[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <A86A7D0E-F304-4322-8204-8F352E67D4E2@earthlink.net>
Date: Thu, 3 Jun 2010 18:13:19 -0700
From: Mitchell Erblich <erblichs@...thlink.net>
To: Hagen Paul Pfeifer <hagen@...u.net>
Cc: Ivan Novick <novickivan@...il.com>, netdev@...r.kernel.org
Subject: Re: 200 millisecond timeouts in TCP
On Jun 3, 2010, at 4:10 PM, Hagen Paul Pfeifer wrote:
> * Ivan Novick | 2010-06-03 15:37:24 [-0700]:
>
>> Using tcpdump and systemtap I am seeing that sometimes retransmission
>> of data is sent after waiting 200 milliseconds. However sometimes
>> retransmissions happen quicker.
>
> Quicker as 200ms? Conservatively the minimum TCP RTO should be 1s (rfc2988),
> Linux differs from this default and define the minimum RTO to 200ms:
>
> #define TCP_RTO_MIN ((unsigned)(HZ/5))
>
> Can you post the tcpdump traces where the relevant retransmission is recognizable?
>
>> Also do you know if the timeout numbers for TCP are configurable parameters?
>
> Some values are documented in Documentation/networking/ip-sysctl.txt, you can
> find the relevant timer implementation in ipv4/tcp_input.c and the definition
> of TCP_RTO_MIN in include/net/tcp.h.
>
>
> Hagen Paul Pfeifer
>
>
> --
> Hagen Paul Pfeifer <hagen@...u.net> || http://jauu.net/
> Telephone: +49 174 5455209 || Key Id: 0x98350C22
> Key Fingerprint: 490F 557B 6C48 6D7E 5706 2EA2 4A22 8D45 9835 0C22
Group, from page 7 of the Linux TCP Congestion Control doc.
"Linux TCP deviates from the IETF specification by al- lowing a minimum limit of 200 ms for the RTO."
Mitchell Erblich--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists