DNS lookup in FC2 still slow.

Ben Vitale bsvitale at comcast.net
Wed Aug 4 00:14:25 UTC 2004


Jack Bowling wrote:

>On Mon, Aug 02, 2004 at 08:53:58PM -0400, Ben Vitale wrote:
>  
>
>>Hi,
>>   Yes, I realize this is an often asked question, and for that I 
>>apologize.  I have the two lines..
>>
>>alias net-pf-10 off
>>alias ipv6 off
>>
>>   ..in my /etc/modprobe.conf, and I have restarted several times since 
>>they were put in place.  They have had no effect, and I don't think IPv6 
>>is the culprit anyway.. my ifconfig shows all regular IP addresses. 
>>   I checked the ping times of my nameservers, and the one at the top 
>>of /etc/resolv.conf has the best times - on average, 8.5ms.  I would 
>>assume this is just a Comcast provider problem, but this was not an 
>>issue until I upgraded to FC2.
>>    
>>
>
>Ben - I wonder if it is a problem with ECN on the nameserver or somewhere
>on the route? First check to see if you have ECN turned on in your kernel:
>
>cat /proc/sys/net/ipv4/tcp_ecn
>
>If you get a 1 back, it is turned on. As a test, turn it off by:
>
>echo 0 > /proc/sys/net/ipv4/tcp_ecn
>
>and see if the slow ping still happens.
>
>BTW, Ian Gulliver wrote a useful program called ecncheck which tests ECN 
>along a route. There is also a companion program called ecnmx which
>compiles from the same tarball that does the same for mailservers. Look for
>it on freshmeat.net
>
>
>  
>
Without any changes to kernel 2.6.6-1.435.2.3,

[bvitale at vandelay ~]$ cat /proc/sys/net/ipv4/tcp_ecn
0
[bvitale at vandelay ~]$


[bvitale at vandelay ~/ecncheck]$ ecncheck 68.48.0.6 53
With ECN:    Connection accepted by ns02.rtchrd01.md.comcast.net 
[68.48.0.6] at hop #7
Without ECN: Connection accepted by ns02.rtchrd01.md.comcast.net 
[68.48.0.6] at hop #7
WARNING: Host doesn't support ECN but fails gracefully
[bvitale at vandelay ~/ecncheck]$

Not sure how to interpret that?

Also, do you think 8.5ms is "slow" for a DNS server ping?

Ben







More information about the fedora-list mailing list