Re: Peculiar behaviour with snmp get - scotty 2.0.2

Juergen Schoenwaelder (schoenw@cs.utwente.nl)
Wed, 11 Sep 1996 01:30:09 +0200

michele@pqtech.com (Michele Chen) said:

Michele> Thanks for the quick response Juergen. Unfortunately,
Michele> due to dependencies on itcl and other extensions, I
Michele> can't yet move to tcl7.5 and tk4.1. However, as you
Michele> requested, I've provided a sample script that
Michele> demonstrates the problem (at least at my site). My
Michele> code does a lot of processing, so I had to send out a
Michele> LOT more requests in this sample to demonstrate the
Michele> problem, but you should see it.

I did run your code against a couple of test agents and the good news
is that scotty 2.1.* is able to handle this test case quite well. Note
that scotty 2.1.* has some flow control session options which limit
the number of asynchronous requests on the wire. This allows an
application to create as many requests as required without worrying
about e.g. UDP buffer overflows or floading intermediate routers.

Your script actually creates 240 request before the event loop is able
to process incoming responses (the after command does not process
events and hence you can even remove it). This works well with scotty
2.1.* since the window mechanism takes care of this. This might not
work well with 2.0.* since I introduced the window option in November
1995. 2.0.* will suffer from UDP input queue size limitations etc.

So the final answer is: Your problem has been solved in the upcoming
2.1.4 version. You now have two options: Find a way to upgrade to Tcl
7.5 or to port these new features from scotty 2.1.* back to 2.0.*
(which might be difficult because I did a lot of changes since 2.0.*).

Juergen

--
Juergen Schoenwaelder schoenw@cs.utwente.nl http://www.cs.utwente.nl/~schoenw
Computer Science Department, University of Twente,   (Fax: +31-53-489-3247)
P.O. Box 217, NL-7500 AE Enschede, The Netherlands.  (Tel. +31-53-489-3678)