how to logoff processes

[Mike Ryder]

Hello,

Its me again!

How do you logoff a process. If I type "LOGOUT 35" to kill pib 35 then
LISTU shows "(logout pending)" and it will stay like that until I do a
unix kill

Thank you for your kind attention to my not-understandingness

Mike

[Ashley Chapman]

Hi Mike,

I've got to admit it does the same for me.  I didn't realise it was leaving so much stuff behind!

:LISTU
  User  Pid          Puid  Origin : Username
   805  3904               /dev/pts/1: root
    46  25920              : root
   115  10277              : root
   132  16518              : root
   133  16521              : root
   140  16543              : root
   155  2307               : root
   205  9946               : root
   226  14792              : root
   244  2554               : root
   246  9401               : root
   439  12597              : root
   377  20004              : root (logout pending)
   445  12879              : root
   624  18893              : root
   644  31911              : root
   647  32448              : root
   683  14908              : root
   712  19071              : root
   728  29606              : root
   946  5051               : root
   947  5061               : root
    41  14831              /dev/pts/0: root
*   56  15983              /dev/pts/2: root

I'm running GPL Openqm 2.3-4 on this laptop, which is running SUSe 9.3 (sorh of Redhat compatible)

[Martin Phillips]

Hi Mike,

I suspect that all of these problems are related. After an abort, QM should
tidy up properly. All locks should be released and you should remain in QM,
not fall out to the Linux shell.

The "logout pending" is because either the process actually doesn't exist or
it is in a loop where it cannot see the logout request.

If the process has actually vanished, the RECOVER.USERS command in the QMSYS
account should be able to perform some degree of recovery.

I have two opposing views of how we should approach this. On the one hand,
we need to identify the fault so some diagnostic work sounds good. On the
other hand, some problems fixed since 2.3-2 may be relevant and upgrading to
the latest release might be useful.

Martin Phillips, Ladybridge Systems

[Martin Phillips]

Hi Ashley,

> I've got to admit it does the same for me. I didn't realise it was
> leaving so much stuff behind!

This certainly should not happen. How did these users terminate their QM
sessions?

Martin Phillips
Ladybridge Systems
17b Coldstream Lane, Hardingstone, Northampton NN4 6DB
+44-(0)1604-709200

[Mike Ryder]

Martin,

Thanks RECOVER.USERS did the trick. The linux process had disappeared.
You are correct, the 3 threads that I started were interconnected - but
for this thread RECOVER.USERS was the command that I was seeking. As
for the reason why I ended up there - that is for another thread. ;-)

Rgds
Mike

[Ashley Chapman]

Hi Martin,

On 21/03/06, Martin Phillips <MartinP...@ladybridge.com> wrote:

Hi Ashley,

> I've got to admit it does the same for me.  I didn't realise it was
> leaving so much stuff behind!

This certainly should not happen. How did these users terminate their QM
sessions?

Various web session crashes, which I cannot identify the cause of.  As soon as I can identify the culprit I'll post it up.

--
Ashley Chapman
Billabong Services Ltd

[Martin Phillips]

Hi Mike/Ashley,

I think (hope?) that we may be seeing examples of the memory corruption that
we have been trying to track down. Previous reports suggested that it was to
do with the debugger. I am now beginning to think that it may be connected
with run time errors. There is an obvious link here - a user might be using
the debugger to track down the run time error.

Time for a bit of digging......

Martin Phillips, Ladybridge Systems

[Ashley Chapman]

On 21/03/06, Martin Phillips <MartinP...@ladybridge.com> wrote:

Hi Mike/Ashley,

I think (hope?) that we may be seeing examples of the memory corruption that
we have been trying to track down. Previous reports suggested that it was to
do with the debugger. I am now beginning to think that it may be connected
with run time errors. There is an obvious link here - a user might be using
the debugger to track down the run time error.

 

If it's any help to you, I have NEVER used the QM debugger.

[Mike Ryder]

Aah de-bugger - that is another part of the notunderstandingness - it
won't let me - but that is for another thread.

[Ashley Chapman]

As they say in Jamaica "de fault wit de system is debugger"

Sorry. Just could not resist that :-)

On 21/03/06, Mike Ryder < mg.r...@gmail.com> wrote:

Aah de-bugger - that is another part of the notunderstandingness - it

[Martin Phillips]

Hi Mike,

> Aah de-bugger - that is another part of the notunderstandingness
> - it won't let me - but that is for another thread.

It's just not your day, is it?.......

A couple of pointers in case this is actually something easy...

1. You need to compile a program in debug mode to be able to use the
debugger. You can do this with the DEBUGGING option to BASIC, by putting a
$DEBUG line at the top of the program source, or by use of a DEBUGGING line
in the $BASIC.OPTIONS record.

2. For the debugger to work best, you need to use QMConsole (Windows only)
or AccuTerm. If you are using AccuTerm, make sure that you have selected an
AccuTerm specific terminal type such as vt100-at. This activates features
that allow decent full screen debugging.

Martin Phillips, Ladybridge Systems

[Tracy Raines]

I've had this happen with the commercial windows version as well. I'm
not sure why, but when I had QM installed on another computer and I was
telneted to it, if I started my VPN client to connect to my system at
work, the QM sessions would stop responding. After disconnecting the
VPN client, I would log back in to QM and try to logout the other
sessions it would just say logout pending. They wouldn't go away until
I rebooted the system.