大叔经验分享（88）jenkins假死

jenkins安装启动后，使用systemctl来进行进程监控

# systemctl enable jenkins

但是还是经常发生jenkins进程挂了，不会自动重启，通过systemctl查看状态为：

# systemctl status jenkins

● jenkins.service - LSB: Jenkins Automation Server

   Loaded: loaded (/etc/rc.d/init.d/jenkins; bad; vendor preset: disabled)

   Active: active (exited) since Tue 2019-09-24 18:52:26 CST; 1 months 1 days ago

     Docs: man:systemd-sysv-generator(8)

    Tasks: 0

   Memory: 0B

查看日志没有任何异常，可能的原因如下：

Possibly the Linux kernel's oom-killer ("out-of memory killer") killed your JVM. It sends a SIGKILL signal to selected processes if the physical memory is no longer sufficient to accomodate all processes. Check that output of journalctl for lines containing the string oom-killer to confirm.

It is normal that the oom-killer strikes only after a certain period. Normally, Linux does not impose restrictions on the amount of virtual memory that a process requests (so your JVM can easily allocate 2 GiB of RAM on a 512 MiB system). However, if the process starts actually using the allocated memory (in your case: amount of used heap grows), then the kernel may find that the amount of physical memory is no longer sufficient. This is the point where the oom-killer will select a promising candiate process to be killed. Jenkins, as a non-system process that consumes a lot of memory, has very good chances to be the one that will be killed.

You can solve the situation by

• adding more physical memory to your system
• reducing JVM memory settings (e.g., smaller heap)

修改jenkins占用内存

# vi /etc/sysconfig/jenkins

JENKINS_JAVA_OPTIONS="-Xmx512m"

参考：https://stackoverflow.com/questions/42607771/jenkins-active-exited