AJP message chunking behaviour

[Tim Whittington <Ti...@orionhealth.com>]

I'm observing sub-optimal AJP messages being sent from Tomcat (5.0.something in this case) to JK running under IIS.
 
I see a consistent pattern (for large response) of message sizes of 8188 and then 12.
This looks like 8184 + AJP header and 8 + AJP header = 8192 + 2 * AJP Headers= 8 * 1024 bytes + 2 * AJP headers.
 
It appears that the AJP Connector on the Tomcat side is calculating message sizes without taking into account the AJP headers, and thus spreading a 8192 byte response message across two AJP messages.
 
Has anyone observed anything like this?
Is this a bug, or expected behaviour?
 
tim

Re: AJP message chunking behaviour

[Bill Barker <wb...@wilshire.com>]

And this, children, is why you don't want to remain on out-dated and 
unmaintained systems :).

If you search the dev@tomcat archives, I believe that Mladen reported this a 
couple of years ago (but don't care enough to actually do the search myself 
;-).  In 5.5, it does much more optimal chunking.

"Tim Whittington" <Ti...@orionhealth.com> wrote in message 
news:45670786.4A17.00AA.0@orionhealth.com...
I'm observing sub-optimal AJP messages being sent from Tomcat (5.0.something 
in this case) to JK running under IIS.

I see a consistent pattern (for large response) of message sizes of 8188 and 
then 12.
This looks like 8184 + AJP header and 8 + AJP header = 8192 + 2 * AJP 
Headers= 8 * 1024 bytes + 2 * AJP headers.

It appears that the AJP Connector on the Tomcat side is calculating message 
sizes without taking into account the AJP headers, and thus spreading a 8192 
byte response message across two AJP messages.

Has anyone observed anything like this?
Is this a bug, or expected behaviour?

tim




---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org