tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject DO NOT REPLY [Bug 50957] Blocking IO can serve wrong response data
Date Tue, 22 Mar 2011 23:17:10 GMT

--- Comment #3 from Brad Plies <> 2011-03-22 19:17:07 EDT ---
(In reply to comment #1)
> Experience has shown that most instances of this type of error are triggered by
> application bugs rather than Tomcat bugs - usually in the form of retaining and
> re-using a reference to the request or response object. One way to test this is
> to set the system property org.apache.catalina.connector.RECYCLE_FACADES to
> true. If you see NPEs then that is indicative of an application bug.

If it were true that this could be caused by application references to request
& response objects, that may not explain why changing to NIO would have any
different behavior.  Also, why would it take an amount of time before
exhibiting the behavior?  If an application did in fact do this, one would
expect a higher occurance rate.

In case I am unable to locate examples of these past instances you describe,
could you provide a few that you know of?  That way I can do a better job
matching characteristics and symptoms.

I will have to evaluate some code to see if any references to request or
response object are being held anywhere.  I would like to try the
RECYCLE_FACADES recommendation but will not be able to put BIO back into the
environment where it was detected.

> It is worth updating to the latest 7.0.x in case you are seeing a variation of
> bug 50189.

50189 differs from this submission in that it relied on AJP and possessed
zero-length messages as symptoms.  By contrast, this submission is using BIO
and not using AJP and positive-length messages are delivered just with the
wrong payloads.  

I cannot yet say at this point if application code is reading from a request
after the response outputstream is closed.  I really doubt it but I'll look

Configure bugmail:
------- You are receiving this mail because: -------
You are the assignee for the bug.

To unsubscribe, e-mail:
For additional commands, e-mail:

View raw message