[AMQ-7336] Service stops with no apparent reason

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 5.15.7, 5.15.10
• Component/s: Broker

Environment

OS : Unix Solaris (I made few minor adaptations on the started script to make it compatible)

OS vers : 11.3.36.7.0

ActiveMQ vers : 5.15.7 at first but I upgraded to the last one today (5.15.10), same result.

I configured log4j with some minor changes :

log4j.rootLogger=DEBUG, console, logfile

log4j.appender.logfile.file=/logs/activemq/activemq.log

log4j.appender.audit.file=/logs/activemq/audit.log

 

I configured /etc/default/activemq with :

ACTIVEMQ_HOME="/produits/activemq"             # I'm used to install new soft in /produits
ACTIVEMQ_BASE="$ACTIVEMQ_HOME"
ACTIVEMQ_CONF="$ACTIVEMQ_BASE/conf"
ACTIVEMQ_DATA="/activemq/data"                     # I dedicated a disk partition to the growing directories
ACTIVEMQ_TMP="/activemq/tmp"

ACTIVEMQ_OPTS_MEMORY="-Xms64M -Xmx1G" # kept untouched

ACTIVEMQ_OPTS="$ACTIVEMQ_OPTS_MEMORY -Djava.util.logging.config.file=logging.properties -Djava.security.auth.login.c
onfig=$ACTIVEMQ_CONF/login.config" # kept untouched

ACTIVEMQ_OUT="/logs/activemq/console.log" # I'm used to write logs on a dedicated disk partition, see the lo4j config

ACTIVEMQ_OPTS="$ACTIVEMQ_OPTS -Dorg.apache.activemq.audit=true"

ACTIVEMQ_SUNJMX_START="$ACTIVEMQ_SUNJMX_START -Dcom.sun.management.jmxremote" # kept untouched

ACTIVEMQ_QUEUEMANAGERURL="--amqurl tcp://0.0.0.0:61616"

ACTIVEMQ_SSL_OPTS="" # kept untouched

ACTIVEMQ_KILL_MAXSECONDS=15

ACTIVEMQ_USER="activemq"

ACTIVEMQ_PIDFILE="$ACTIVEMQ_DATA/activemq.pid"

JAVACMD="auto"

• Assignee: Jean-Baptiste Onofré
• Reporter: Thomas Ricou
• Created: 05/Nov/19 17:12
• Updated: 06/Nov/19 11:44

Description

Hi,

First of all, thanks for having developed this great tool, it helps me a lot.

ActiveMQ is having a strange behavior concerning this issue. I configured a service to run <install_dir>/bin/activemq start (or stop). I'm using the solaris SMF system which launch the start command and monitors the child PID. If it ends, the SMF considers a failure and start back the service. The main point of the issue is that the ActiveMQ process stops but nothing seems to be related to this stop in the logs. I activated DEBUG level.

While the SMF system saves me, it appears that activeMQ clients are not behaving very well when it happens. This is my business to make them able to "survive" the ActiveMQ restart. What I'm considering a an important bug that I can't figure what is stoping the ActiveMQ service. The only audit log I could find says :

INFO | anonymous called org.apache.activemq.broker.jmx.BrokerView.terminateJVM[0] at 05-11-2019 14:20:17,921 | RMI TCP Connection(1)-192.168.3.72
INFO | anonymous called org.apache.activemq.broker.jmx.BrokerView.terminateJVM[0] on localhost at 05-11-2019 16:35:51,658 | RMI TCP Connection(1)-192.168.3.72

(the final IP is the local server's running the ActiveMQ service)

 

My last clue is that it doesn't happen on the test environment, only on production where activity is far heavier.

 

Would you have any idea to find out what is causing the service to stop ?

 

Thanks a lot.