[nsp] just fyi - 12.2(13) & snmp

[Tomas Daniska]

hi there,

this is just a notice in case somebody has snmp-related problems with
12.2(13)

i've seen a change in the way ios replies to some snmp gets. following
is a snippet from a mail i was sending that describes the issue - my
apologies for not rewriting it before posting here:


*************
<...snip...>
for the ciscoMemoryPoolFree variable:

erixon$ snmpget -c XXXXX non-122-13-box 1.3.6.1.4.1.9.9.48.1.1.1.6.0
cisco.ciscoMgmt.ciscoMemoryPoolMIB.ciscoMemoryPoolObjects.ciscoMemoryPoo
lTable.ciscoMemoryPoolEntry.ciscoMemoryPoolFree.1 : Gauge32: 36979812

erixon$ snmpget -c XXXXX 122-13-box 1.3.6.1.4.1.9.9.48.1.1.1.6.0
snmpget: Agent reported error with variable #1.
.iso.org.dod.internet.private.enterprises.cisco.ciscoMgmt.ciscoMemoryPoo
lMIB.ciscoMemoryPoolObjects.ciscoMemoryPoolTable.ciscoMemoryPoolEntry.ci
scoMemoryPoolFree.0: SNMP: Variable does not exist or access is denied.


well, pre-12.2(13) seems to be a bit non-standard as it replies the way
it should reply to GET-NEXT (not GET which actually was issued), but
it's still better than what 12.2(13) does (_VARIABLE_ does not exist)



however, if you take some more standard mib, e.g. the ifTable, then we
have this behaviour with GET:


erixon$ snmpget -c XXXXX non-122-13-box 1.3.6.1.2.1.2.2.1.1.0
interfaces.ifTable.ifEntry.ifIndex.0 : VARBIND EXCEPTION: No Such
Instance

erixon$ snmpget -c XXXXX 122-13-box 1.3.6.1.2.1.2.2.1.1.0
snmpget: Agent reported error with variable #1.
.iso.org.dod.internet.mgmt.mib-2.interfaces.ifTable.ifEntry.ifIndex.0:
SNMP: Variable does not exist or access is denied.



in other words, 12.2(13) says the VARIABLE does not exist (which is not
true) while pre-12.2(13) says INSTANCE does not exist (which is actually
the case). so, assuming i am the software that needs the information on
free memory and that i don't know this variable is actually a list of
some stuff, i'd do these conclusions:

- if i was querying a 12.2(13) box, i'd think the device does not allow
the whole MIB string (either it has not implemented it or it refuses
to). IPM (yes, as already stated, i mean internetwork performance
monitor) is a bit more strict on this in means that if the device does
not reply to this it denies any actions as it is not aware if it would
eat a significant ammount of the target devices' memory

- if i was querying a pre-12.2(13) box, i'd notice that .0 is not a
valid index so i'd try GET-NEXT or whatever else method to get the
values from the list. so - at least i know i have to ask somewhere
else...
<...snip...>
*****************

--
 
Tomas Daniska
systems engineer
Tronet Computer Networks
Plynarenska 5, 829 75 Bratislava, Slovakia
tel: +421 2 58224111, fax: +421 2 58224199
 
A transistor protected by a fast-acting fuse will protect the fuse by
blowing first.