Skip to contentSkip to conversation

separation of sysctl and tcp-window-tracking patch?

DISCUSSION [email protected] 4 participants 5 messages Nov 1, 2002 – Dec 13, 2002 netfilter

lore
Share & tools
lore
#0 lore raw

There seems to be varied needs for netfilter specific /proc entries. The tcp-window-tracking patch seems to me to have placed a /proc entry point for netfilter most sanely at /proc/sys/net/ipv4/netfilter/. I would also like to have a sysctl/proc node to diddle with tunables in my Amanda conntracking module.

Specifically the default of 180 seconds for a UDP stream is not long enough on the control (master in netfilter parlance) connection of an Amanda session, so I would like to override the master's timeout value in my conntrack helper. I have tested my theory with a hard coded value and it work. Now I would like to make it tunable by the user/administrator.

I suppose I could pass an argument while loading the module, but using sysctl (and/or /proc) just seems so much more sane. But I digress, greatly.

I am wondering if Jozsef and/or Harald would like to separate out the creation of the /proc/sys/net/ipv4/netfilter/ node into a separate patch for the rest of us to use, rather than having to make tcp-window-tracking a prerequisite just to get a proc entry to use.

Thots?
b.

Signature · 1 line
Brian J. Murrell
#1 lore raw

I was catching up on back netfilter-devel messages and saw this.

I absolutely agree. I would *really* like to be able to tune various netfilter-related settings via /proc, regardless of my opinions on the tcp-window-tracking patch.

(My specific need is related to DNS service: namely, in many cases, 30 seconds to establish a UDP session simply isn't enough time to permit a reply to an outstanding DNS query. I want to be able to up that timeout to something closer to 60 or 120 seconds.)

Signature · 3 lines
James Ralston, Information Technology
Software Engineering Institute
Carnegie Mellon University, Pittsburgh, PA, USA
Quoted 25 lines · On 2002-10-31 at 21:12:36-0500 [email protected] wrote:

There seems to be varied needs for netfilter specific /proc entries. The tcp-window-tracking patch seems to me to have placed a /proc entry point for netfilter most sanely at /proc/sys/net/ipv4/netfilter/. I would also like to have a sysctl/proc node to diddle with tunables in my Amanda conntracking module.

Specifically the default of 180 seconds for a UDP stream is not long enough on the control (master in netfilter parlance) connection of an Amanda session, so I would like to override the master's timeout value in my conntrack helper. I have tested my theory with a hard coded value and it work. Now I would like to make it tunable by the user/administrator.

I suppose I could pass an argument while loading the module, but using sysctl (and/or /proc) just seems so much more sane. But I digress, greatly.

I am wondering if Jozsef and/or Harald would like to separate out the creation of the /proc/sys/net/ipv4/netfilter/ node into a separate patch for the rest of us to use, rather than having to make tcp-window-tracking a prerequisite just to get a proc entry to use.

Thots?
b.

#2 lore raw
Quoted 8 lines · James Ralston

I absolutely agree. I would *really* like to be able to tune various netfilter-related settings via /proc, regardless of my opinions on the tcp-window-tracking patch.

(My specific need is related to DNS service: namely, in many cases, 30 seconds to establish a UDP session simply isn't enough time to permit a reply to an outstanding DNS query. I want to be able to up that timeout to something closer to 60 or 120 seconds.)

Please note, that the timeout settings via /proc introduced in the tcp-window-tracking patch are global. You cannot raise the UDP timeout values just for DNS.

Also, we have to handle the backward compatibility issue of /proc/sys/net/ipv4/ip_conntrack_max, if the introduction of /proc/sys/net/ipv4/netfilter/ is accepted.

Regards,
Jozsef
-
E-mail : [email protected], [email protected] PGP key : http://www.kfki.hu/~kadlec/pgp_public_key.txt
Address : KFKI Research Institute for Particle and Nuclear Physics

          H-1525 Budapest 114, POB. 49, Hungary
#3 lore raw
Quoted 3 lines · Jozsef Kadlecsik

Please note, that the timeout settings via /proc introduced in the tcp-window-tracking patch are global. You cannot raise the UDP timeout values just for DNS.

That's true. But on dedicated nameservers (the situation I care about), I don't care if all UDP connections on the system receive the higher limit.

Quoted 3 lines

Also, we have to handle the backward compatibility issue of /proc/sys/net/ipv4/ip_conntrack_max, if the introduction of /proc/sys/net/ipv4/netfilter/ is accepted.

True.

Signature · 3 lines
James Ralston, Information Technology
Software Engineering Institute
Carnegie Mellon University, Pittsburgh, PA, USA
#4 lore raw
In reply to James Ralston
Quoted 4 lines · James Ralston

(My specific need is related to DNS service: namely, in many cases, 30 seconds to establish a UDP session simply isn't enough time to permit a reply to an outstanding DNS query. I want to be able to up that timeout to something closer to 60 or 120 seconds.)

If your DNS servers serve zones then make them send requests from the udp/53 port, then late replies are accepted as requests by the filter.

Denis Ducamp.