Hi, to your question regarding the Http/1.0:
I am using protocol="HTTP/1.1".
I send the message using this command:
wget http://172.18.206.103:8089/bundles/AOS701b318-hd.tg

I have no idea why the log shows HTTP/1.0.

-----Original Message-----
From: André Warnier [mailto:a...@ice-sa.com] 
Sent: ה 30 יוני 2011 18:45
To: Tomcat Users List
Subject: Re: fail to download large static files in tomcat

Michal Singer wrote:
> This was the configuration I had in the server.xml which did not work for 
> large files:
> 
> <Connector executor="AgentExecutor" port="8088" 
> protocol="org.apache.coyote.http11.Http11NioProtocol" SSLEnabled="false" 
> scheme="http" secure="false" redirectPort="8080" connectionTimeout="2000" 
> enableLookups="false" disableUploadTimeout="true" allowTrace="false" 
> maxKeepAliveRequests="1" processorCache="800" acceptorThreadCount="1" />
> 
> And I changed this to:
> <Connector executor="AgentExecutor" port="8088" protocol="HTTP/1.1" 
> SSLEnabled="false" scheme="http" secure="false" redirectPort="8080" 
> connectionTimeout="2000" enableLookups="false" disableUploadTimeout="true" 
> allowTrace="false" maxKeepAliveRequests="1" processorCache="800" 
> acceptorThreadCount="1" />
> 
> And now it does work for 500MB files, though when comparing the time it takes 
> (for smaller files which work on both configurations),
> the performance degraded with the HTTP/1.1 connector. So of course, I prefer 
> it to work on nio.
> 
> 
Just to clarify one more aspect :
In previous messages,

a) you showed an access log line like this :
10.206.90.6 - - [30/Jun/2011:11:37:16 +0300] "GET 
/bundles/vacc-AOS701b318-fl.tgz 
HTTP/1.0" 200 -

(indicating HTTP/1.0 was being used)

and then later you said :

b) "I changed the connector to HTTP/1.1 and now the download works"..

So what exactly did you change, to make HTTP/1.0 become HTTP/1.1 instead ?
Or am I now confused ?
What does the access log show now, for a download ?

  ----------------

On a separate issue : in the AccessLog Valve, you can specify a "pattern", so 
as to add to 
the access log lines the total time it took to process the request (see the 
"pattern" 
attribute and for instance the "%D" pattern string).
That may be useful in order to really compare objectively the times for 
downloading files.

Remember that there are other elements than the server itself that determine 
how fast the 
file is downloaded (the network, how many other things are going on, the 
client, etc..).
So you would need to repeat the test a number of times in order to really be 
sure that 
there is a solid difference on the server side.


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org

Reply via email to