wrapper-user

[Wrapper-user] wrapper.daemonize: signal trap problem

[John Q. <jqu...@jq...>]

Hey Folks:

I'm having some difficulty debugging a problem I'm seeing when daemonizing my application with the wrapper.  I'm hoping that someone more knowledgeable in these things may be able to point me in the right direction, or bring any errors of mine to light.  Thanks in advance for your consideration.  Here are the details.

I've implemented a 'Daemonizer' class that my Application must extend:

 http://pastebin.com/m54ac98cd

Here's a test class:

 http://pastebin.com/m520ea3e4

On a 'CentOS release 5 (Final)' machine, when I run the DaemonizerTest:

 "$WRAPPER_CMD" "$WRAPPER_CONF" wrapper.syslog.ident=$APP_NAME wrapper.pidfile="$PIDFILE"

 (this is actually run using the wrapper init.d script specifying the 'console' run mode)

Everything works correctly, and when I send a SIGINT via the keyboard, or a SIGTERM via kill(1), the Daemonizer.shutdown() hook is properly called.  However, when I run the application as a daemon, such as:

 "$WRAPPER_CMD" "$WRAPPER_CONF" wrapper.syslog.ident=$APP_NAME wrapper.pidfile="$PIDFILE" wrapper.daemonize=TRUE

 (this is actually run using the wrapper init.d script specifying the 'start' run mode)

And subsequently send a SIGTERM via kill(1), the shutdown() method is not called.  I've tried various things to debug this, but am still left scratching my head.

What am I doing wrong?

Thanks so much.

- John Quigley

Re: [Wrapper-user] wrapper.daemonize: signal trap problem

[Leif M. <le...@ta...>]

John,
Can you add the wrapper.debug=true property and then post the
wrapper.log file for the second "start" case?

Cheers,
Leif

John Quigley wrote:
> Hey Folks:
>
> I'm having some difficulty debugging a problem I'm seeing when daemonizing my application with the wrapper.  I'm hoping that someone more knowledgeable in these things may be able to point me in the right direction, or bring any errors of mine to light.  Thanks in advance for your consideration.  Here are the details.
>
> I've implemented a 'Daemonizer' class that my Application must extend:
>
>  http://pastebin.com/m54ac98cd
>
> Here's a test class:
>
>  http://pastebin.com/m520ea3e4
>
> On a 'CentOS release 5 (Final)' machine, when I run the DaemonizerTest:
>
>  "$WRAPPER_CMD" "$WRAPPER_CONF" wrapper.syslog.ident=$APP_NAME wrapper.pidfile="$PIDFILE"
>
>  (this is actually run using the wrapper init.d script specifying the 'console' run mode)
>
> Everything works correctly, and when I send a SIGINT via the keyboard, or a SIGTERM via kill(1), the Daemonizer.shutdown() hook is properly called.  However, when I run the application as a daemon, such as:
>
>  "$WRAPPER_CMD" "$WRAPPER_CONF" wrapper.syslog.ident=$APP_NAME wrapper.pidfile="$PIDFILE" wrapper.daemonize=TRUE
>
>  (this is actually run using the wrapper init.d script specifying the 'start' run mode)
>
> And subsequently send a SIGTERM via kill(1), the shutdown() method is not called.  I've tried various things to debug this, but am still left scratching my head.
>
> What am I doing wrong?
>
> Thanks so much.
>
> - John Quigley
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> Wrapper-user mailing list
> Wra...@li...
> https://lists.sourceforge.net/lists/listinfo/wrapper-user
>
>   

Re: [Wrapper-user] wrapper.daemonize: signal trap problem

[John Q. <jqu...@jq...>]

I believe I've discovered the root of my problem; this has led me to a few other questions that I hope you'll be good enough to answer for me.

The problem appears to have been the result of my sending the TERM signal to the wrong process.  We don't want to be signaling the binary 'wrapper' but rather the Java process itself that the wrapper spawns.  Is this correct?  I expected the wrapper process to be the entity that would catch these signals.

My original assumption that we should be signaling the binary itself came from the behavior of the init.d script provided with the binary distribution.  This script creates a pid file based upon the process number of the wrapper binary, and sends the kill to the process number gathered from this file at time of shutdown.  Is this a bug?

Thanks so much for you help!

- John Quigley