Would excessive errors cause a gateway interface to go offline?

[bad_joojoo]

Hi,

just wondering if there is any possibility that an excessive amount of traffic (may not exceed the capability in Mbps of the device so let’s say at 150Mbps as a general number) could impact an interface or a pair of interfaces so that the number of errors and CRC’s exceeds billions that the driver or kernel would take the interface offline.

The only way to successfully restore the service is to either reboot the node so that it rejoins the cluster or alternatively, reload the drivers (modprobe) to bring it back online. . . 

any thoughts/comments most welcome 😳. . .

[PhoneBoy]

Sounds like either a hardware issue or a bug in the relevant driver.
Either way I would engage with the TAC.

[bad_joojoo]

Hey PhoneBoy,

Many thanks for the thoughts, gratefully appreciated.

I have engaged with TAC. Unfortunately, they were a little slow in the reciprocal response and hence having to work through some elements on my own steam. Some interesting points you raise. Through digging around, I found that the messages log has actually captured the point when the two interfaces go down (they are not connected and not a bond), the log then goes on to identify a message with MNG_VETO. I've dug around into a few Intel driver articles and it looks like it may be similar to some of the known features/bugs with the same chipset. Watch this space.. 😁