a cluster of an ES40 and a DS20e also running OpenVMS Alpha V8.3. When I issue the following commands, I get an error:

$ monitor cluster
^C
MONITOR> set default/node=hcpb ! different node
MONITOR> monitor process/topcpu
%NONAME-W-NOMSG, Message number 00000000
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000000
0041, PC=FFFFFFFF80E94978, PS=0000001B

$ monitor
MONITOR> set default/node=hcpa ! this node
MONITOR> monitor process/topcpu
%NONAME-W-NOMSG, Message number 00000000
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000000000
0041, PC=FFFFFFFF80E94978, PS=0000001B

This worked for me prior to upgrading from VMS 7.3-2 to VMS 8.3. This problem also is seen when running monitor on a standalone node running OpenVMS Alpha 8.2. And installation of the latest Monitor patch kit does not change the behavior. Is this a bug in Monitor, or something else not being set up correctly?

Re: OpenVMS Alpha V8.3 MONITOR problem in cluster

Hein van den Heuvel — Thu, 22 Nov 2007 17:29:55 GMT

Please raise this issue with HP support.
This is certainly a bug.
No standard tool should ever cause an ACCVIO problem.
In the mean time, just SET HOST and monitor localy?

Hein.

Re: OpenVMS Alpha V8.3 MONITOR problem in cluster

Hoff — Thu, 22 Nov 2007 20:07:50 GMT

...Or load AMDS or Availability Manager, and monitor host activity that way.

Re: OpenVMS Alpha V8.3 MONITOR problem in cluster

Karl Rohwedder — Fri, 23 Nov 2007 01:56:56 GMT

Just confirmed that the error also occurs on V8.2. A monitor/node=nnn (with nnn being the local node) surprisingly works in opposite to set default.

regards Kalle

Re: OpenVMS Alpha V8.3 MONITOR problem in cluster

Volker Halle — Fri, 23 Nov 2007 02:06:28 GMT

Jim,

I can reproduce this on an OpenVMS Alpha V8.2 system, also on Alpha V8.3 and I64 V8.2. It works fine on VAX V6.2 ;-) Did you install the recent V83A_MONTOR-V0100 patch ? Nothing in the release notes, but just give it a try, before logging a call.

The next-best workaround would be to use:

$ MONI PROC/TOPCPU/NODE=node

The MONITOR> SET DEF command seems to be a seldom-used code path. Log a low-priority call with HP.

Volker.

Re: OpenVMS Alpha V8.3 MONITOR problem in cluster

Jim Geier_1 — Fri, 23 Nov 2007 18:12:58 GMT

I have installed the recent VMS83A_MONTOR-V0100 patch kit. I have been using the SET DEF command in Monitor for many years -- I certainly can Set Host and monitor from a process on the local node. It is rather convenient to use the SET DEF in Monitor at times. For example, if I use MONITOR CLUSTER, then see a large CPU usage on another system, it is faster to press control-C, and issue the commands SET DEF nnnn and MON PROC/TOPC to get a quick view of what is happening.

I also use AMDS, but because the data center is 35 miles away across a couple of routers, I have to run it using an X terminal to the Java-based AM on an AlphaServer in the data center, and the performance of the Java AM is painful to use.

Re: OpenVMS Alpha V8.3 MONITOR problem in cluster

Jim Geier_1 — Mon, 26 Nov 2007 13:52:44 GMT

The case has been submitted to HP OpenVMS support, case number 3600132603-361. The response a request for additional information and that it would have to be elevated to OpenVMS Engineering.