Conference virke::mrmemo

.0> Begging your help for the hereafter described problem :

.0> we very often find the MEMO server on a VAX ( at the "G�n�rale de
.0> Banque" in Brussels ) beeing in a "starting, connecting to MR and MEMO
.0> Gateway" state and remaining in this state forever.The current operation 
.0> is "logging on to MEMO Gateway".
.0> This seems to happen everytime the IBM on the other side has been stopped.

    What does "the IBM on the other side has been stopped" mean?
    How often does this happen?


.0> Taking a SNA trace on the DEMSA ( tracing the LU used by MRMEMO ) doesn't
.0> help that much as nothing happens over this LU.
 
    Is it possible for you to provide me with an SNA trace (i.e. one
    continuous trace) that starts when there is a normal working connection
    and lasts through the events:

      o IBM is stopped
      o IBM is restarted
      o MRMEMO hangs when trying to reconnect
      o The MRMEMO server is SHUT down and reSTARTed and the conection works
        again


.0> The server process on the VAX is in a LEF state.

    It is probably waiting for the connect to MEMO to complete (a call to
    the SNA API - SNALU0$REQUEST_CONNECT_W).


.0> A look at the MRMEMO1.LOG file shows that the IBM has been gone but does 
.0> not provide much more informations.

    If you can produce the SNA trace described above, I would also like to
    have the MRMEMO1.LOG from the session to match the trace against.


.0> Usually, a SHUT / START of the server "fixes" the problem.

.0> Is this a normal behaviour ? Or is there a way to fix this so that the 
.0> communication can restart automatically ?

    This is not how it should work. The connection should be reestablished
    automatically when the IBM comes up again.


    What are the version numbers of VMS, MRMEMO and DECnet/SNA API?


    Sorry for the delay in answering. We've been away from the office for a
    week.

    If there is some info you don't want to post here (such as a big SNA
    trace), you can mail it (or a pointer) to VMSmail STKHLM::OLSSON.

    Anders

So glad to have a correspondent !!!

Here we go ...

.1>    What does "the IBM on the other side has been stopped" mean?
.1>    How often does this happen?

	The IBM is usually stopped ( IPL ) every week during the night 
	 from friday to saterday.
	We so find the memo gateway in a "starting" state nearly every 
  	 monday morning ( I think it happens systematically but as I'm 
 	 not 100% sure, I prefer to say "nearly" ). 

.1>    Is it possible for you to provide me with an SNA trace (i.e. one
.1>    continuous trace) that starts when there is a normal working connection
.1>    and lasts through the events:
.1>
.1>      o IBM is stopped
.1>      o IBM is restarted
.1>      o MRMEMO hangs when trying to reconnect
.1>      o The MRMEMO server is SHUT down and reSTARTed and the conection works
.1>        again

	I will manage to get such a trace the next time the IBM is stopped.
	This is planned for the coming week-end.

	Can you confirm me that a trace at the session level on the DEMSA
	 gateway is what you expect ?

.1>    If you can produce the SNA trace described above, I would also like to
.1>    have the MRMEMO1.LOG from the session to match the trace against.

	Sure ...

.1>    This is not how it should work. The connection should be reestablished
.1>    automatically when the IBM comes up again.

	OK. This means I was right to enter this note.

.1>    What are the version numbers of VMS, MRMEMO and DECnet/SNA API?

	VMS    : 5.3 on a VAX 6410.
	MRMEMO : 2.0
	API    : 2.2

Regards.

- Philippe -

.3>	Can you confirm me that a trace at the session level on the DEMSA
.3>	 gateway is what you expect ?

    Yes

    Anders

    Sorry, but it seems the IPL of the IBM has been cancelled this
    week-end. Our gateway is still up and running and I have today no
    valuable informations in the DEMSA trace log file.
    
    I will keep on tracing and will come back here as soon as I have
    relevant infos.
    
    Thanks and regards.
    
    - Philippe -  

    I do not forget this one, but I've been off for a few weeks.
    The traces are taken again. Hoping I get a valuable one soon ...
    
    Regards.
    
    
    - Philippe Wauters ( TSC Belgium ) -

    Could you please tell me if we have to stop the MRMEMO gateway during
    system shutdown ( in SYS$MANAGER:SYSHUTDWN.COM ) ? Up to now, there is
    no gateway shutdown when our systems are stopped, and I wonder if this
    does'nt induce problems when the systems are restarted.
    
    The reason why I ask this is because our cluster has been stopped this
    saterday morning and restarted a bit later. The IBM's have not been 
    stopped.
    
    I have found the gateway in the "Starting - Connecting to MEMO" state
    this morning. We had to deactivate/reactivate the LU on the IBM to be
    able to restart the gateway.
    
    So : MC MRMMAN SHUT 1 or no SHUT 1, that's the question ...
    
    Regards.
    
    
    - Philippe -
            

    RE .7

    There shouldn't be any problems to take down a node running MRMEMO without
    doing a clean SHUT first. If you have problems with it, it is probably
    another symptom of the problem you described in .0 (it is the same
    system, isn't it?).

    Are they using "real" IBM stuff in the other end or some "compatible"
    equipment?

    Anders