Re: tooBig {} problem ...

Juergen Schoenwaelder (schoenw@cs.utwente.nl)
Wed, 13 Nov 1996 09:05:57 +0100

jorge1@swat.West.Sun.COM (Jorge A. Osorio) said:

Jorge> I'm running into a problem that when walking a MIB I get a
Jorge> tooBig{} error message for one of the MIB objects we have.

Jorge> I realize this MIB object can vary in size (up to 256 bytes),
Jorge> so I wonder if there is a size-limitation in this? Though, I
Jorge> don't get this error if I use the SNMP_Walk proc. Any ideas?
Jorge> Suggestions? Is there something wrong?

SNMP is based on UDP and hence every SNMP message must fit into one
UDP datagram. The maximum size of UDP datagrams differs from agent to
agent. You are walking through five columns of the table "in parallel"
and the resulting response messages that exceed the datagram size
limit of your agent.

The management application is responsible to deal with tooBig errors
by making the requested varbind list smaller. The SNMP_Walk proc asks
only one varbind per request which avoids tooBig errors (but is very
inefficient). Note that requesting only one varbind also allows
inconsistencies in one row because the values are read at a different
time (the agent might even reboot during the walk).

The inefficiency can be easily solved by using SNMPv2. The walk
command of the Tnm extension uses getbulk if available and this
reduces the number of SNMP requests. The second problem requires some
non-trivial logic on the manager side to deal with sparse tables, to
check sysUpTime during the walk (which is not well supported in the
walk command right now as someone posted a couple of days ago) and to
deal with tooBigs etc.
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)