T.R | Title | User | Personal Name | Date | Lines |
---|
1867.1 | a dumb idea from a dumb manager | TOOK::MATTHEWS | | Tue Dec 03 1991 11:50 | 13 |
| I recently read the Simple Book by Marshall Rose. I distinctly remember
a discussion of the use of the GET operation vs. GET NEXT operation.
The behavior described fits the GET operation semantics. You might
want to check with MOLAR::BOSE and see if the SNMP AM uses the
GET operation for a list of attributes. If so, then if any attribute
within the list forces an error in the agent, the complete operation
is aborted. Since the FCL seems to work right and the iconic map
doesn't there may be a small difference in how they use the AM which
triggers the use of GET vs GETNEXT.
I only claim to be a layman trying to offer a possible insight.
wally
|
1867.2 | In the PMs | MKNME::DANIELE | | Tue Dec 03 1991 12:21 | 17 |
| The AM has successfully requested and retrieved the attributes.
The FCL and/or MAP cannot display them. FCL's error message is (and has been
for 3 versions now) particularly misleading. It means the dictionary
definitions do not match a value that has been returned across the call
interface.
A typical case is the agent returns a value for an attribute of type
enumeration that is not in the range defined in the mcc dictionary.
In any event you can check this out using DAP, etc.
And your point is well taken that the 'raw value' should be presented to the
user even if it isn't present in the dictionary. MIBs as known to the manager
will always be out of synch in some manner with the agents.
Regards,
Mike
|
1867.3 | | NISKY::SHREFFLER | I'll have a Caffeine free VMS please | Tue Dec 03 1991 13:41 | 7 |
| Thanks for the help. It turns out that the object is a feature (multisessions)
that is supposed to return a value of 1 for disabled and 2 for enabled. If I
enable this feature for a port the snmp get receives a valid value although it
says the feature is disabled. It looks like someone goofed. I bet that when
the feature is disabled the terminal server is sending a zero and that is out
of range. If I had 3 hours to kill while my parse table is rebuilt I would try
to verify my assumption.
|
1867.4 | use FCL log bits to verify values being returned | TOOK::CALLANDER | MCC = My Constant Companion | Fri Jan 10 1992 10:00 | 10 |
| don't waste three hours rebuilding just to test the assumption. Instead
define MCC_FCL_PM_LOG to a value of 8. You will be able to see exactly
what is being passed back across the call interface. Change from
enabled to disabled and you will be able to pick up both values and
then you can make the change. If you are running the V1.2 kit the
parse table build will ONLY rebuild the changed entity class which
means minutes not hours on the rebuild.
jill
|