Bug#477174: #477174

yes, I used iptables. I had this experience on debian/etch 4.0 r0 to r3. Since then I removed debian and am using rhel 4, on which this problem doesn't occur.

Seems like this problem is debian specific. It hasn't happened on rhel 3 or rhel 4 on the same box.

Also, I have not tried upgrading to later kernels since then. I'm using the stock, out of the box kernels that come with rhel ws3 and rhel 4

M. Istehbab

On Sat, Sep 27, 2008 at 9:30 PM, M. Istehbab <hariskhan@gmail.com> wrote:

I filed a bug few weeks back. Detalis can be seen at;

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=477174

I also wrote to kernel / netdev mailing list. That post can be seen here;

http://marc.info/?l=linux-netdev&m=120920147425850&w=2

What happened was, I had squid on a transparent linux box, which acted as default gateway for a 60 box network.

At random times, squid would suddenly not get connections from LAN clients or squid would not GET the connection attempts from clients.

Nothing would show up in tcpdump output against LAN clients who would try to hit squid normally to browse the internet.

So, the connection from LAN clients was not being registered at any level on the linux box with the mentioned kernel versions. And this happened at random time.

M. Istehbab

On Sun, Sep 14, 2008 at 1:57 AM, Bastian Blank <waldi@debian.org> wrote:

Hi

Please show what you mean with "random TCP connection failure". For
example with a dump from tcpdump.

Do you use netfilter/iptables?

Bastian

--
Military secrets are the most fleeting of all.
-- Spock, "The Enterprise Incident", stardate 5027.4