tomcat-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject [Bug 58241] JVM crash when trying to flush output buffer
Date Wed, 12 Aug 2015 17:48:53 GMT

Mark Thomas <> changed:

           What    |Removed                     |Added
             Status|NEW                         |RESOLVED
         Resolution|---                         |INVALID

--- Comment #5 from Mark Thomas <> ---
The Tomcat project does not provide patches for issues in old releases, we only
provide new releases so if a bug isn't reproducible on the latest stable
release (i.e. 7.0.63 as I write this) then it isn't going to get looked at.

If a bug includes a test case then we can run it on the latest release and test
it ourselves. For a bug like this where no test case is provided, the onus to
test on the latest release is with the user reporting the bug.

Note: In this particular case, you would need to test with the latest tc-native
release (1.1.33 as I write this).

Having looked at this more closely this isn't a Tomcat bug. An instance of is using the OutputStream obtained
from ServletResponse.getOutputStream() (and in one case the stack trace
suggests it has somehow obtained direct access to the underlying OutputBuffer).
If this OutputStream is used outside of the lifecycle of the associated
Response things will go wrong. In the case of APR crashes are likely since
there is an attempt to write to a socket that no longer exists. You'd see
errors but not crashes if you switched to the NIO connector.

Note there doe snot appear to be any need for you to be using APR/native since
you are not using TLS. NIO should give comparable performance.

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