I submitted a bug on this new behavior when running Resin 4.x on Windows 

The problem is that when you issue a "stop" command to the resin service either 
via the sc (service controller) command, e.g. "sc \\serverName stop resin" or 
via "net stop resin", you can't then issue a "start" command to start up resin 
because Windows still thinks the service is running. We use this process for 
doing deployments or just general restarts from a central server.

I haven't tested this scenario, but another problem with this new behavior 
might be that if the watchdog doesn't know that a resin stop was specifically 
requested via the service controller, it might try to restart resin and that 
would be bad.


On Mar 30, 2010, at 11:49 , Emil Ong wrote:

> Hi Aaron,
> This is the correct behavior. In 4.0.x, we changed the watchdog to be a
> long-lived process.  If you'd like to shut it down, along with all
> running resin instances, use the "shutdown" command.  We're in the
> process of updating the documentation this week to reflect the change.
> Thanks,
> Emil
> On Tue, Mar 30, 2010 at 11:01:23AM -0500, Aaron Freeman wrote:
>> When we try to stop the resin-4.0.5 processes using:
>> $JAVA_HOME/bin/java \
>> -server \
>> -Djava.util.logging.manager=com.caucho.log.LogManagerImpl \
>> -Djava.security.egd=/dev/urandom \
>> -Dhost=${SERVER} \
>> -Dresin.home=${RESIN_HOME} \
>> -jar ${RESIN_HOME}/lib/resin.jar \
>> -conf ${SERVER_ROOT}/conf/resin.xml \
>> $*
>> Where we pass in "start" to start it and "stop" to stop the server.  The 
>> main resin java process stops, but the watchdog does not.  Is that 
>> expected behavior (and thus a change from 3.0.x)?  If so, how do we stop 
>> the watchdog java process?  I believe Rick Mann reported the same 
>> issue.  Also we are running uncompiled, if that matters.

resin-interest mailing list

Reply via email to