xt_connlimit 20070620_2

[Patrick McHardy]

Jan Engelhardt wrote:
> On Jun 22 2007 14:42, Patrick McHardy wrote:
> 
>>>Conntrack states are needed. For TCP this is TCP_CONNTRACK_TIME_WAIT.
>>>Hence I think all states that could possibly exist (as far as conntrack is
>>>concerned) are listed in netfilter/nf_conntrack_<proto>.h
>>
>>Indeed, its in include/linux/netfilter/nf_conntrack_sctp.h.
>>I'm wondering why its looking at the states though, a conntrack
>>in TIME_WAIT state is still a conntrack.
> 
> 
> The original author thought that a TIME_WAIT socket does not count
> towards the limit anymore (interpreting the code). Seems reasonable.
> Does SCTP does not have the concept of TIME_WAIT - does SCTP end
> with that is known in TCP land by "LAST_ACK"?
> If so, then all is well with SCTP and no extra state check if needed for SCTP.


I think the shutdown states correspond to TIME_WAIT/LAST_ACK.

>>>Yes there is a user-administrator race. For example,
>>> * host opens N connections (and hits the connlimit limit) and new
>>>   connections are rejected
>>> * administrator reloads ip tables
>>> * now, the next N that are seen -- which may not even be the *original*
>>>   connections -- are allowed, so the older ones could get magically
>>>   disconnected.
>>
>>That sucks a bit.
> 
> 
> Well give me your stance on this. Do I have to change that now?


If you have a good idea how to fix that ..