lagg problems on diskless client

[Beeblebrox]

@Peter:
My DC nodes already mount /tmp as tmpfs (a ramdisk). I have hence
modified the script and defined
lagg_tmp:=/tmp/mnt and added the line
mkdir "$lagg_tmp" before the md creation.

lagg still breaks, so I commented out the lines following "Actually
flip the interface". Checking the DC status shows that /rescue
contents have been copied to /tmp/mnt, so the script has worked up to
that point. When I manually try
cd /tmp/mnt && ./ifconfig re0 delete, I get the familiar "cannot
delete re0" message.

Regards.