Lab 12: Restart policies in practice

Objective

Demonstrate how each restart policy behaves when the container exits and when the host reboots.

Tasks

Task 1: ‘always’ — restart even after manual stop

docker run -d --name pol-always --restart always alpine sh -c 'sleep 30; exit 1'
docker stop pol-always
# Wait 30s
docker ps -a --filter name=pol-always
# pol-always was restarted even though we stopped it

Task 2: ‘unless-stopped’ — respect manual stop

docker run -d --name pol-unless --restart unless-stopped alpine sh -c 'sleep 30; exit 1'
docker stop pol-unless
# Wait 30s
docker ps -a --filter name=pol-unless
# pol-unless stays in 'Exited' state; not restarted

Task 3: ‘on-failure’ — restart on non-zero exit

docker run -d --name pol-fail --restart on-failure alpine sh -c 'sleep 30; exit 1'
docker ps -a --filter name=pol-fail
# RestartCount keeps incrementing

docker run -d --name pol-ok --restart on-failure alpine sh -c 'sleep 30; exit 0'
# Wait 30s
docker ps -a --filter name=pol-ok
# pol-ok stays in 'Exited' state; exit 0 is success

Task 4: ‘no’ — no restart

docker run -d --name pol-no --restart no alpine sh -c 'sleep 30; exit 1'
# Wait 30s
docker ps -a --filter name=pol-no
# Exited, not restarted

Task 5: Simulate host reboot (skip in shared labs)

In a private lab only:

docker run -d --name reboot-test --restart always nginx:1.27
sudo reboot
# After reboot
docker ps --filter name=reboot-test
# 'always' container is back; 'no' / 'on-failure' may not be

Cleanup

docker rm -f pol-always pol-unless pol-fail pol-ok pol-no 2>/dev/null