Google Chrome blocks RWIN ?

**Rollingstone** · 07-10-09, 04:36 AM

Default TCP Receive Window (RWIN) = 65536
RWIN Scaling (RFC1323) = 1 bits (scale factor: 2^1=2)
Unscaled TCP Receive Window = 32768

RWIN is not fully optimized (even though it is a comparatively large number). The unscaled RWIN value is lower than it should be. Also, RWIN being close to and above 65535 does not justify the header overhead of enabling TCP 1323 Options. You might want to use one of the recommended RWIN values below.

If you use Google Chrome, TCP Analyzer will show TCP Window 65536 while in Firefox + IE + Safari + Opera, it displays an another number (same 127776 in my case) !

What a strange issue !

**akbarri** · 07-10-09, 04:25 PM

but my TCP Analyzer Result are normal

Chrome 2.0.172.33 = Firefox 3.5 = IE8
(i'm not checking @ Opera & Safari)

What is the version of that Google Chrome anyway??

**Rollingstone** · 07-10-09, 08:30 PM

What is the version of that Google Chrome anyway?

Chrome 3.0.187.1 and Chrome 2.0.174.0 !

Win XP SP2 !

**Streetspirit** · 07-11-09, 12:27 AM

Originally Posted by Rollingstone

Chrome 3.0.187.1 and Chrome 2.0.174.0 !

Win XP SP2 !

This is truly odd. I just got the same thing, a 64K TCP RWIN. It was however, correctly analyzed by DSLReports' JAVA based tweak analyzer however I am seeing the 64K window here using the TCP Analyzer.

Perhaps the fault lies with some interaction between Chrome's Javascript engine and the code here. Definitely something to watch...

I'm using XP SP3 / Chrome 3.0.193.0 (Latest development version.)

Ref:

DSLReports results: http://www.dslreports.com/tweakr/blo...nXP&via=normal

Speedguide results: " Default TCP Receive Window (RWIN) = 65536
RWIN Scaling (RFC1323) = 1 bits (scale factor: 2^1=2)
Unscaled TCP Receive Window = 32768

RWIN is not fully optimized (even though it is a comparatively large number). The unscaled RWIN value is lower than it should be. Also, RWIN being close to and above 65535 does not justify the header overhead of enabling TCP 1323 Options. You might want to use one of the recommended RWIN values below."

--
Rgds

**Streetspirit** · 07-11-09, 12:32 AM

@ Rollingstone ,

I'm curious what a JAVA based analyzer would report on your system. Could you please visit http://www.dslreports.com/tweaks and test with the Java tweak test tool. Many thanks in advance.

--
Regards

**Rollingstone** · 07-11-09, 03:50 AM

It was however, correctly analyzed by DSLReports' JAVA based tweak analyzer however I am seeing the 64K window here using the TCP Analyzer.

Same thing here !

RWIN correctly analyzed by DSLReports Analyzer & Nitro NDT tested in Google Chrome 3 ! But TCP Analyzer didn't !

**Streetspirit** · 07-11-09, 01:30 PM

Originally Posted by Rollingstone

Same thing here !

RWIN correctly analyzed by DSLReports Analyzer & Nitro NDT tested in Google Chrome 3 ! But TCP Analyzer didn't !

I'm fairly sure it's not accurate and is a script interaction of some sort. However, it would be nice if the powers that be here take a look.

Here are my results with Chrome and with Firefox:

BROWSER: Firefox 3.0.11

Code:

« SpeedGuide.net TCP Analyzer Results » 
Tested on: 07.11.2009 11:20 
IP address: 24.190.xx.xxx 
Client OS: Windows XP 
 
TCP options string: 020405b40103030401010402 
MSS: 1460 
MTU: 1500 
TCP Window: 921600 (NOT multiple of MSS) 
RWIN Scaling: 4 bits (2^4=16) 
Unscaled RWIN : 57600 
Recommended RWINs: 64240, 128480, 256960, 513920, 1027840 
BDP limit (200ms): 36864kbps (4608KBytes/s)
BDP limit (500ms): 14746kbps (1843KBytes/s) 
MTU Discovery: ON 
TTL: 116 
Timestamps: OFF 
SACKs: ON 
IP ToS: 00000000 (0)

Browser: Google Chrome v3.0.193.0

Code:

« SpeedGuide.net TCP Analyzer Results » 
Tested on: 07.11.2009 11:19 
IP address: 24.190.xx.xxx 
Client OS: Windows XP 

TCP options string: 020405b40103030101010402 
MSS: 1460 
MTU: 1500 
TCP Window: 65536 (NOT multiple of MSS) 
RWIN Scaling: 1 bits (2^1=2) 
Unscaled RWIN : 32768 
Recommended RWINs: 64240, 128480, 256960, 513920, 1027840 
BDP limit (200ms): 2621kbps (328KBytes/s)
BDP limit (500ms): 1049kbps (131KBytes/s) 
MTU Discovery: ON 
TTL: 116 
Timestamps: OFF 
SACKs: ON 
IP ToS: 00000000 (0)

Hope that the admins look at this problem.

--
Cheers-

**akbarri** · 07-11-09, 01:41 PM

Originally Posted by Rollingstone

Chrome 3.0.187.1 and Chrome 2.0.174.0 !

Win XP SP2 !

Both are Development Version, maybe u can this issue to Google Chrome Forum
http://www.google.com/support/forum/p/Chrome
http://googlechromecommunity.com
http://chromespot.com

**Rollingstone** · 07-11-09, 08:51 PM

Both are Development Version, maybe u can this issue to Google Chrome Forum
http://www.google.com/support/forum/p/Chrome
http://googlechromecommunity.com
http://chromespot.com

But tested in Java based Analyzer , RWIN in Chrome correctly analyzed ! (same as other browsers)

**Philip** · 07-12-09, 03:36 AM

Interesting, I'll take a look.

Can you please PM me or post here the full URL of the Analyzer resuts page under the different browers ?

Thanks

**Rollingstone** · 07-12-09, 04:02 AM

Did you mean TCP Analyzer or Java-based Analyzer ?

OK, here we go :

TCP Analyzer URLs under different browsers :

Chrome 3.0.187.1

http://www.speedguide.net/analyzer.php?
DATA_OFFSET=40&TCP_Options_string=020405ac0103030101010402&IP_MTU_DISCOVER=1&WIN=32768&RWIN=65536&MSS=1452&SCALE=1&TTL=42&TSOPT=0&SACK_PERM=1&IP_TOS=0&IP=123.21.x&timestamp=1247389731

Firefox 3.0.11

http://www.speedguide.net/analyzer.php?
DATA_OFFSET=40&TCP_Options_string=020405ac0103030101010402&IP_MTU_DISCOVER=1&WIN=63888&RWIN=127776&M SS=1452&SCALE=1&TTL=42&TSOPT=0&SACK_PERM=1&IP_TOS=0&IP=123.21.x&timestamp=1247389745

Opera 9.64

http://www.speedguide.net/analyzer.php?
DATA_OFFSET=40&TCP_Options_string=020405ac0103030101010402&IP_MTU_DISCOVER=1&WIN=63888&RWIN=127776&M SS=1452&SCALE=1&TTL=42&TSOPT=0&SACK_PERM=1&IP_TOS=0&IP=123.21.x&timestamp=1247389775

IE 6

http://www.speedguide.net/analyzer.php?
DATA_OFFSET=40&TCP_Options_string=020405ac0103030101010402&IP_MTU_DISCOVER=1&WIN=63888&RWIN=127776&M SS=1452&SCALE=1&TTL=42&TSOPT=0&SACK_PERM=1&IP_TOS=0&IP=123.21.x&timestamp=1247389820

Safari 4

http://www.speedguide.net/analyzer.php?
DATA_OFFSET=40&TCP_Options_string=020405ac0103030101010402&IP_MTU_DISCOVER=1&WIN=63888&RWIN=127776&M SS=1452&SCALE=1&TTL=42&TSOPT=0&SACK_PERM=1&IP_TOS=0&IP=123.21.x&timestamp=1247389920

**Philip** · 07-12-09, 04:52 AM

Thanks for the quick reply...

From this, it seems that Chrome is doing somehing different during the TCP handshake betwen the Analyzer server and the browser, I'll have to take a closer look but it will take some time, it is not trivial.

**northraleighguy** · 10-25-11, 02:50 PM

Ah, seems if I found the culprit of my issue

This seems to effect only Server 2003 in our situation.

When using IE on Server 2003

« SpeedGuide.net TCP Analyzer Results »
Tested on: 10.25.2011 15:46
IP address: xx.xx.xxx.xx
Client OS: Windows Server 2003

TCP options string: 020405ac0103030201010402
MSS: 1452
MTU: 1492
TCP Window: 261360 (multiple of MSS)
RWIN Scaling: 2 bits (2^2=4)
Unscaled RWIN : 65340
Recommended RWINs: 63888, 127776, 255552, 511104, 1022208
BDP limit (200ms): 10454kbps (1307KBytes/s)
BDP limit (500ms): 4182kbps (523KBytes/s)
MTU Discovery: ON
TTL: 52
Timestamps: OFF
SACKs: ON
IP ToS: 00000000 (0)

Then using Google Chrome 14.0.835.202 m Windows

« SpeedGuide.net TCP Analyzer Results »
Tested on: 10.25.2011 15:11
IP address: xx.xx.xxx.xx
Client OS: Windows Server 2003

TCP options string: 020405ac0103030101010402
MSS: 1452
MTU: 1492
TCP Window: 65536 (NOT multiple of MSS)
RWIN Scaling: 1 bits (2^1=2)
Unscaled RWIN : 32768
Recommended RWINs: 63888, 127776, 255552, 511104, 1022208
BDP limit (200ms): 2621kbps (328KBytes/s)
BDP limit (500ms): 1049kbps (131KBytes/s)
MTU Discovery: ON
TTL: 52
Timestamps: OFF
SACKs: ON
IP ToS: 00000000 (0)

Other OS with Google Chrome work fine XP XP X64

**Philip** · 12-02-11, 10:14 AM

Hmm, note the two different "TCP Option strings"... The problem seems to be different RWIN value advertised by your browser, not a different value reported by the tool.

Seems to me that Chrome is changing TCP options, that is overstepping what a browser should be doing if that is the case, IMHO...

Thanks for sharing that.