IPv6 erratic SNMP

[xxiii]

v2.4.5:

I'm having a lot of trouble trying to monitor SNMP via IPv6. Its just now started working again after restarting zabbix.

I suspect zabbix is binding/using the global temporary dynamic IP instead of the server's permanent IP, and it works up until that temporary one expires.

This is a guess however. I tried to assign a specific SourceIP but ran into a problem there (addressed in a separate thread).

I had a packet capture showing that zabbix was definitely sending snmp out on the temporary dynamic. The packet capture from before restarting zabbix is gone unfortunately. There were far fewer requests than their should have been, but possibly this was because zabbix had declared most of the items "became not supported", or, it was trying to send out using a no longer valid sourceIP (not sure if linux would allow that or not)?

Also, I tried querying some of the relevant items from the zabbix server with snmpget and snmpwalk, and they all worked just fine.

[xxiii]

Its looking like this might actually be a get-bulk issue with the device in question.

I wish there was a way to specify via template (and whether it should apply to inherited/discovered templates or not) whether get-bulk should be used;

Or alternatively, via host groups (other than mass-update, which would only apply at the time and not to future hosts).

I've looked a little bit through the feature requests and there seems to be a lot of contention around how to configure usage of get-bulk.