I found similar reports (see 0002062: Freenet exits and wrapper fails to start it again - MantisHub) where the JVM could
not be restarted by Java wrapper when it got stuck. The JVM should not
have crashed. This could be caused by some JVM bug. The wrapper however
should have been able to restart Djigzo. It could be that the default
restart delay time is not long enough for all situations.
I will change the restart delay to 30 seconds for the new release. You
can add the following value to:
/usr/share/djigzo/wrapper/djigzo.wrapper.conf
wrapper.restart.delay=30
When the Java wrapper process detects that the JVM hangs it kills the
JVM and waits 30 seconds before restarting Djigzo.
Why the JVM got stuck in your case that remains a mystery. I have never
seen it happen.
Kind regards,
Martijn Brinkers
Andreas Schubert wrote:
···
i use your Djigzo VMware virtual appliance.
in the syslog from yesterday i cannot find any relevant information.
the kern.log has only entries after the today reboot.
Mit freundlichen Grüßen
Andreas Schubert
Dipl.-Ing. (FH)
Leiter EDV
Tel. +49 7121 9463-360
Fax +49 7121 9463-150
Transline Deutschland Dr.-Ing. Sturz GmbH
Transline Deutschland ist ein Unternehmen der Sturz Gruppe (www.sturz-gruppe.de)
Martijn Brinkers
<martijn(a)djigzo.com
> To
Sent by: users(a)lists.djigzo.com
users-bounces(a)lists cc
.djigzo.com
Subject
Re: djigzo does not
06.11.2009 11:13 deliver mails
Somehow the Java Virtual Machine got stuck which normally should not
happen (perhaps a JVM bug?).
Java wrapper which is used to start Djigzo 'injects' a Java process into
the JVM and periodically checks whether the JVM still responds. Java
wrapper itself is a Linux application (written in C). It detected that
the JVM was no longer responding and tried to restart the JVM. Somehow
Java wrapper was unable to restart Djigzo.
JVM appears hung: Timed out waiting for signal from JVM.
JVM did not exit on request, terminated
JVM exited in response to signal SIGKILL (9).
Unable to start a JVM
<-- Wrapper Stopped
--> Wrapper Started as Console
You test that the wrapper process restarts Djigzo after an unexpected
exit by killing the Djigzo java process (using killl -9)
If I explicitly kill the Djigzo Java process Djigzo is automatically
restarted by the wrapper process. djigzo.log after killing the Djigzo
process:
JVM exited unexpectedly.
JVM exited in response to signal SIGKILL (9).
Launching a JVM...
In your case the Java wrapper was unable to restart Djigzo so it seems
that there were problems with your Linux system. The big question is
what kind of problems.
Is there anything relevant in syslog or kernel.log?
Kind regards,
Martijn Brinkers
Andreas Schubert wrote:
hello,
yesterday we had a strange behavior.
djigzo accepts all emails but did not deliver anyone.
in the syslog i find entries like this:
"......lost connection with 127.0.0.1[127.0.0.1] while receiving the initial
server greeting"
and later in the syslog many entries
"connect to 127.0.0.1 .... connection refused"
i tried to open http-management interface, but i got this error message:
(See attached file: check.htm)
has someone an idea what has happend and what can we do to prevent this
behavior?
It should not be difficult to create a plugin for any monitoring system
you use (like Nagios, Cacti, Munin etc.).
Kind regards,
Martijn Brinkers
Martijn Brinkers wrote:
···
I found similar reports (see 0002062: Freenet exits and wrapper fails to start it again - MantisHub) where the JVM could
not be restarted by Java wrapper when it got stuck. The JVM should not
have crashed. This could be caused by some JVM bug. The wrapper however
should have been able to restart Djigzo. It could be that the default
restart delay time is not long enough for all situations.
I will change the restart delay to 30 seconds for the new release. You
can add the following value to:
/usr/share/djigzo/wrapper/djigzo.wrapper.conf
wrapper.restart.delay=30
When the Java wrapper process detects that the JVM hangs it kills the
JVM and waits 30 seconds before restarting Djigzo.
Why the JVM got stuck in your case that remains a mystery. I have never
seen it happen.
Kind regards,
Martijn Brinkers
Andreas Schubert wrote:
i use your Djigzo VMware virtual appliance.
in the syslog from yesterday i cannot find any relevant information.
the kern.log has only entries after the today reboot.
Mit freundlichen Grüßen
Andreas Schubert
Dipl.-Ing. (FH)
Leiter EDV
Tel. +49 7121 9463-360
Fax +49 7121 9463-150
Transline Deutschland Dr.-Ing. Sturz GmbH
Transline Deutschland ist ein Unternehmen der Sturz Gruppe (www.sturz-gruppe.de)
Martijn Brinkers
<martijn(a)djigzo.com
> To
Sent by: users(a)lists.djigzo.com
users-bounces(a)lists cc
.djigzo.com
Subject
Re: djigzo does not
06.11.2009 11:13 deliver mails
Somehow the Java Virtual Machine got stuck which normally should not
happen (perhaps a JVM bug?).
Java wrapper which is used to start Djigzo 'injects' a Java process into
the JVM and periodically checks whether the JVM still responds. Java
wrapper itself is a Linux application (written in C). It detected that
the JVM was no longer responding and tried to restart the JVM. Somehow
Java wrapper was unable to restart Djigzo.
JVM appears hung: Timed out waiting for signal from JVM.
JVM did not exit on request, terminated
JVM exited in response to signal SIGKILL (9).
Unable to start a JVM
<-- Wrapper Stopped
--> Wrapper Started as Console
You test that the wrapper process restarts Djigzo after an unexpected
exit by killing the Djigzo java process (using killl -9)
If I explicitly kill the Djigzo Java process Djigzo is automatically
restarted by the wrapper process. djigzo.log after killing the Djigzo
process:
JVM exited unexpectedly.
JVM exited in response to signal SIGKILL (9).
Launching a JVM...
In your case the Java wrapper was unable to restart Djigzo so it seems
that there were problems with your Linux system. The big question is
what kind of problems.
Is there anything relevant in syslog or kernel.log?
Kind regards,
Martijn Brinkers
Andreas Schubert wrote:
hello,
yesterday we had a strange behavior.
djigzo accepts all emails but did not deliver anyone.
in the syslog i find entries like this:
"......lost connection with 127.0.0.1[127.0.0.1] while receiving the initial
server greeting"
and later in the syslog many entries
"connect to 127.0.0.1 .... connection refused"
i tried to open http-management interface, but i got this error message:
(See attached file: check.htm)
has someone an idea what has happend and what can we do to prevent this
behavior?
Zitat von Martijn Brinkers <martijn(a)djigzo.com>:
Why the JVM got stuck in your case that remains a mystery. I have never
seen it happen.
I have occasionally see stuck process/crash of OS with Linux in VMWare
which cannot be reproduced on real hardware by any means. Maybe it was
such a glitch. One can try to update the VM software if possible and
run a memchecker against the hardware of the Host.
