[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference csc32::consolemanager

Title:POLYCENTER Console Manager
Notice:Kits, Scans, Docs on CSC32:: as PCM$KITS:,PCM$DOCS:, PCM$SCANS:
Moderator:CSC32::BUTTERWORTH
Created:Thu Aug 06 1992
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:1541
Total number of notes:6564

637.0. "How To Handle Terminal Server Failure" by PRMS01::HCARTER (M. Hill Carter @DTN 339-7781) Mon Mar 06 1995 15:18

   I have customer who wants to use Console Manager to manage a set of
   systems which will be used together as an Air Traffic Control System.
   The customer is concerned about the potential failure of the terminal
   servers which are connected to the console ports on the systems and wants
   Digital to state a failover methodology for the terminal servers.

   I see two potential ways to handle this:

   1) Monitor all terminal servers used by Console Manager with the 
      POLYCENTER DECnet Manager product. When a terminal server fails,
      use a spare terminal server on the shelf to replace the failed
      terminal server.

   2) Have a redundant terminal server connection (separate terminal server)
      for every system being managed. Then use a switch box to move the
      console connection over to the backup server when the primary server
      connection fails. This would require that the Console Manager map
      have two icons (primary and backup connections) for every managed
      system.

   Is there a preferred way to handle the failure of a terminal server?

   Thanks in advance for any help,

   Hill...
T.RTitleUserPersonal
Name		DateLines
637.1CSC32::BUTTERWORTHGun Control is a steady hand.Mon Mar 06 1995 18:1340
    Hill,
       Very interesting question!! To be honest there hasn't ever been an
    "official" preferred method to resolve the problem you describe.
    Terminal servers just don't break often enough, IMHO, for it to have
    ever been a real issue, i.e., only a handfull have ever asked a similar
    question. Personally, I like your second idea but it is of course
    the most expensive. As usual the elegance of any solution is
    proportional to the amount of money you can throw at it...;-}. You
    would literally have to configure each system in the database twice but
    with a different name. This of course requires redundant licenses as PCM
    sees each redundant node record as a separate connection. 
    
    I see another solution with some creative programming. I'm not familiar
    with that other network management product you mentioned but assuming:
    
    1. It has an API or some way to trigger an action routine.
    
    2. It has a user interface that can be run under control of PCM as a 
       pseudo node.
    
    In both cases you will have two config files that differ only in the
    connection details to the terminal servers. When a failure occurs you 
    will rename the backup config file to CONSOLE_CFG.DAT and issue a 
    reconfigure command. You will do this vai a command file.
    
    In case #1. You could write an application that detects the terminal
    server has failed. You would then perform the rename and reconfig.
    This method require no additional PCM licenses but you need to
    determine if the management application has an API.
    
    In case #2, the user interface to the product runs under control
    of PCM as a pseudo node. It has a scan profile that will contain an
    event that scans for the string that the user interface produces when a 
    server has failed. The command file mentioned above would then be run 
    as an action routine as we will setup an ENS filter to look for the
    event. This method will require only 1 additional license for the 
    application pseudo-node. 
    
    Regs,
      Dan
637.2Thanks for the quick reply!PRMS01::HCARTERM. Hill Carter @DTN 339-7781Mon Mar 06 1995 23:257
    
    Dan,
    
       Thanks for the reply to my base note. I will discuss your idea
       with the customer and see what they think.
    
    Hill...
637.3OPG::SIMONSimon Jackson @reo 830 X3879Tue Mar 07 1995 09:478
Hi,
   bear in mind also that we do detect breaks in the link between the PCM host
and the terminal server as a normal part of the product. The Daemon processes
have a timer routine which checks at a predetermined interval the state of
each of the connections and notifies if a connection breaks. Ensure the customer
understands that this functionality is in the product!!!

Cheers SImon...
637.4another terminal server failure questionICEKRM::KUNGFri Mar 24 1995 20:3811
    Another question along the same lines:
    
    Customer is reporting the following:
    	when a user connects through a terminal server to the node running
        PCM and then does a console connect to another node, the terminal 
        server they were connected through goes down.
    
    I looked through other notes and replies and could not find this.
    Is this pilot error or something real in PCM?
    
    Thanks.
637.5Yet Another Terminal Server Failure QuestionGRANPA::MSTENNETTFri Apr 14 1995 00:296
    I am also having this same problem at my customer site.  Every time we
    try to connect to a console from a terminal server that server crashes.
    My customer is really after me about this and any help or pointers that
    you can provided would be very much appreciated.
    
    Thanks in advance for your help.
637.6CSC32::BUTTERWORTHGun Control is a steady hand.Fri Apr 14 1995 22:027
    Is it a real decserver and if so what type? The onlything I can imagine
    is when PCM does a SETMODE or SETCHAR QIO against the users terminal
    that the server gets something from the lat driver that it can't
    handle. I've never had this kind of a problem from VCS or PCM.
    
    Regs,
      Dan
637.7DECservers continuedGRANPA::MSTENNETTMon Apr 17 1995 02:523
    The customer is using DECserver 500s and 200s.  The console services
    are hosted by DECserver 200s and the connection point of failure is a
    DECserver 500.