Conference azur::mcc

4298.0. "automatically restart of mcc_dna4_evl after dying?" by MUNICH::SCHWEMMER () Mon Dec 21 1992 12:09

Whenever EVL dies, MCC_DNA4_EVL also dies.
NETACP-Process is responsible to start EVL again all the time this happens.

Are there any plans for future releases of MCC, that MCC_DNA4_EVL will be
restarted automatically after dying?

Thanks in advance, 

Mathilde Schwemmer, DSC Munich

***

Please have a look to the following notes-entry and to the answer on it 
(just for background-info):


           <<< BULEAN::BULEAN$USER:[NOTES$LIBRARY]DECNETVAX.NOTE;1 >>>
                           -< DECnet-VAX Conference >-
================================================================================
Note 4772.0            EVL dies with NETASN and LINKABORT                1 reply
MUNICH::SCHWEMMER                                    48 lines  14-DEC-1992 10:12
--------------------------------------------------------------------------------
    

Our customer has problems with sporadically (1 times a week or more frequently)
dying EVL-process under VMS V5.4-2 (HW).

The evl-logfile contains the following message:

%EVL-E-NETASN, unable to assign a channel to NET
-SYSTEM-F-LINKABORT, network partner aborted logical link

Could somebody please give me a hint how to find out, why this message
occurs?

Could I find out more by turning on some bits for EVL$LOG? When yes,
which one makes sense?

 
        Bit 0 = any data base changes. 
        Bits 1-3 refer to messages: 
           Bit 1 = any time link to sink is opened. 
           Bit 2 = any time link is confirmed to a sink node receiver. 
           Bit 3 = any time link to event monitor process. 
        Bits 4-6 refer to events: 
           Bit 4 = all raw enents. 
           Bit 5 = events queued to a "sink". 
           Bit 6 = events received by a receiver. 
        Bits 7+ = reserved. 


Or does somebody have other ideas how to find the reason?
(I've read the entries 3939, 2180, 2069, etc. It doesn't help me so much).
    
I've already checked the parameter PQL_DFILLM and PQL_MFILLM. Both are
set to 100. This seems not to be a bottleneck.
Customer told me also, that there aren't double-existing DECnet-addresses.
Performance of the machine is o.k.


Background-info:
Because EVL is dying, also MCC_DNA4_EVL-Process (for DECMCC BMS V1.2) dies.
Customer has to start the process again and this isn't reasonable.


Any hint would be appreciated,

Thanks in advance.

Mathilde Schwemmer, DSC Munich
================================================================================
Note 4772.1            EVL dies with NETASN and LINKABORT                 1 of 1
BULEAN::LEE                                          11 lines  18-DEC-1992 11:32
--------------------------------------------------------------------------------
    >%EVL-E-NETASN, unable to assign a channel to NET
    >-SYSTEM-F-LINKABORT, network partner aborted logical link
    
    When EVL failed to get the sink data from NETACP via QIO call
    issuing a read from the mailbox, EVL will signal the error status
    returned from the QIO call.  The logical link between NETACP and
    EVL probably went away.
    
    NETACP is responsible to restart the EVL process whenever there is
    an event to log. May be DECmcc should do the same for MCC_DNA4_EVL
    if it can monitor EVL ???