Unpingable switch

[Quatermass]

What would be the first thing to check with a switch which intermittently can't be pinged, but is still basically functioning, i.e. clients still happily connected? It's a Netgear FSM726. Fixable with a reboot but something's not right...

[catch21]

Its management IP address has been changed? Can you get a console lead onto it and show the IP address that way?

[sonofsanta]

ARP table may be full/getting filled, so clients already in the table get connectivity, but when an additional client attempts to ping, the switch doesn't have space in the table to plan a route and so doesn't reply. I don't know if you can clear it in a Netgear, but if not, a reboot of the switch would sort it.

 

A good clue that this is happening is when some computers can ping a switch but not others. Those that can ping will already be in the ARP table.

 

You may be able to see the ARP table in the GUI/CLI; if it has a binary number of entries (e.g. 256, 512) then it's likely it's full.

[Duke5A]

It will respond to ping intermittently? When it does, is the ping reply time high? First thing I would think of is excessive CPU load if no other switch exhibits the same symptoms. If other switches do then I would be think broadcast storm. If the switch is manageable then go into the console and check the CPU load, and if it's high check to see what process is eating it. If its STP you could have a loop somewhere.

[Quatermass]

The fault itself is intermittent, but once it stops being pingable, that's it until it's rebooted. No pings get through. It's a managed switch but in that state you can't use web GUI or telnet to check stats. Next time it happens we'll try console direct and see if anything looks obviously wrong.

[jbunoomally]

I think you have to reboot to get the console to connect on the 726s. It might be worth leaving one connected on your next reboot to look at when it subsequently fails.

 

JB.