Conference azur::mcc

>    I'm using GRAPH from the operation menu in MCC v1.3 to monitor the
>    lines of a DECbridge620. I've started graph on the 4 lines on
>    the multicast throuhghput. After sometime (some hours) of monitoring 
>    the little man puts his hands on his head (I do this all the time) and 
>    monitoring stops. The message window says:
>    
>    			Invalid Time Specified
>    
>    What could be the reason of this error. 

  There could be something going on in the Performance Anaylzer (PA), but
  I don't know what .. I've seen this error too when using the Graph
  widget.  But I've never seen it while using FCL - have you?  Could you
  try the equivalent FCL show command and see if you get the same error?
  If not .. there could be something going on in the Graph Widget.

  Thanks /keith

Hi,

Just to say that I've seen the problem as well when graphing on phaseIV circuits
and node hdlc link * log stat *. As .0 states it is random and can occur after
a long period (12 - 24hrs).

Ian

    re .2
    
 > I don't know what .. I've seen this error too when using the Graph
 > widget.  But I've never seen it while using FCL - have you?  Could you
 > try the equivalent FCL show command and see if you get the same error?
    
    What do you mean with the 'equivalent FCL show command'? The show        
    statistics?
    
    -Geert-
    

re: .3

>> What do you mean with the 'equivalent FCL show command'? The show
>> statistics?

  Yes - the FCL command equivalent to graphing statistics ... well,
  the FCL doesn't graph them, just lists them out.

    show bridge <name> line <line-number> all statistics, -
    for start +00:00 dur 00:00:58, at every 00:01:00

  This says you want Bridge Line statistics calculated every 1 minute;
  collecting counters 58 seconds apart.  The 'at every' and 'for' are
  a few seconds apart to allow the calculations to be done and write
  the output to the screen.  

  Crank this up in a window to see if you get the 'invalid time specified'
  error over the same period of time which you saw it during graphing.  If
  you don't get the error, there could be something up with the graph tool.

  Ok ?

/keith

    
    I've started the fcl command 5 days ago and it is still running without
    problems. Does this means that there's a problem with the window
    interface of MCC?
    
    -Geert-
    

RE: .5

Geert,

>    I've started the fcl command 5 days ago and it is still running without
>    problems. Does this means that there's a problem with the window
>    interface of MCC?

  This would be my guess .. that something is going wrong in the Graph
  Widget.  Or maybe the FCL is just more forgiving to whatever is going
  wrong.

/keith

  Yes - a QAR or CLD should be filed.  I suspect the proper component
  would be the Iconic Map.

  We need to identify 'where' the error is coming from -- there are a lot
  of MCC components executing between the Graph request and the actual
  calculation of statistics.

  Could you try setting up a debug logical?  This will instruct the
  Information Manager/Dispatcher to log the details of information flow
  through the MCC system, and hopefully, pinpoint which component is
  generating the 'Invalid Time' error.  At the DCL/CSH prompt:

  VMS:      define mcc_im_log 68000000
            man/ent/inter=decw

  ULTRIX:   setenv MCC_IM_LOG 0x68000000
            mcc_iconic_map

  This will cause *A LOT* of output to be written to sys$output (stdout).
  I'd expect the last page or so of output to have the details of the
  error .. Please post the results if you could.

  I'll try running this test too - we can compare results.

  Thanks /keith

    Keith,
    
    I have defined the logical and assigned sys$output to a file. I started
    Graph to monitor 3 types of statistics on a line of bridge.
    The Graph was started at 15:30 and the error occured at 20:00:08 on
    a Graph called 'count of invalid protocol messages'.
    The resulted logfile has a size of 12000 blocks! I think its a little
    bit to large to put it here. Do you now a way to calculate the actual
    time from the hex timestamps in the logfile?
    Here is an extract from the logfile:
    
    RETURNED Function HANDLE = 5F3814, state = MORE,   context = CCE8B0,
    substate =
    0
              time_stamp = 5F37D4, out_entity = CCE770
              out_p = 5F39D4, out_q = 0,  returned CVR: RESPONSE
    
    
    MCC_CALL Function HANDLE = 5F3814, state = MORE,   context = CCE8B0,
    substate =
    0
              verb = 1, in_entity = CCE450, partition = 4
              time_spec = 0, in_p = 5F3754, in_q = 0, out_entity = CCE770
    MCC_CALL dispatch HANDLE = 5F3814, module is MCC_DOMAIN_FM
    
    
    MCC_CALL Function HANDLE = 5F5794, state = FIRST
              verb = 1, in_entity = BA0890, partition = 4
              time_spec = 5F6BD4, in_p = 5F58D4, in_q = 0, out_entity =
    ACA930
      TIME_SPEC:
    time frame at 5f6bd4 hex
    
    
    If I can calculate the time, I can try to locate where the error
    occured and I could place that part in the notesfile.
    
    -Geert-