vp340 commented on PR #3554:
URL: https://github.com/apache/cxf/pull/3554#issuecomment-6068692617
Hi @reta ,
It's ok for me.
Obviously not ` leaking / keeping temp files` was the absolute priority!
I understand that we found these types of problems only with
LoggingOutputStream.class so spreading the change could be more dangerous than
beneficial.
> the user will get real exception (if happen during write) or closed
exception (if the stream was closed and write happens).
Yeah it's not a problem itself.
Maybe we should at least document what is the case if someone find this
error "The channel has been closed already" coming from the logging module.
From experience in my company (unfortunately) if there were a pick of
"connection reset" Exception, seeing also the same number of the new channel
exception thrown... the first instinct of the average IT guy could be to
disable log... if they see
> Caused by: java.io.IOException: The channel has been closed already
at
org.apache.cxf.ext.logging.LoggingOutputStream.write(LoggingOutputStream.java:107)
... We shouldn't code around user assumptions (we are not the frontend guys
;D ) , but I recommend to provide at least a clear explanation in the
documentation, that hopefully will prevent people from thinking the logging
feature is causing the failure, when it's actually just reacting to an already
closed connection."
If U are interested @ffang , @reta in what the new stack actually can look
like with a spring-camel-tomcat integration, here I upload the log of a
complete reproduced issue (REQ_IN, REQ_OUT, RESP_IN, RESP_OUT + Connection
Reset by peer | stack + The channel has been closed already | stack ) :
[connection-RST-new-stack.txt](https://github.com/user-attachments/files/33223953/connection-RST-new-stack.txt)
Thanks again to both of you for the precious lessons.
I'm glad I helped even a little bit.
See U in the next problem!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
