Hi, AFAIK security in SEMS (using the SIP stack) is done directly via
iptables, this is, just allow SIP messages from our proxy.
The example B2BUA applications (as auth_b2b) don't allow setting an
outbound proxy so they contact directly the target. Linux conntrack
makes possible UDP responses to come back from the target during some
seconds (~120 sec).

So we have a security issue here:

- We secure SEMS by Iptables and just allow initial SIP UDP requests
comming from our proxy. We use a SEMS b2b application that contacts
directly the SIP target (any target/IP). After 120 seconds the target
tries to send BYE to SEMS but the UDP datagram is rejected/dropped by
Iptables since conntrack already deleted that UDP "connection".

- As an insecure solution we open SEMS SIP port to the world, so
anyone from anywhere could send a INVITE with "P-App-Name" and so to
SEMS bypassing our proxy (very insecure).

- A non always feasible solution would be adding the possible SIP
targets IP to Iptables ACCEPT list, but:
  - Maybe we cannot know which IP they will be (DNS can change...).
  - This "solution" would allow hacked from these target IP's.


IMHO the best solution would be allowing an "outbound" parameter to
SEMS applications, how easy is it? Thanks.

-- 
Iñaki Baz Castillo
<[EMAIL PROTECTED]>
_______________________________________________
Semsdev mailing list
[email protected]
http://lists.iptel.org/mailman/listinfo/semsdev

Reply via email to