Mysterious switch to IPv6

[Mads]

We have added a line to the up case in the ifplugd.script that runs ethtool to advertise 10/100 Full only, and it was working fine (this was the only change done to the out of the box script). Then we changed the mode code in the call to advertise 10 Full Only, restarted - and something went very wrong... The link was still there, but still at 100 Full, but the device had now begun to only respond to IPv6 for some reason. We were able to connect with SSH on IPv6 and configure an IPv4 address, and push an image of the device.

The scenario is rather complex and we did not have time to gather much details, but it would be interesting to know if anyone would know anythin g that could have triggerd this. We speculated that perhaps the script file had been corrupted, and that for some reason the system was then defaulting to an IPv6 only setup...but that seems like a strange thing to do by default...or?

[ScotSalmon]

Did you happen to keep a copy of the possibly-corrupt image?

[AdiPaqq]

Could you post ethtool command that you added to the script?

[Mads]

I have been able to reproduce this once now. Editing the file caused the device a second time to revert to IPv6 only traffic. I do not think it's the ethtool command, but the fact that we edit it on a Windows machine and copy it back (the access rights are still OK though). But the command we have added to the default script is as follows:

case "$2" in

   up)

      # On link-up, we're not guaranteed to be on the same network as

      # we were before, so we should reprobe for an address if it's

      # automatically configured.

            /usr/sbin/ethtool -s $INTERFACE advertise 0x002