T.R | Title | User | Personal Name | Date | Lines |
---|
4015.1 | | YAHEY::BOSE | | Tue Nov 03 1992 17:15 | 7 |
|
Dave,
Do you encounter this sort of behaviour with SNMP entites
only? Can you try this out with other entities.
Rahul.
|
4015.2 | try fullname | CTHQ::WOODCOCK | | Tue Nov 03 1992 22:53 | 8 |
| Greetings,
You may want to try specifying the FULLNAME in the alarm expression and
any specific NOTIFY commands. This worked for me after encountering
similar problems.
kind regards,
brad...
|
4015.3 | re -.* | ANOSWS::COMFORT | Spent a little time on the hill | Wed Nov 04 1992 09:16 | 26 |
|
Hi,
< re .-2 >
Node4 and Bridges work fine. The bridges are in a subdirectory 3
levels deep. I am attempting to try TSAM, however TSAM is generating a
license check fail against the DECMCC-EMS license. As soon as that is
resolved, I can try it with TSAM. I will also test with the Data
Collector today. Note that the domain icon of the domain that the SNMP
entity resides in will light the appropriate color.
< re .-1 >
By fullname I assume you mean something along the lines of:
.test.phred where test is the DNS subdir and phred is the entity
All my alarm rules use the DNS syntax, with the exception of Node4
alarms. I have tried using the fullname and the UCX name for both
the alarm rules and targeting. But I will test this again.
Thank you.
Dave
|
4015.4 | include NS | CTHQ::WOODCOCK | | Wed Nov 04 1992 23:25 | 12 |
|
> By fullname I assume you mean something along the lines of:
>
> .test.phred where test is the DNS subdir and phred is the entity
Dave, I think (memory fogged) I used the full fullname including namespace. It
might be worth the shot.
LOCAL_NS:.test.phred
best regards,
brad...
|
4015.5 | Problem solved, thanks | ANOSWS::COMFORT | Spent a little time on the hill | Thu Nov 05 1992 10:22 | 8 |
|
Brad -
Thanks very much, this indeed solved the problem. I apologize for
being dense about the fullname thing. Once again many thanks.
Dave
|
4015.6 | snmp fullname vs internet name is the problem | MCC1::DITMARS | Pete | Thu Nov 05 1992 12:50 | 28 |
| This problem crops up over and over again.
We should clearly document this, clearly communicate it to our support
organizations, and in V.next, make the code smart enough to minimize
the possibility of it happening.
When you enter "SNMP .foo", you may think ".foo" is a fullname.
But the parser thinks it's an internet name, and in the internal
AES representation, the datatype is set to internet name.
The SNMP AM must be robust enough to try interpretting the instance
as either internet name or fullname, regardless of the datatype.
Or it's just lucky. :^)
The notification FM and IMPM, however, believe the datatype in the
AES. When it comes time to do targeting (in the FM) or notification
correlation (in the IMPM), the fact that the datatype is "wrong"
prevents the "right" things from happening.
The simple solution, as Brad points out, it to always use the namespace
when refering to an SNMP entity by its fullname in an alarm rule.
This forces the parser to interpret the entity's name as a fullname.
Leading dots are legal in an internet name, and are not enough to make
the parser understand that this is a fullname.
A more complicated solution is to refer to the SNMP entity by the exact
string that was used when registering it, as shown by the DIRECTORY
command (not SHOW!).
|
4015.7 | | YAHEY::BOSE | | Thu Nov 05 1992 15:10 | 13 |
|
>> The SNMP AM must be robust enough to try interpretting the instance
>> as either internet name or fullname, regardless of the datatype.
>> Or it's just lucky. :^)
Pete,
The SNMP AM first tries to use the instance as an Internet Name.
If that fails, then it is interpreted as a FullName. This was done
to overcome the ambiguity between Internet Name and Full Name. So
it is not just plain luck that it works that way. :-)
Rahul.
|