topic UCX memory problem on VMS in Operating System - OpenVMS

UCX memory problem on VMS

Sunil_uk — Wed, 23 Feb 2005 11:59:18 GMT

I am maintaining multiple VAX servers and workstations connected on WAN....ocassionaly
most of the worksations gives UCX memory failure when they load picture files ro open multiple window session...

VMS ver is 7.1 and ucx ver is 4.2

Is it a VMS bug?

Re: UCX memory problem on VMS

Volker Halle — Wed, 23 Feb 2005 13:41:48 GMT

Sunil,

if you would like help from this forum, please be very specific about the operation/command performed and the exact error message returned.

The commands UCX SHOW COMM and UCX SHOW COMM/MEM display information about UCX-specific memory.

SHOW MEM/POOL/FULL displays information about the OpenVMS nonpaged pool.

Consider to include the output of the above commands in a text attachment to your next reply. Capture this memory information at a time where everything is working and at the time of (or immediately after) the error.

Volker.

Re: UCX memory problem on VMS

Sunil_uk — Fri, 25 Feb 2005 10:06:06 GMT

I have attached the output (ucx show comm & sh ucx comm/mem) from a running system...
But I am not able to get the out put after the crash as the system goes down..

Waht is the trouble free version of TCPIP on VAXVMS 7.1

Re: UCX memory problem on VMS

Volker Halle — Fri, 25 Feb 2005 10:13:32 GMT

Sunil,

the counters look o.k. - the peaks have not reached the maximum values and there are no waits or drops.

You did not provide the SHOW MEM/POOL/FULL output for nonpaged pool.

If the system crashes (and you have got a valid SYSDUMP.DMP file), please provide the contents of the file CLUE$OUTPUT:CLUE$LAST_node.LIS (summary of system crash data). Getting the UCX memory data from a VAX dump might be hard...

You can look at the nonpaged pool data in the dump with:

$ ANAL/CRASH SYS$SYSTEM:SYSDUMP.DMP
SDA> SHOW POOL/NONPAGED/SUMM

and look at the amount of pool used.

Volker.

Re: UCX memory problem on VMS

Sunil_uk — Tue, 01 Mar 2005 05:40:51 GMT

Hi,

i have attached the clue file (as per your advice)..please examnine and suggest me if you get some idea..

thnking you

Sunil

Re: UCX memory problem on VMS

Volker Halle — Tue, 01 Mar 2005 07:21:52 GMT

Sunil,

your OpenVMS VAX V7.1 system has incurred a HALT restart crash. The HALT-PC seems to be A67EB, a P0 address in current image DKA100:[AGMS.MCSYS.][IMG]SYSIPC.EXE;56 - the PSL shows: Kernel Mode, IPL 0 (i.e. process context). You can examine the failing instruction with SDA:

$ ANAL/CRASH SYS$SSTEM:SYSDUMP.DMP
SDA> EXA/INS A67EB

SDA> EXA/INS A67EB-20;30 ! instr-stream

This type of crash happens, when a HALT instruction (HALT = 0x00) is executed in kernel mode. OpenVMS itself does not use HALT instructions to signal problems, it will always uses BUGW or BUGL to initiate a crash. Some application programs may use a HALT to signal a problem. A HALT can also happen, if the program counter (PC) ends up pointing into a data area instead of into an instruction stream.

There have been no nonpaged pool expansion problems before (PMS$GL_NPAGDYN* = 0).

If this system has crashed in the same way before, you can look at the crash history with:

$ CLUE:==$CLUE
$ CLUE/DISPLAY
CLUE_DISPLAY>DIR

If there have been HALT crashes before, look at their CANASTA parameters with CLUE_DISPLAY> n (n= crash # from column 1)
and look for the name of the current image.
Also look at CLUE_DISPLAY> SHOW STACK n for the failing PC/PSL for those HALT crashes.

Please report your findings here.

Volker.