And by "disguise" i've meant "distinguish" :-)
T.
Dne 2016-12-06 06:54, mudrunka@xxxxxxxxx napsal:
Hello,
currently in iptables i can set NOTRACK (-j CT --notrack) only for
OUTPUT and PREROUTING. Because the routing decision is made after the
conntracking.
I need stateful firewall on INPUT, but conntrack on FORWARD is
performance drawback for me. And i can imagine that someone might have
exact oposite of this problem.
When i want to enable conntrack for input, but not for forwarding, i
have to list all the ip adresses on local interfaces. This is big
administrative PITA for several reasons. i have routers with hundreds
of vlans and each of these vlans have multiple ip adresses - both ipv4
and ipv6. Disabling conntrack for FORWARD only means listing all of
them in PREROUTING to disguise INPUT traffic from the FORWARDed one.
This is annoying and prone to error.
It would be super useful if one can simply use "-j CT --notrack" in
INPUT and FORWARD. (it already works in OUTPUT)
If it's impossible to postpone conntrack after routing decision, it
might be possible to add some macro that would match any of local
adresses that are currently on any of interfaces. like "--src local"
or "--dst local". Currently i am using ipset filled by cron script
with all these adresses parsed from "ip a s". But that's far from
being elegant or reliable.
I am planning to switch over to nftables, so it might be another
solution...
Is this planned to be fixed in nftables? If not can you please
consider fixing it?
Thanks
Best regards
Tomas Mudrunka
--
To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
