Linux Netfilter discussions
 help / color / mirror / Atom feed
From: "Holger Brueckner [c]" <hb@ciphirelabs.com>
To: netfilter@lists.netfilter.org
Subject: Re: problem with conntrack loosing state [signed]
Date: Tue, 17 May 2005 12:30:11 +0200	[thread overview]
Message-ID: <1116325811.3383.3.camel@localhost> (raw)
In-Reply-To: <1114533395.2969.39.camel@localhost>


just for the record:

problem has been solved by running kernel > 2.6.12-rc3

h.


On Tue, 2005-04-26 at 18:36 +0200, Holger Brueckner [c] wrote:
> hello,
> 
> (please cc, i'm not a regular on the list)
> 
> we're experiencing some strange problems with the conntrack engine
> loosing state. following setup:
> 
> fw with several interfaces
> kernel 2.6.11.X
> iptables v1.2.11 (debian)
> 
> all ips have a /32 netmask so that every traffic is routed through the
> firewall. this is assured by corresponding vlan setup on the switches.
> 
> FORWARD is:
> Chain FORWARD (policy DROP)
> target     prot opt source               destination
> DROP       all  --  anywhere             anywhere            state
> INVALID
> ACCEPT     all  --  anywhere             anywhere            state
> RELATED,ESTABLISHED
> .....
> LOGDROP    all  --  anywhere             anywhere
> 
> eventually after a day or two pakets pakets which should be matched by
> established coming in from the same interface as they go out will get
> dropped and logged.
> 
> e.g.     srv1 --+-- fw -- srv3
>          srv2 --|
> 
> "established" packet from srv1 to srv2 will get dropped after some days.
> it looks like the syn flags don't trigger the conntrack engine although
> the syn "pakets" go through the fw as expected, only pakets with no syn
> flag set get dropped.
> while this is the case the fw works perfectly for host which are not on
> the same interface. so conntrack for connections from srv1 to srv3 or
> srv2 to srv3 work as expected. rebooting the firewall is the only
> solution to the problem.
> 
> there's not very much load on the server yet, last time i checked there
> were about 250 conntrack entries. it looks like this might be realted to
> Daniel Wittembergs "NAT stops working (more)" thread, at least the
> symptoms are quite similar.
> 
> any suggestions to further debug this ? we just upgraded to 2.6.12-rc3
> to see if this is solved. if not we will downgrade and see if this
> happens again.
> 
> holger brueckner
> 
> 
> 
> 



--
------------------------ [ SECURITY NOTICE ] ------------------------
To: netfilter@lists.netfilter.org.
For your security, hb@ciphirelabs.com
digitally signed this message on 17 May 2005 at 10:29:37 UTC.
Verify this digital signature at http://www.ciphire.com/verify.
------------------- [ CIPHIRE DIGITAL SIGNATURE ] -------------------
Q2lwaGlyZSBTaWcuAVduZXRmaWx0ZXJAbGlzdHMubmV0ZmlsdGVyLm9yZwBoYkBjaXBoa
XJlbGFicy5jb20AZW1haWwgYm9keQAJBgAAfAB8AAAAAQAAAJHHiUIJBgAArwIAAgACAA
IAIPIh6RkYmeZcQRIJZYy7b0sQ3bMYMvm8+Po9XBPY0pNnAQAPjqtled7ohMXyims3EKz
yrw6OVIEN+mt2h+UbCMAsTCv2wOnTCiMfYPjb8mnpo4EOy0ccziFlSwuEZM4elmcPU2ln
RW5k
--------------------- [ END DIGITAL SIGNATURE ] ---------------------



      reply	other threads:[~2005-05-17 10:30 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2005-04-26 16:36 problem with conntrack loosing state [signed] Holger Brueckner [c]
2005-05-17 10:30 ` Holger Brueckner [c] [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1116325811.3383.3.camel@localhost \
    --to=hb@ciphirelabs.com \
    --cc=netfilter@lists.netfilter.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox