Conference azur::mcc

4022.0. "How do I ALARM on "Test Successful" ??" by MSDOA::REED (John Reed @CBO - DTN: 367-6463 = DNIS) Tue Nov 03 1992 15:01

    Help !!
    
    
    I have a very simple problem, and I can't seem to find anyone who is
    capable of answering it for me.  I have a customer with a DECmcc V1.2.0
    VMS system.  He has several NODE4 nodes, and several STATION nodes, and
    about five DECserver x00 TERMINAL_SERVER nodes, and four or five Xyplex
    terminals server SNMP nodes.
    
    The customer wants the icons that are UNREACHABLE to change color.  For
    instance, I can make the Xyplex servers come and go, using the SNMP 
    ipreachability test for "up".  I can test the NODE4 nodes for CHANGE_OF
    STATE, ON,*) etc....   But I can't figure out what to use to make the
    TERMINAL_SERVER change color if it gets unplugged, or something.
    
    I tried:
    
    CREATE DOMAIN ATLANTA RULE SERVER_UNREACHABLE_MAJOR -
     EXPRESSION = (CHANGE_OF (TERMINAL_SERVER * Counter Creation Time, *,*), -
     AT EVERY 00:00:20), DESCRIPTION = "The Counters have been Zero'd on a -
    	Terminal Server.  This could be because the Server is unavailable -
    	or rebooted, or Software Zeroed."         ,-
     BATCH QUEUE = "MCC_ALARMS" , SEVERITY = MAJOR   , -
     CATEGORY = "Terminal Server"   , -
      Alarm Fired Procedure = DKA0:[MCC]MCC_ALARMS_MAIL_ALARM.COM  , -
      Alarm Exception Procedure = DKA0:[MCC]MCC_ALARMS_MAIL_EXCEPTION.COM,-
      Alarm Fired Parameters = "SOUL::JAMIE"
    
    But this told me that the data was an unsupported data type.  I looked
    at some other parameters, but could not decide which to alarm on.
    
    Any suggestions for a simple "TEST SERVER" type of alarm?
    I am looking for something that can be happy if the "Test Successful"
    comes back happily.
    
    JR
    

    I have worked this issue with several customers.  The best way I have
    found to do this is to set up an alarm which will always evaluate to
    false, one example would be to test for 'bytes sent < 0'.  This will
    always evaluate to false, but then if the server was unreachable, such
    as powered down or ethernet cable unplugged, then the exception
    procedure would fire.
    
    Steve Woestemeyer
    CSC/CS Network Support Team

    What about the attribute "software status" = normal. You can define an
    alarm to poll for the specific attribute..
    
    It's works
    
    

I've left phone-mail for John and talked to the CSC about this.

I think we have figured out a partial solution to this problem,
as Steve Woestemeyer has suggested.

In answer to the question about the Counter Creation Time attributes
not being alarmable, the answer to that is to use the Seconds Since
Zeroed attribute instead which is alarmable, and in fact is more desirable.
Since the Counter Creation Time is a 'synthetic' attribute derived
from the Seconds Since Zeroed information.  (We actually take the
Uptime value from the server, and back-calculate the creation time.
Since this can 'drift' up and down a bit due to timing, it would
not be constant anyway.)

-Dave