https://bz.apache.org/bugzilla/show_bug.cgi?id=57749
Mark Thomas changed:
What|Removed |Added
Resolution|--- |FIXED
Status|NEW
https://bz.apache.org/bugzilla/show_bug.cgi?id=57749
--- Comment #7 from Mark Thomas ---
I opted for WARN until we were more sure what the root cause was. If it is
concurrent close then no objections to making it debug.
--
You are receiving this mail because:
You are the assignee for the bug.
https://bz.apache.org/bugzilla/show_bug.cgi?id=57749
--- Comment #6 from Remy Maucherat ---
IMO, the logging level should be debug, there are other places where the socket
is checked for null due to possible concurrent close and there's no logging at
all.
--
You are receiving this mail because:
https://bz.apache.org/bugzilla/show_bug.cgi?id=57749
--- Comment #5 from Konstantin Kolinko ---
Created attachment 32608
--> https://bz.apache.org/bugzilla/attachment.cgi?id=32608&action=edit
access_log.2015-03-25
Access log file for the failed test.
As I mentioned earlier, I am using
test.ac
https://bz.apache.org/bugzilla/show_bug.cgi?id=57749
--- Comment #4 from Konstantin Kolinko ---
Created attachment 32607
--> https://bz.apache.org/bugzilla/attachment.cgi?id=32607&action=edit
TEST-org.apache.catalina.nonblocking.TestNonBlockingAPI.NIO2.txt
Reproduced.
This is Tomcat 8 @r16689
https://bz.apache.org/bugzilla/show_bug.cgi?id=57749
--- Comment #3 from Mark Thomas ---
I've added some logging and I'll set the TestNonBlockingAPI test running in a
loop for NIO2 until it fails.
--
You are receiving this mail because:
You are the assignee for the bug.
---
https://bz.apache.org/bugzilla/show_bug.cgi?id=57749
--- Comment #2 from Mark Thomas ---
That was going to be my first question - how easy is this to reproduce.
I took a quick look at the code. The only way the socket could be null is if
the output buffer had been recycled. That would be trigger
https://bz.apache.org/bugzilla/show_bug.cgi?id=57749
Konstantin Kolinko changed:
What|Removed |Added
OS||All
--- Comment #1 from Konstanti