Large scale NAT with PF - some weird problem

[Ian FREISLICH]

Milan Obuch wrote:
> As a first step, I did small upgrade, so now I run FreeBSD 9.3-STABLE
> #0 r284695: Mon Jun 22 08:55:29 CEST 2015.
> 
> I still see the issue, but I found simpler workaround when bad state
> ocurs - using
> 
> pfctl -k <ip.of.affected.client>
> pfctl -K <ip.of.affected.client>
> 
> in this order seems to remedy the issue for this one affected client
> without affecting other clients. This still does not solve the problem,
> just eases the reaction.

How is your NAT rule defined?  I had a closer look at the way I did it:

nat on vlan46 from 10.8.0.0/15 to !<on-our-net> -> xx.xx.xx.xx/24 round-robin sticky-address

I think you may be missing the "round-robin" that spreads the mapping
over your pool.  The manual says that when more than 1 address is
specified, round-robin is the only pool type allowed, it does not
say that when more than 1 address is specified this is the default
pool option.

You can check your state table to see if it is indeed round-robin.

#pfctl -s sta |grep " ("
...
all tcp a.b.c.d:53802 (10.0.0.220:42808) -> 41.246.55.66:24       ESTABLISHED:ESTABLISHED
all tcp a.b.c.e:60794 (10.0.0.38:47825) -> 216.58.223.10:443       ESTABLISHED:FIN_WAIT_2

If all your addresses "a.b.c.X" are the same, it's not round-robin
and that's your problem.

Ian

-- 
Ian Freislich