[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

1338.0. "details on PCm Characteristics : Timeout" by ULYSSE::BAUDELLE () Wed Jul 10 1996 02:44

    
    Could you tell me how Ens detects that it can't join an action to
    understand when it could be useful to modify the Pcm timeout
    (manual Pcm: 3-14)
    
    Thanks 
    
    Pierre
T.RTitleUserPersonal
Name		DateLines
1338.1CSC32::BUTTERWORTHGun Control is a steady hand.Wed Jul 10 1996 13:299
    Pierre,
      In a nutshell, ENS expects the action routine process to assign a
    channel to a listener "socket" (mailbox) created specifically for
    that action routine process. This parameter should really never have to
    be modified. If there becomes a need I would argue that the PCM engine
    is overloaded, i.e. you need more horsepower.
    
    Regards,
       Dan
1338.2More detailsULYSSE::BAUDELLEThu Jul 11 1996 04:2232
    Dan,
    
    Thanks for your quick reply.
    I know you are the Console manager Ipc specialist so let's give you
    more details.
    
    Infact i am facing a kind of overload problem as at some period of the
    day (when many events are received by Ens) Console notify goes to Lef
    state (old problem for me, Ipmted for weeks). Then if i have a look at the log files i
    see some actions failing to run correctly (like re-sending Multi-line
    window to a Pc which is down for example). If i stop this action then
    Console notify goes back to HIB state and everything works fine again.
    
    My question is: I know that action routines are run in some parallel
    mode but it looks like the feeding of action routines mailbox By Ens is
    done in sequential way. So if an action   routine runs slow for some
    reasons (link problem) and that many events are coming at the same time
    could we end up with the mailbox becoming full and Ens going to LEF ?
    
    Decreasing the timeout then forcing a new image to be activated (and
    the previous mailbox deleted ) could be a solution or increasing
    DEFMBXMXMSG to reach this timeout before the mailbox gets full would be
    the right workaround ? Or am i wrong ?
    
    
    Christian Heusbourg , Oscint Project leader , has done some testing
    showing this kind of behaviour (Oscint Notes file , Maria menicacci
    Entry)
    
    Thanks and regards
    
    Pierre
1338.3CSC32::BUTTERWORTHGun Control is a steady hand.Thu Jul 11 1996 13:187
    Pierre,
       I fixed a problem where the ENS process could accvio or hang in this
    manner. The code fix is in ECO 3. ECO 3 is kitted so it's probably
    worth your while to try it.
    
    REgards,
       Dan
1338.4CSC32::BUTTERWORTHGun Control is a steady hand.Thu Jul 11 1996 13:244
    And see note 1308.4 for the kit location.
    
    Regards,
       Dan