Conference azur::mcc

4313.0. "MCC_EVC_SINK object goes away...fails to restart T1.3" by GLFNUT::LICAUSE (Al Licause (338-5661)) Wed Dec 23 1992 12:58

I've had the MCC_EVC_SINK process running for several days now under T1.3
and have noticed many errors in the log files.

Periodically a window will open that says "One or more events were lost for
Domain * Rule *".

MCC_EVC_SINK.LOG has many failures....

16-DEC-1992 15:07:13.81, Disconnected link to COLLECTOR
16-DEC-1992 15:07:13.87, Waiting for connect request from a COLLECTOR
16-DEC-1992 15:09:13.90, Connected to COLLECTOR
16-DEC-1992 15:09:13.94, Received COLLECTOR event over DECnet
>>Error translating MCC event 0, status = 1000, 16-DEC-1992 15:09:13.95
16-DEC-1992 15:09:13.94, Disconnected link to COLLECTOR
16-DEC-1992 15:09:13.96, Waiting for connect request from a COLLECTOR
16-DEC-1992 15:11:14.09, Connected to COLLECTOR
16-DEC-1992 15:11:14.12, Received COLLECTOR event over DECnet
>>Error translating MCC event 0, status = 1000, 16-DEC-1992 15:11:14.13
16-DEC-1992 15:11:14.12, Disconnected link to COLLECTOR
16-DEC-1992 15:11:14.14, Waiting for connect request from a COLLECTOR
16-DEC-1992 15:13:14.23, Connected to COLLECTOR
16-DEC-1992 15:13:14.25, Received COLLECTOR event over DECnet
>>Error translating MCC event 0, status = 1000, 16-DEC-1992 15:13:14.26
16-DEC-1992 15:13:14.25, Disconnected link to COLLECTOR
16-DEC-1992 15:13:14.34, Waiting for connect request from a COLLECTOR
16-DEC-1992 15:15:14.35, Connected to COLLECTOR
.
.
.

Today the object went away even though the process was still running.

The most recent error log showed the following:

$ if f$mode() .nes. "INTERACTIVE" then exit
$manage/enterprise/presentation=mcc_evc_sink
>> Error waiting for internal event, at step 5, status = 52877426, 23-DEC-1992 
23-DEC-1992 12:50:26.06, Terminating sink management thread
23-DEC-1992 12:50:26.06, Terminating COLLECTOR sink management thread,status = 
23-DEC-1992 12:50:26.07, Terminating COLLECTOR event sink
%MCC-E-EVENT_LOST, events matching the request were lost
  SYSTEM       job terminated at 23-DEC-1992 12:50:27.98
<CR><LF>  Accounting information:
  Buffered I/O count:             118         Peak working set size:    8026
  Direct I/O count:                49         Peak page file size:     12991
  Page faults:                   3065         Mounted volumes:             0
  Charged CPU time:           0 00:00:05.37   Elapsed time:     0 00:00:17.26


Any one seen this before or has it been reported?

Al

As you might guess, many people are away for the holidays at the moment.
The best way to make sure this get propper attention is to file a QAR
against the notification component.

-- Erik

    events lost means that the event pool was full, and that some were
    lost. The sink should not stop on this condition and should be qared.
    This could happen if you had a process out there blasting a ton
    of collector events, or if you had an mcc process out that some
    how got shot (killed) while it had an outstanding getevent request
    to the event manager.  The outstanding request would cause the event
    manager to hang onto the events so that it could report them to
    the getevent requestor (the one that was hung).
    
    As to how to check for these conditions, you got me...I usually
    just
    kill all my sinks and mcc processes and start again. As the release
    notes and manuals used to state, dont' go around shooting your mcc
    processes you don't know what they might be in the middle of....
    

thanks for the information.

I always get confused when reconfiguring a WANrouter to sink events to DECmcc.

I could never remember if the Number = 82 field should be Name = MCC_EVC_SINK
or Name = MCC_EVL_SINK.

I think I have it straight now......EVL is the one that picks up DECnet or
OSI events and EVC picks up events from/for the collector....right?

In any case, I had some routers pointing to EVC sink which tended to flood
this process with the wrong kind of event.

The problem seems to have subsided.

Al