Snmpget succeeds but snmpwalk and snmpgetnext fail

First, we were contracted to write software to a Red Hat distribution 
with system monitoring through SNMP using an Agent-X module, the OS is 
RHEL 5.2 using net-snmp 5.3.1-24 (the one provided by the distribution); 
we are not allowed to request a change to these specifications.

The Agent-X module appears to load correctly and mostly functions; we 
can query individual leaf nodes of the database such as:

[root@imgsvr bin]# ./snmp_agent &
[2] 13508
[root@imgsvr bin]# snmpget -c localCom -v2c localhost 
.1.3.6.1.4.1.36383.1.1.1.0
SNMPv2-SMI::enterprises.36383.1.1.1.0 = STRING: "1.0"
[root@imgsvr bin]# snmpget -c localCom -v2c localhost 
.1.3.6.1.4.1.36383.1.1.2.0
SNMPv2-SMI::enterprises.36383.1.1.2.0 = STRING: "20110118"
[root@imgsvr bin]# snmpget -c localCom -v2c localhost 
.1.3.6.1.4.1.36383.1.1.3.0
SNMPv2-SMI::enterprises.36383.1.1.3.0 = INTEGER: 1

If we then try to use either snmpwalk or snmpgetnext the response is the 
same:

[root@imgsvr bin]# snmpwalk -c localCom -v2c localhost 
.1.3.6.1.4.1.36383.1.1
Timeout: No Response from localhost

[root@imgsvr bin]# snmpgetnext -c localCom -v2c localhost 
.1.3.6.1.4.1.36383.1.1
Timeout: No Response from localhost.

When the system fails to respond in this fashion snmpd itself appears to 
freeze waiting a response from our Agent-X module. Stopping the module 
(e.g. kill 13508) will normally free up snmpd to respond to other 
queries, but if we wait long enough the only way to kill the module is 
with a kill -9. In addition, once the module is killed the only way to 
re-enable support for querying our program information requires fully 
stopping and restarting snmpd, then starting our Agent-X module again.

The data in our MIB is fairly static, specific fields that are updated 
semi-regularly by the program with no dynamic tables or other such 
fields, as such we made use of the netsnmp_register_scalar() function 
with the understanding that it would insert the proper helper functions 
eliminating the need to specifically handle getnext or other serial 
requests.

If you need any other information to diagnose this problem please ask. 
Any help resolving this issue will be greatly appreciated.

Patrick

------------------------------------------------------------------------------
Achieve unprecedented app performance and reliability
What every C/C++ and Fortran developer should know.
Learn how Intel has extended the reach of its next-generation tools
to help boost performance applications - inlcuding clusters.
http://p.sf.net/sfu/intel-dev2devmay
_______________________________________________
Net-snmp-users mailing list
[email protected]
Please see the following page to unsubscribe or change other options:
https://lists.sourceforge.net/lists/listinfo/net-snmp-users