Conference azur::mcc

5239.0. "graphing and other problems" by CGOOA::KUHNEN () Tue Jun 22 1993 19:05

    I have recently upgraded to version 1.3 of mcc with the expectation
    that some of the problems with 1.2 would be fixed. The 1.2 problems are
    still evident along with a number of new problems.
    
    Problem 1.
    
    You cannot graph historical SNMP data. This actually worked in version
    1.2 of mcc but does not function now. I set the scope of interest time
    to a period in the passed with a duration that finishes before the
    current time. The interval time I used coicides with the interval time
    used to record the data. When I graph the data such as a count on an 
    SNMP interface I get a value of 1 repeated on the graph. Using the same
    scope of interest time using the operations menu returns the
    appropriate data.
    
    Graphing historical statistics does not work at all . The graph seems
    to now go off into never never land and does not graph any data even 
    though the little man continues to run as if the graph was functioning.
    
    The data is generally available through the command line interface
    without any problems.
    
    Problem 2.
    
    When I use the ip reachability poller to check for reachability I will
    occassionally get spurious failures. I have set the retries to 5 and
    the timeout to 15 seconds and am only polling every 10 minutes. These
    spurious failures also occur when a "real" failure occurs. That is if
    I loose a link that provides connectivity to 10 remote stations that I
    am polling, in addition to the 10 failures associated with the link
    going down there are usually a number of other reachability events that
    will occur at the same time as the first events. These will clear the
    next time they are polled 10 minutes later.
    
    Problem 3.
    
    I now find that some of my recorded data seems to be invalid. I
    recently went to look at the past utilization on an SNMP interface. I
    found that the utilization remained exactly the same for a number of
    hours. When I went to look at the counters using the command line
    interface I found that the counters returned what I will have to
    describe as "odd" values. Could this be a result of the upgrade
    procedure and should I restart from scratch the recording and delete
    the old recorded data ?
    
    Problem 4.
    
    The system seems to get a large number of "rule * encountered an
    exception" errors when there is more than one user of DECmcc on the
    system. The help desk here has a window open to MCC to keep them
    informed of major network events. When they are on the system there is
    a great increase in the number of these errors. I have looked at the
    system and cannot seem to find any indication that the system is low on
    resources and both accounts accessing DECmcc are set up with account
    resources greater than those suggested in the manuals and release
    notes.
    
    Any suggestions ?
    
    I have already entered some of these problems into the version 1.2 QAR
    system last year without any response. Should I re-enter them into the
    1.3 QAR system ?
    
    Thanks
    
    Harry
    
    
    
    
    
    Thanks
    
    Harry Kuhnen 
    
    
    

Re Problem 1

Graphing of snmp interface historical statistics is broken. The problem has been
identified and an engineer is looking at it.  We hope to fix it for the MUP.

The problem has already been CLD'd.

Verna

    What harware and operating system are you using?
    
    s/rob

    Operating system VMS 5.5-1
    
    MCC V1.3
    
    Vaxstation 4000 Model 60
    
    56 MB of memory
    
    I was able to get rid of the errors with two processes running by
    increasing my working set to 35000 from 25000 even though the MCC
    process that is running alarms only has on average 23000 pages in it's
    working set. Could there be a problem in the code related to paging
    under certain conditions ?
    
    Thanks
    
    Harry