On Wednesday 24 May 2006 21:32, Gergely CZUCZY wrote: > i've met a very strange issue with NATting. > > i've noticed that only every second outgoing SSH connections succeed, and > this was a bit strange. i've started a few, and tcp dumped them, applied > a filter for S/SA tcp flags, and i've got the following result: > > No. Time Source Destination Protocol > Info 31 4.513136 213.178.116.238 195.56.55.204 TCP > 53480 > ssh [SYN] Seq=0 Len=0 MSS=1460 WS=1 TSV=2969214 TSER=0 32 6.542201 > 213.178.109.103 195.56.55.204 TCP 56051 > ssh [SYN] > Seq=0 Len=0 MSS=1460 WS=1 TSV=2971243 TSER=0 73 8.293252 213.178.116.238 > 195.56.55.204 TCP 61535 > ssh [SYN] Seq=0 Len=0 MSS=1460 > WS=1 TSV=2972994 TSER=0 74 9.834288 213.178.109.103 195.56.55.204 > TCP 59672 > ssh [SYN] Seq=0 Len=0 MSS=1460 WS=1 TSV=2974535 > TSER=0 115 11.384353 213.178.116.238 195.56.55.204 TCP > 60708 > ssh [SYN] Seq=0 Len=0 MSS=1460 WS=1 TSV=2976085 TSER=0 > > take a look at the source address > now i've checked the interface configuration: > > # ifconfig tun0 > tun0: flags=8051<UP,POINTOPOINT,RUNNING,MULTICAST> mtu 1492 > inet 213.178.109.103 --> 195.70.32.11 netmask 0xffffffff > Opened by PID 208 > > for my information i looked them up: > 238.116.178.213.in-addr.arpa domain name pointer > caracas-4334.adsl.interware.hu. 103.109.178.213.in-addr.arpa domain name > pointer caracas-2407.adsl.interware.hu. > > so it appears that's just an other user-IP from my ISP's ADSL-pool. > > now the ppp.log looked like really interesting, here comes the point: > --- chop with axe here --- > May 24 18:08:02 beeblebrox ppp[208]: tun0: IPCP: IPADDR[6] changing > address: 213.178.116.238 --> 213. 178.109.103 > --- chop with axe here --- > as you can see, one source IP is the old one i had before, and the other on > is that i'm using currently. i've tried to re-read pf.conf with pfctl -f, > but that didn't helped, nor -d/-e (disabling and then enabling it). > > this solved it: > # pfctl -d > # pfctl -F nat > # pfctl -F state > # pfctl -F Sources > # pfctl -f /etc/pf.conf > # pfctl -e > > i'm using userland ppp service, as it seems from the tun0 interface. > > is this issue alread known, and is it really a bug, or i'm doing something > wrong? the pf.conf is availabe from here. this is my home gateway, it's > also a testbox, some kind of playground. > > uname -a: > FreeBSD beeblebrox.harmless.lan 6.1-STABLE FreeBSD 6.1-STABLE #0: Fri May > 19 14:25:03 CEST 2006 > [EMAIL PROTECTED]:/usr/obj/usr/src/sys/BEEBLEBROX i386 > > pf.conf: > http://phoemix.harmless.hu/pf.beeblebrox.conf
Try using: (tun0:0) in "to", "from" and "->" statements. The ":0" after the interface name will make sure that we don't use alias addresses on the interface. In fact this is a bug in ppp, but it was decided that it was non-trivial to fix it. I don't remember all the details, but http://www.freebsd.org/cgi/query-pr.cgi?pr=69954 was the PR back then. btw, you seem to be missing "()" around $if_ppp in the ftp-proxy rule. -- /"\ Best regards, | [EMAIL PROTECTED] \ / Max Laier | ICQ #67774661 X http://pf4freebsd.love2party.net/ | [EMAIL PROTECTED] / \ ASCII Ribbon Campaign | Against HTML Mail and News
pgpSRG2wBMkc9.pgp
Description: PGP signature
