netdev - Re: [PATCH] tcp: account SYN-ACK timeouts & retransmissions

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <201001121855.32018.opurdila@ixiacom.com>
Date:	Tue, 12 Jan 2010 18:55:31 +0200
From:	Octavian Purdila <opurdila@...acom.com>
To:	David Miller <davem@...emloft.net>
Cc:	netdev@...r.kernel.org, mpyatkovskiy@...acom.com
Subject: Re: [PATCH] tcp: account SYN-ACK timeouts & retransmissions

On Tuesday 12 January 2010 02:15:37 you wrote:
> From: Octavian Purdila <opurdila@...acom.com>
> Date: Tue, 12 Jan 2010 00:16:33 +0200
> 
> > On Friday 08 January 2010 03:25:07 you wrote:
> >> From: Octavian Purdila <opurdila@...acom.com>
> >> Date: Wed,  6 Jan 2010 22:10:58 +0200
> >>
> >> > Signed-off-by: Maksim Pyatkovskiy <mpyatkovskiy@...acom.com>
> >> > Signed-off-by: Octavian Purdila <opurdila@...acom.com>
> >>
> >> RETRANSSEGS is meant to count data segments retransmits, not pure
> >> control frames.
> >>
> >> Similarly for TCPTIMEOUTS, it's mean for data retransmit timeouts.
> >>
> >> If you overload these statistics with other similar events, they
> >> become less meaningful.
> >
> > But we do increment RETRANSSEGS & TCPTIMEOUTS for SYNs. This creates an
> > imbalance in client/server stats which makes things harder to diagnose.
> 
> Interesting.
> 
> Can you do me a favor and look into the code history and see
> if we used to account both sides?
> 

Sure. 

> I wonder if we accidently lost the statistic bumps when the
> generic inet connection socket layer was added.
> 

I don't think so because we are seeing this issue in 2.6.7 as well. I'll 
download the CVS history and drill down further.
--
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