Yes, both Netview and HPOV has a history.
But, for me atleast. Netview changed alot to the better from HPOV.
HP cant handle events at all, IBM supplied rules and dynamic eventgroups.
I am forever greatfull. Netview is for me better then anything else on
the market today.
But things could be even better.
Of course I can write my own state engine that handles the things I ask
for. (Thank you IBM for the API's that HP charge for)
Rulesets and nvcorrd changed alot for us who wants to process events.
But their are pieces missing. Like Selection wich makes DB queries possible.
Instance id and a "Query snmp" node which could handle that instance.
And best of all give me the power to create my owv messages with popen().
The developers of Netview havent fix PRINTF() in "T/EC slot mappping"
which is scary. That function is a key to beeing able to generate customized
T/EC events. 5.1 is the second version which states that you can use it.
And again next-version-will-fix-it-all.
But shouldnt we agree on that alot of your customers maybe spends
100-1000 of hours on just creating their own solution. These things I
ask for, could do the job within a ruleset without creating alot of
zombie process.
/Terje
Terje, we have to remember that the format of the basic netview traps was
in place years
before there was such a thing as a Ruleset editor, and were designed for
use by Netview
(or Openview) itself. We have all kicked around various ways of getting
straight to the answer
of which interface is involved and is it significant; I think if it were
trivial it would have been done
already. The more people who send such requests to netview@tivoli.com, the
better chance
it has of getting put into the product plan. So send your note, by all
means.
One simple approach that has not yet been discussed in this thread is
re-issuing your own
event. An Interface Down event could execute an snmptrap command, or event
command.
It might be kind of messy if you put it all in the command window on the
trap itself, so I would
put it in a little script. Parse off the second word of $3. This saves you
a database lookup.
Send a new event with $2 and this piece in it (or not, if you want to thin
it out at this point).
In the ruleset you now have something to compare. The more smarts you put
in that little script,
the less processing done in the ruleset. But it does not require a lot of
code. Also, I would
make the customized event (ifdown, etc) a copy that is used only for
devices in a collection
to further reduce the number of events processed in this way. Make your own
enterprise id
while you are at it....
Cordially,
Leslie A. Clark
IBM Global Services - Systems Mgmt & Networking
(248) 552-4968 Voicemail, Fax, Pager
==========================================================================
==
OK! You miss what I am missing.
1. I want to use a ruleset to forward events to TEC, no external scripts.
2. I want to filter out certain if-types (ISDN, Async, Ethernets, Token)
3. I want to be able to do SNMP queries on instances i.e interfaces.
(Sometimes we have backup interfaces we want to verify
that its up and running, this is done by setting corrstat to
instance of backup interface)
4. Fetch info from external source and load a variable.
This is things that could be done if you guys:
1. Supplied Selection name of interface in down trap, perhaps in var 4.
2. Supplied interface SNMP instance id in traps, perhaps in var 4.
3. Changed "Compare MIB var.." node to allow us to supply instance as a
var.
4. Created a node which runs popen() and retrives stdout into varxxx.
Fairly basic requirements I think, does anyone agree ??????
/Terje
|