git clone https://github.com/xmattstrongx/healthcheckrestart.git
The script acts as a CLI which takes in a
-uor--urlflag which is the healthcheck url being exposed by the foobar service.
Example:
./monitor.sh -u http://localhost:8080/healthcheck
-
For this project I am assuming that the foobar service is an executable binary located in the same folder as the
monitor.shfile. Had this been a live fire exercise most like foobar would be setup as a process in the init.d folder. This way the status checks and restarts would have been more like:sudo service start|status|restart -
Also I am logging out straight to the command line. In a real app this would not be as useful as logging to STDERR which would be shipped off to an ELK stack or something.
The monitor script will take action if:
- If the foobar service is not running it will start it.
- If the healthcheck endpoint responds with a 500 it will restart foobar.
- I spent approxiamately 1 hour on this problem.
I wrote a tiny go app to simulate the randomness of either a 200 or a 500 responce. If you would like to use it to test with simply run go build -o foobar
Note: The monitor script does not gracefully shut down as is so it may leave foobar processes stranded. killall foobar command will take care of that.