SNMPv3 polling failed after host router is restarted

[Liki Lee]

Our customer uses SNMPv3 polling to get sysUpTime from some router hosts.
For some routers, SNMPv3 polling becomes failed after the router is restarted, although snmpwalk from the machine of zabbix proxy works well.
SNMPv3 Polling becomes available after SNMP cache reload on the machine of zabbix proxy.

Does anyone know what the problem is? Or is it a bug of zabbix proxy?

[Liki Lee]

To add some detail, we are using zabbix 5.0 on RedHat linux 8.5.

[Markku]

Have you checked that the engineIDs are unique in all routers?

https://assets.zabbix.com/files/events/meetup_20200813/Arturs_Lontons_Zabbix_SNMPv3.pdf

Markku

[cyber]

Maybe... just maybe... it is because after router restart its boot count changes and then it wont match the cached values. After snmp cache reload it is all reset and started all over and polling resumes...
Its pretty much same with credentials, if you happen to change them , then you should reload snmp cache also...

[Liki Lee]

Thanks for the above replies.
Finally the problem is found. The router's implementation of SNMPv3 is incomplete. EngineBoots is not correctly implemented. When the router is restarted, EngineBoots
is not increased.