On Thu, 17 Mar 2016 13:39:04 +0100 Morten Brekkevold morten.brekkevold@uninett.no wrote:
On Mon, 14 Mar 2016 10:19:34 +0200 Mattias Söderholm mattias.soderholm@malax.fi wrote:
Sorry, yes it's two rows. The correct chassi comes from ENTITY-MIB (this one is marked as DOWN i NAV) and the wrong empty chassi comes from SNMPv2-MIB (this one is ok in NAV).
[...]
For some reason, the plugin doesn't see that a chassis did come from ENTITY-MIB, and creates an extra chassis instance. We haven't properly tested how NAV behaves when multiple MIBs are used to source entity information; it seems it's time to take a closer look at just that.
I did some digging and added more debug statements to the code. While we still need to take a better approach at supporting multiple sources for entity information, it seems to me that your problem must come from the fact that there is some issue collecting information from the ENTITY-MIB on this switch of yours.
This is why the chassis previously seen in the ENTITY-MIB is marked as down (the `entity` plugin can no longer see it), and why the `system` plugin creates a "new" chassis. It seems the code will delete the SNMPv2-MIB sourced chassis once the `entity` plugin finds its chassis info again.
Does `ipdevpoll.log` contain any messages about your switch, other than the regular "I ran some job"-messages?