Re: [Wrapper-user] Stop wrapper from sending "kill -9"?

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

Graham,
What version of the Wrapper are you using? And what platform is it running on?
Could you please send me your wrapper.conf file so I can have a look at it?

The Wrapper is first of all timing out because the JVM is frozen for
longer than the ping timeout. (wrapper.ping.timeout)
That is what is initiating the shutdown at 07:53:51.
The Wrapper should then start to shutdown the JVM and then kill it
after it fails to exit on its own.  That is all happening immediately
in your case.

To avoid the ping timeouts, you should set it to a very large value or
0 to disable it all together.
wrapper.ping.timeout=true

The current version looks like this:
---
INFO   | jvm 1    | 2013/10/08 14:05:53 | WrapperManager: WARNING:
Making JVM appear to be hung...
STATUS | wrapper  | 2013/10/08 14:06:27 | JVM appears hung: Timed out
waiting for signal from JVM.  Restarting JVM.
ERROR  | wrapper  | 2013/10/08 14:07:02 | Shutdown failed: Timed out
waiting for signal from JVM.
ERROR  | wrapper  | 2013/10/08 14:07:02 | JVM did not exit on request,
termination requested.
STATUS | wrapper  | 2013/10/08 14:07:02 | JVM received a signal SIGKILL (9).
STATUS | wrapper  | 2013/10/08 14:07:02 | JVM process is gone.
STATUS | wrapper  | 2013/10/08 14:07:02 | JVM exited after being
requested to terminate.
---
See that the original hang was detected at 14:06:27.  The JVM is asked
to shutdown on its own. But that fails after 30 seconds and the
Wrapper decides to kill it forcibly.
This second timeout is the wrapper.shutdown.timeout and
wrapper.jvm_exit.timeout, depending on which part of the shutdown is
getting stuck.
In your case, it is best to simply disable the ping timeouts all together.

In recent versions, you can request that slow pings be logged for your
reference as follows:
wrapper.ping.alert.threshold=1
Also see the jvm_ping_slow and jvm_ping_response_slow events to be
able to respond to these slow pings in your own way:
http://wrapper.tanukisoftware.com/doc/english/props-event.html#events

Sincerely,
Leif Mortenson
Tanuki Software, Ltd.

On Tue, Oct 8, 2013 at 9:08 PM, Graham Leggett <mi...@sh...> wrote:
> Hi all,
>
> I am currently trying to get the service wrapper to not send the -9 signal at all to (in my case) activemq, as activemq reverts to an older version of the queue and the additional messages being repreocessed is killing our system.
>
> INFO   | jvm 1    | 2013/10/08 07:53:19 | [Full GC
> INFO   | jvm 1    | 2013/10/08 07:53:22 |  2048661K->2038472K(2092928K), 3.5312760 secs]
> ERROR  | wrapper  | 2013/10/08 07:53:51 | JVM appears hung: Timed out waiting for signal from JVM.
> ERROR  | wrapper  | 2013/10/08 07:53:51 | JVM did not exit on request, terminated
> INFO   | wrapper  | 2013/10/08 07:53:51 | JVM exited on its own while waiting to kill the application.
> STATUS | wrapper  | 2013/10/08 07:53:51 | JVM exited in response to signal SIGKILL (9).
>
> I don't mind the service wrapper complaining that the JVM is too slow, what I want is the wrapper to just ask for a graceful restart or no restart at all.
>
> Any ideas?
>
> Regards,
> Graham
> --
>
>
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
> _______________________________________________
> Wrapper-user mailing list
> Wra...@li...
> https://lists.sourceforge.net/lists/listinfo/wrapper-user



-- 
Leif Mortenson
President & CEO
Tanuki Software, Ltd.
6-16-7-1001 Nishi-Kasai, Edogawa-ku
Tokyo 134-0088 Japan
Tel: +81-3-3878-3211
Fax: +81-3-3878-0313
モートンソン リーフ
代表取締役
タヌキソフトウェア有限会社
〒134-0088 東京都江戸川区西葛西6-16-7-1001 第2 白子ビル
Tel: 03-3878-3211
Fax: 03-3878-0313
http://www.tanukisoftware.com
lei...@ta...