Deliver unfinished HTML pages

Hi Bjoern,

 

 

A GTM WIP only provides answers to DNS queries. An LTM VIP can send an HTTP response to a client using an iRule.

 

 

What exactly are you trying to do?

 

 

Aaron

a user requests a html side over the f5 to the web server. the web server begin to generate the html page and begin the delivering the page (over the F5). The problem is, that the f5 only begin with the delivering when the web server has the complete page has generated. the page the the web server deliver is a kind of waiting room

I assume you have a VIP defined within the local traffic section of the configuration. Can you post an anonymized version of the output from 'b virtual VIP_NAME list'? Do you have an HTTP profile added to the VIP? If so, can you also post the output from 'b profile HTTP_PROFILE_NAME list'?

 

 

Is the server using chunked encoding in responses? If so, SOL5379 (Click here) might be relevant to this issue.

 

 

Aaron

here is the data:

 

----------------snip-------------

 

[root@lb1:Active] config b virtual vs_23_test_http list

 

virtual vs_23_test_http {

 

destination 1.1.1.1:http

 

ip protocol tcp

 

profile http tcp

 

pool pool_test_23_http

 

}

 

----------------snip------------------

 

 

[root@lb1:Active] config b profile http list

 

profile http http {

 

basic auth realm none

 

oneconnect transformations enable

 

compress disable

 

compress uri include none

 

compress uri exclude none

 

compress content type include "text/" "application/(xml|x-javascript)"

 

compress content type exclude none

 

compress min size 1024

 

compress buffer size 4096

 

compress vary header enable

 

compress http 1.0 disable

 

compress gzip memory level 8k

 

compress gzip window size 16k

 

compress gzip level 1

 

compress keep accept encoding disable

 

compress browser workarounds disable

 

compress cpu saver enable

 

lws width 80

 

max header size 32768

 

pipelining enable

 

adaptive parsing disable

 

ramcache disable

 

ramcache size 100mb

 

ramcache max entries 10000

 

ramcache max age 3600

 

ramcache min object size 500

 

ramcache max object size 50000

 

ramcache uri exclude none

 

ramcache uri include none

 

ramcache uri pinned none

 

ramcache ignore client cache control all

 

ramcache aging rate 9

 

ramcache insert age header enable

 

}

 

profile http http_fansale {

 

defaults from http

 

oneconnect transformations disable

 

}

 

So it looks like a standard HTTP VIP configuration. What are the symptoms of the problem? How did you determine that the BIG-IP is buffering the full response before sending it?

 

 

Is the server sending chunked responses? Is chunking set to preserve on the HTTP profile? I believe that's the default setting.

 

 

Aaron

from client side, the first seconds happened nothing, then he will get the whole page in a very short time. The chunked setting in the http profile is a idea - will check it

 

 

It would help to capture a tcpdump of the client connecting directly to the server and compare that with a tcpdump on the BIG-IP through the VIP. For the BIG-IP, you can use syntax like this to capture traffic on all interfaces involving the client or server IP's:

 

 

tcpdump -i 0.0 -s0 -w/var/tmp/trace.dmp host CLIENT_IP or host SERVER_IP

 

 

Replace CLIENT_IP and SERVER_IP with the actual addresses. The output is written to /var/tmp/trace.dmp or whatever you name the output file. You can use Wireshark to view the trace. If you get stuck, F5 Support can help you review the tcpdumps.

 

 

Aaron

Do you have iRules on this VS? What happens if you disable HTTP profile processing (Either remove it, or put in an iRule to run last that disables it at the end of the request, before the response). (Assuming you don't have any iRules that use the HTTP_RESPONSE that is).

 

 

H

sorry, i´m today total bussy to follow your hints - i´ll be back :-)

 

(on monday)