Skip to contentSkip to conversation

Strange behavior of conntrack helper sip: rules do not always match as expected

DISCUSSION Jürgen Schmidt 1 participant 1 message May 18, 2016

lore
Share & tools
lore
#0 lore raw

Hello!

Given is an Asterisk server running on a Linux 3.10.101 machine with a global IP address (-> no NAT).

The following rules are given:

INPUT
=====

# SIP
iptables \
-A INPUT -i ppp0 -p udp -s 217.0.18.0/23 \
-d 217.232.238.65 --sport 5060 --dport 5060 -m conntrack \
--ctstate RELATED,ESTABLISHED -j ACCEPT

# RTP
iptables \
-A INPUT -i ppp0 -p udp -s 217.0.0.0/13 -d 217.232.238.65 \
--dport 11500:12000 -m conntrack --ctstate RELATED,ESTABLISHED \
-m helper --helper sip -j ACCEPT

OUTPUT
======

# SIP
iptables -A OUTPUT -p udp -o ppp0 -s 217.232.238.65 -d 217.0.18.0/23 \
--dport 5060 --sport 5060 -j ACCEPT

# RTP
iptables -A OUTPUT -p udp -o ppp0 -s 217.232.238.65 -d 217.0.0.0/13 --sport 11500:12000 -m conntrack --ctstate RELATED,ESTABLISHED \
-m helper --helper sip -j ACCEPT

This configuration works pretty fine as long as there is no ringgroup defined in Asterisk / FreePBX, which contains an *outbound* phone number. If the ringgroup contains an outbound phone number, the following problem happens, if the incoming call is accepted *locally*:

- Asterisk stops the signaling to the outbound phone through the trunk.
- Asterisk gives the call to the local phone.

Now the problem:
===============
All RTP packages between provider and Asterisk trunk are now blocked in both directions by netfilter.

To get it working again, these two additional rules are necessary for RTP handling *without* sip helper (but otherwise mostly the same as the RTP rules above besides slightly different states):

iptables -I INPUT -i ppp0 -p udp -s 217.0.0.0/13 -d 217.232.238.65 --dport 11500:12000 -m conntrack --ctstate NEW,ESTABLISHED -j ACCEPT

iptables -A OUTPUT -p udp -o ppp0 -s 217.232.238.65 -d 217.0.0.0/13 \
--sport 11500:12000 -m conntrack --ctstate ESTABLISHED -j ACCEPT

Analyzing w/ conntrack -E didn't show any abnormality - the relevant and correct rules have been generated - but obviously aren't matched by the standard rules w/ helper sip any more although the rules would have been completely suited. I too tested the sip helper rules using all states NEW,RELATED,ESTABLISHED - this doesn't match, too.

Do you have any idea about what's going on there?

Thanks for any hint,
kind regards,
Jürgen