https://bz.apache.org/bugzilla/show_bug.cgi?id=63306
Mark Thomas <ma...@apache.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Resolution|--- |WORKSFORME
Status|NEW |RESOLVED
--- Comment #4 from Mark Thomas <ma...@apache.org> ---
I can't re-create the scenario you describe. I have been using the Snake game
from the WebSocket examples and disabling the WiFi on the client device. What I
observe is:
- Tomcat continues to be able to write to the client without any apparent error
- There is (eventually) an I/O exception on read but it takes about 15 minutes
to occur
If I give it long enough, everything does get tidied up. I have been unable to
trigger a permanent memory leak.
I strongly suspect that the time taken for the connection to fail will be
highly dependent on the network.
It is not clear whether you are seeing the same behaviour as I am or if you are
seeing a genuine memory leak.
I have also done some testing with the TesterFirehoseServer class. Using the
async API appears to trigger this issue more readily than the basic API. This
time the I/O exception was on write but it was still after about 15 minutes.
If you believe you have found a genuine memory leak then feel free to re-open
this issue but you will need to provide the steps to reproduce it (possibly
with a simple test web application) on a clean install of the latest 7.0.x
release.
Generally, I'd expect the WebSocket implementation in 8.5.x onwards to perform
better. It uses the I/O layer directly rather than using the Servlet
non-blocking I/O API.
--
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@tomcat.apache.org
For additional commands, e-mail: dev-h...@tomcat.apache.org
