service tomcat7 start fails, but the process exist

2020-07-02 22:06发布

问题:

I have been trying to install tomcat7 on ubuntu docker images with apt-get install tomcat7. The installation works fine and starting tomcat from the catalina.sh works as well.

I need to start tomcat from "service tomcat7 start", which instead fails. Regardless the failure result, if I wget localhost:8080 I can see tomcat answering and if I ps -ef | grep tomcat I can see the process.

Similarly if I run service tomcat7 status it says the tomcat is not running even when it is and the PID file does exist.

I have noticed that when I start tomcat from the catalina scripts, the pid file created is called tomcat.pid, but the /etc/init.d/tomcat script would try to read tomcat7.pid.

However, forcing the name in the script to peek up the right pid file, does not solve the problem.

Has anyone else experienced this?

The ubuntu version of the docker file I am trying is not really relevant, since I have been trying with several.

Anyway the one I am mostly using are 12.10 and 14.04.

Thanks!

回答1:

In the docker Ubuntu image I am using (5506de2b643b - 14.04.1 LTS), the start-stop-daemon with the --test argument is working incorrectly and reports that tomcat is not running, even when it is.

The tomcat7 init.d script starts tomcat because start-stop-daemon --test says (correctly) tomcat is not running, but then a little later in the startup process it checks that tomcat started successfully and is running. start-stop-daemon --test now incorrectly says tomcat is not running, which causes the tomcat7 init.d script to remove the PID file.

As a result, service tomcat7 status returns false when tomcat is running because the PID file is gone, but it will return false even if the PID file is there with correct PID due to the bug in start-stop-daemon --test.

Here's an example session demonstrating the bug:

#TOMCAT PID IS 43
root@a2cf26ade2a9:/# ps -eaf | grep tomcat7
tomcat7     43     1  0 14:06 ?        00:00:04 /usr/lib/jvm/java-7-oracle/bin/java -Djava.util.logging.config.file=/var/lib/tomcat7/conf/logging.properties -Djava.util.logging.manager=org.apache.juli.ClassLoaderLogManager -Djava.awt.headless=true -Xmx128m -XX:+UseConcMarkSweepGC -Djava.endorsed.dirs=/usr/share/tomcat7/endorsed -classpath /usr/share/tomcat7/bin/bootstrap.jar:/usr/share/tomcat7/bin/tomcat-juli.jar -Dcatalina.base=/var/lib/tomcat7 -Dcatalina.home=/usr/share/tomcat7 -Djava.io.tmpdir=/tmp/tomcat7-tomcat7-tmp org.apache.catalina.startup.Bootstrap start

#PID FILE HAS CORRECT PID
root@a2cf26ade2a9:/#  cat /var/run/tomcat7.pid
43

#START-STOP-DAEMON --TEST REPORTS THAT IT WOULD START TOMCAT
root@a2cf26ade2a9:/#  start-stop-daemon --test --start --pidfile /var/run/tomcat7.pid --user tomcat7 --exec /usr/lib/jvm/java-7-oracle/bin/java
Would start /usr/lib/jvm/java-7-oracle/bin/java .

root@a2cf26ade2a9:/# echo $?
0


回答2:

Docker does not start up any of the OS services, only the command given in the Dockerfile or explicitly on the command line.

My understanding is that service and related commands depend on an init process which does not exist.

You should start you tomcat explicitly from the Catalina scripts. You can wrap that with something like supervisord to get the "restart-if-it-crashes" behavior.



回答3:

I encountered this. I fixed it (finally) by adding JAVA_HOME=/usr/lib/jvm/java-8-oracle to /etc/default/tomcat7. For some reason it was using that JAVA_HOME to actually run the process, but it was using /usr/lib/jvm/default-java to check if the process was running. This confused it, so it never detected that the process was running.



回答4:

I happens have the some situation, it caused by deleting /var/lib/tomcat7/logs directory, where I remake the directory, it owned by another user, so that tomcat7 can not restart, and not logs output, I change own, now it works fine.

sudo chown tomcat7.tomcat7 /var/log/tomcat7
sudo chown tomcat7.tomcat7 /var/lib/tomcat7/logs