engine boots value is always 1 if specified engineID with -e flag in trapsess

[Vijendra Agarwal (vijagarw)]

Hi All,

I am using the net-snmp 5.5-54 version from centOS.

My issue is related to SNMPv3 trap message. I have multiple agent configured with different engine id for each.
Inside snmpd.conf of each I added trapsess entry with -e <engine_id> flag.

I have created the user with each engine_id of the agent in snmptrapd.conf. The users gets created successfully.

When my agent start it send some clear traps for the resources it monitors by monitor plugin.

I am able to receive the trap on host where my snmptrapd application is running.

I am facing issue when any of my snmpd agent get restarted. If it get restart and snmptrapd is still running then the traps where not successfully authenticated by snmptrapd and it gives issue as "message too old" and drop the packet.

I check on tcpdump and found that the engine boot value always comes as 1. The snmpd agent does not send the correct snmp engine boot value.

Then I did the change that 0 removed the -e <engine_id> from trapsess entry of snmpd.conf, allow the engine to add the engine_id to the packet. After this change I checked the tcpdump again, now this time engine boot value given by agent is correct. Everytime it gives B=5 or B=16 ...

I have a query here
why it does not give the actual engine boot value when we specify the engine_id specifically in trapsess entry?
Is it bug in net-snmp code / it is as per RFC?
Are there any way to solve the above issue?

Thanks
Vijendra.

[b***@gmail.com]

Post by Vijendra Agarwal (vijagarw)
Hi All,
I am using the net-snmp 5.5-54 version from centOS. 
My issue is related to SNMPv3 trap message. I have multiple agent configured with different engine id for each. 
Inside snmpd.conf of each I added trapsess entry with –e <engine_id> flag. 
I have created the user with each engine_id of the agent in snmptrapd.conf. The users gets created successfully. 
When my agent start it send some clear traps for the resources it monitors by monitor plugin. 
I am able to receive the trap on host where my snmptrapd application is running. 
I am facing issue when any of my snmpd agent get restarted. If it get restart and snmptrapd is still running then the traps where not successfully authenticated by snmptrapd and it gives issue as “message too old” and drop the packet. 
I check on tcpdump and found that the engine boot value always comes as 1. The snmpd agent does not send the correct snmp engine boot value. 
Then I did the change that 0 removed the –e <engine_id> from trapsess entry of snmpd.conf, allow the engine to add the engine_id to the packet. After this change I checked the tcpdump again, now this time engine boot value given by agent is correct. Everytime
it gives B=5 or B=16 …
I have a query here 
why it does not give the actual engine boot value when we specify the engine_id specifically in trapsess entry?
Is it bug in net-snmp code / it is as per RFC?
Are there any way to solve the above issue?
Thanks
Vijendra.

Did you get resolution to this. What is the fix?