[Observium] Device up/down conditions

Can someone help me understand the algorithm/conditions that will cause a device to be flagged up/down in Observium?

Reason I ask is I'm seeing instances where devices will get flagged as 'down' however when I login to the Observium host and manually test via fping/snmpwalk, I get valid responses back. The devices come back a few minutes later.

I don't see anything in observium.log aside from poller completion results, so I could use some more info/ideas to help debug why these devices keep going offline intermittently.

Cheers, -Chris