0004969: ServletContextListener.contextDestroyed() not called when resin is told to shutdown

Viewing Issue Simple Details [ Jump to Notes ]

[ View Advanced ] [ Issue History ] [ Print ]

Category

Severity

Reproducibility

Date Submitted

Last Update

0004969

[Resin]

major

always

02-28-12 07:32

06-19-12 13:03

Reporter

aiman_a

View Status

public

Assigned To

ferg

Priority

normal

Resolution

fixed

Status

closed

Product Version

4.0.24

Summary

0004969: ServletContextListener.contextDestroyed() not called when resin is told to shutdown

Description

I notice there is a bug (4331, and 4325 related) already logged and closed for this, but it was flagged as 'can not reproduce'. I can reproduce this reliably in my environment.

I have a ServletContextListener implementation that prints to StdOut, StdErr and throws an Exception in the contextDestroyed() method. I have added it as a listener in my web.xml, and the resin logs show that the context initialized message is being printed when I start the server.

The contextDestroyed() method does not get executed when I exit the server using "resin.sh shutdown".
It does however get executed when I call "resin.sh stop" and also when I dump a new war file into the web-apps directory forcing a restart.

Another interesting thing I can see is that killing the child server process (the one that the watchdog starts) manually using 'kill' calls the contextDestroyed() method. However, killing the parent watchdog process does NOT call it.

I do not have "shutdown-wait-max" defined anywhere in my resin.xml, so I assume this allows me to have 60s to perform shutdown code. I should be able to write to StdOut quite easily in that time.

I have also tested this using JVM shutdown hooks (Runtime.getRuntime().addShutdownHook) instead of contextDestroyed, and I get the exact same results.

This does not matter if you use resin.sh or pass "shutdown" as an argument directly to resin.jar

Additional Information

This should behave as I have described as only "kill" should close uncleanly. This can be seen in this blog post: http://blog.caucho.com/2011/08/29/stopping-resin-stop-shutdown-or-kill/ [^]

A workaround is to simply call stop on all your servers instead of shutdown.

Attached Files

Relationships

Notes
(0005883) ferg 06-19-12 13:03	server/6h10

Issue History
Date Modified	Username	Field	Change
02-28-12 07:32	aiman_a	New Issue
06-19-12 13:03	ferg	Note Added: 0005883
06-19-12 13:03	ferg	Assigned To	=> ferg
06-19-12 13:03	ferg	Status	new => closed
06-19-12 13:03	ferg	Resolution	open => fixed
06-19-12 13:03	ferg	Fixed in Version	=> 4.0.29

Mantis 1.0.0rc3[^]

29 total queries executed.
26 unique queries executed.