Intermittent "No response to DHCP broadcast" alerts but DHCP server is fine; any ideas?

[davebainum]

Hi all,

We've been getting some intermittent alerts from one of our ThunderBolt locations for a while - it's Event ID 33029872, "No responses to DHCP Broadcast".

Tends to happen several times a day.  Strangely, when you go into TBolt and then click the "Test DHCP" monitor, it responds saying that everything is fine.  We've tried the usual "easy fixes" such as rebooting the Thunderbolt, etc., rediscovering topology - doesn't seem to change anything.

I'm almost tempted to change out the PLG1000 at this point and see if that makes a difference.  We haven't seen this behavior at any other of our TBolt or OnPlus sites to date...

Thanks in advance for any suggestions/advice.

-- Dave Bainum, PMP* (dbainum@ritetech.net)

RiteTech LLC / www.ritetech.net / Tel. +1 (703) 561-0607

[*PMP=PMI Certified Project Management Professional]

[Michael Holloway]

Hi Dave,

The default threshold for this monitor has been adjusted in the next release to prevent false positives like this. We've seen a few networks where the DHCP responses from the server were taking more than 10 seconds to be returned, but only intermittently, and so would occasionally set off this monitor. The new default threshold will wait 20 seconds for a response from a DHCP server.

Like other monitors, the service is checked every 5 minutes, but as soon as a failure is detected we recheck the service every 60 seconds until we reach the maximum number of check attempts, at which point we consider the failure a true event. Most monitors are configured for 2 check attempts before considering the event to be true, however the DHCP and WAN Bandwidth Performance monitors check a total of 3 times to prevent brief spikes in network activity from creating a false positive.

You can change the current default timeout of 10 seconds by editing the OnPlus Agent device, go to the 'Monitors' tab, and open up the 'DHCP Service' monitor. I recommend setting this to 20 seconds rather than the default of 10. FYI we've also corrected the text for this sensor to read as seconds rather than milliseconds.

-mike