Re: connbytes patch eliminated

[Michael McCallister <mikemc-netfilter@contactdesigns.com>]

Andy Furniss wrote the following on 05/04/2006 08:57 AM:

> Michael McCallister wrote:
>
>> Hello,
>>
>> First, a warning - I am a newbie to netfilter, so I may ask some 
>> stupid questions here.  I believe the connbytes patch offers exactly 
>> what I am looking for - granted it is listed as experimental, but I 
>> am willing to test it out since if offers the functionality I think I 
>> need - mainly depriotizing bulk transfers.  I am concerned because it 
>> appears it was dropped from the main linux kernel, the last kernel I 
>> found with it was linux-2.6.15.7.  Also, it is not in pom-ng - at 
>> least I could not find it in pom snapshots or cvs 
>> (http://svn.netfilter.org/cgi-bin/viewcvs.cgi/trunk/patch-o-matic-ng/).  
>> So I get the impression there may be plans to get rid of the 
>> connbytes patch.  The latest iptables still does checks for it though 
>> "[ -f $KERNEL_DIR/net/ipv4/netfilter/ipt_connbytes.c ] && echo 
>> connbytes".  Was there a decision that it was not suitable anymore 
>> and it is being eliminated in favor of another approach?  If so, any 
>> advice as to the new approach is greatly appreciated.  Also, if it 
>> was dropped from the kernel/pom because it was highly unstable and 
>> caused system crashes - that would be great information too :-)
>>
>> Thanks for any help - my apologies if I missed something obvious.
>> Michael
>>
>
> Still there new name - the whole netfilter config has changed since I 
> last did one.
>
> [andy@amd ~]$ grep -i connbytes /boot/config-2.6.16.11
> CONFIG_NETFILTER_XT_MATCH_CONNBYTES=m
>
> Andy.

Thanks Andy,

I can see that my problem is I need iptables from CVS.  I guess things 
have moved around in the kernel recently: 
http://svn.netfilter.org/cgi-bin/viewcvs.cgi/trunk/iptables/extensions/.connbytes-test?rev=6579&view=markup

I generally try avoid building custom kernels (I'm a "rpm -ivh 
kernel-xxx.rpm" kind of guy) so I didn't know things changed that 
often.  Thanks again for the insight.

Michael