Conference csc32::consolemanager

1221.0. "Stack dump when issuing a BREAK" by 45769::CLEPHANEB () Thu Feb 29 1996 07:11

 
 OpenVMS V6.1
 PCM V1.6-110
 
 Here at British Gas E&P, Digital FM system manage over 40 systems around the 
 country. 31 of these systems are connected to PCM over a wide area network 
 using TELNET and DS700's. In the last week or so I have added a further 7 
 systems and we are now experiencing the following problems.
 
 When connecting to a system and issuing a break my process stack dumps with 
 the following. 

Do you want to transmit a break? [Y/N] y
%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=0000002C, P4

  Improperly handled condition, image exit forced.

        Signal arguments              Stack contents

        Number = 00000005                00000000
        Name   = 0000000C                00000000
                 00000000                0000B58B
                 0000002C                00000000
                 0000B719                20000000
                 03C00004                7FE7B9B4
                                         7FE7B984
                                         0000B5AD
                                         00000001
                                         0015C2B8
  
        Register dump

        R0 = 00129B60  R1 = 00000005  R2 = 00000000  R3 = 0000000D
        R4 = 7FE7B994  R5 = 00001A58  R6 = 00000CC8  R7 = 0015B300
        R8 = 00002ECC  R9 = 00000000  R10= 00000000  R11= 00002EC8
        AP = 7FE7B818  FP = 7FE7B7D8  SP = 7FE7B854  PC = 0000B719
        PSL= 03C00004

 
 
 Sometimes when issuing a break, I get the system is unavailable or disabled 
 message and get left at the command prompt. On investigation I discover that a 
 controller process has died with a similar stack dump in its logfile. 
 
 This has only started since adding these extra systems.
 
 nb. These extra systems caused a third controller process to be created.
 
 Also related to this (as it started happening at the same time) is another 
 problem. The system running (VAX 4000) PCM has a VAXstation 3100 booted into 
 it using EWS which we use to display the C3 and ENS displays. When starting 
 the C3 display, clicking onto the C3 window causes the whole workstation to 
 hang. We can only clear this by rebooting. To stop this from happening I 
 iconize the display immediately it comes up and just leave the ENS window 
 running. This means we can't use the C3 display to connect to anything but we 
 can get round this for now by doing a CONNECT CONSOLE from a decterm.
 
 Any ideas anyone ? Are these problems are fixed in ECO2 ?
 
 Regards,
 
 Bruce Clephane
 
 

    The ACCVIO problem is new, however, the problem with the systems being
    disabled is not. The code checked the wrong status value when it issued
    the break command and thought it had lost connection to the system
    which in turn caused it to shutdown the connection. The break is
    acutally issued though. This problem is definitely fixed in ECO2.
    
    Now the bit with the EWS booted station is interesting but I have a
    question: The C3 that you start - is it being displyed to the 3100 or
    to the local PCM engines display?
    
    Regs,
      Dan

    
    Dan,
    
    Thanks for the info. on ECO2.
    
    The C3 window is being displayed to the EWS served 3100. The 3100
    (Vaxstation M38) starts up 3 items on startup when logging in. A System
    Watchdog display, A decterm monitoring backup jobs and the C3 display with
    the Events window. The ENS window works fine but if you don't iconize the
    C3 display almost immediatley it starts, the workstation hangs.
    
    Sometimes after about 10 minutes or so the workstation seems to catch
    up and free up all the windows. It's only 'recovered' in this way 2 or
    3 times out of about 50 or 60 hangs.
    
    Regards,
    
    Bruce
    
    ps.
    
    I've just recreated the situation again while typing in this note, an
    SDA> SHOW PROCESS/CHANNEL shows the process busy on NET4496:
    and MBA4651:, could this be some kind of mailbox quota/size problem ?
    

    
    ... more on .2
    
    Just rebooted the 3100, the process hung around until the workstation
    had reloaded EWS and terminated with an exit status of 12DB821C ?
    
    Bruce
    

    >mailbox quota size problem
    
    Doubtful. The busy channels you mentioned are more than likely expected
    as PCM uses mailboxes for inter-process communication and since you
    are sending the display across decnet we would expect the NET device
    as well.
    
    >had reloaded EWS and terminated with an exit status of 12DB821C 
    
    This message code translates to:
    
    %XLIB-F-IOERROR, xlib io error
    
    which is a very general status for X apps when they cannot communicate
    with the server via whatever transport is being used. I will have to
    try this on a VT1300 and see what happens. The C3 may be giving the EWS
    Xserver a hernia trying to handle the complexity of the screen or some
    such.
    
    Regs,
      Dan