Since some times, we noticed that one of our pppoe server running FreeBSD
had multiples tun interfaces stuck with a never stoping ppp process. The
logs clearly shows the user is logging on a new session with with the same
router... We ensured it's not a script kiddy trying to suck some ip
Then we tried to figure out what's the link with these users and we found
it was their router. In fact, the Linksys BEFSR41 ver. 4 (maybe others
too) makes the ppp process kinda freeze and lock a tun interface with it's
IP address until you kill the ppp process. Our ppp.conf file sets a max of
1472 for the mtu. Fortunately, we can find these locked interfaces by
their MTU which is at 1500 and abnormal for our tun interfaces.
For the moment we can only monitor it and tell our clients not to use this
kind of router but we think FreeBSD's ppp should manage this threat and
release the precious and limited ressource (IP addresses).
Note that this problem can take some hours to show... To reproduce it,
take a BEFSR41 router from Linksys, connect it to your FreeBSD server with
pppoed and wait 12 hours. It can happen anytime from 5 seconds to 12
hours..but in 12 hours, it really should happen ;)
Has anybody noticed this problem? We experienced it on release 5.x and 6.x.