Steve,

FWIW, this is not just a James issue.  It is a SIGSEGV, and has been
experienced with many applications.  It isn't even just a linux issue; the
same thing is documented on Solaris.  Please note the following from a quick
google search:

ref:

http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q=%22Unexpected+Signal+
%3A+11%22+libjvm
 http://www.google.com/search?hl=en&lr=&ie=UTF-8&oe=UTF-8&q=SEGV+libjvm
 http://www.google.com/search?hl=en&lr=&ie=UTF-8&oe=UTF-8&q=SIGSEGV+libjvm

The only reference I found to any possibility of application error was a
note from Scott Ferguson on Caucho: "There are some cases where a missing
synchronization can cause a SEGV if a thread is reading a object's field
while another is modifying it. These would be very hard to track down."

So far, I have not received a SIGSEGV, and I do use -server.  I am curious
as to the percentage of people who do and do not see the problem.  However,
since it is easy to add options via the environment, it doesn't make sense
to put the -server option in the script.  Anyone who wants it can do "export
PHOENIX_JVM_OPTS=-server" when running James.

With respect to the LD_ASSUME_KERNEL export, it has been recommended.  I
don't know whether or not Sun has addressed the need in later releases of
the JVM.  Sun is aware of the problem, but I think that they should should
be notified of the problem.  The more times the issue comes up, perhaps the
more likely they will be to do something that either resolves it or helps to
identify the cause.

ref:
http://www.google.com/search?hl=en&lr=&ie=UTF-8&oe=UTF-8&q=LD_ASSUME_KERNEL+
java

My suggestion was simply to give it a try and see if it helped; it would at
least give us another data point.

In any event, you can run James without the -server option.  Some people had
the opposite experience as you've had, and found that James was more stable
for long runs with it, so they can just use the export if they want to use
it.

        --- Noel


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to