search

arquillian / arquillian-cube

Control (docker, kubernetes, openshift) containers in your tests with ease!
http://arquillian.org/arquillian-cube/
120 stars 98 forks source link

Failure to cleanup generated docker networks #1097

Closed AndyGee closed 5 years ago

AndyGee commented 5 years ago
Issue Overview

Failure to cleanup docker networks created by the cube framework when the container does not start

Expected Behaviour

Resources that are created automatically (particularly network) should be cleaned up, even on an error.

Current Behaviour

Every failed start still leaves networks in a bad state:

NETWORK ID NAME DRIVER SCOPE dd4e7f1d6c26 bridge bridge local 63bd4b9e59dd e2e_default bridge local 8090aa381f39 e2e_default bridge local 684be06d1553 e2e_default bridge local 6f9ef479dc05 e2e_default bridge local 7dbaf8babc4d host host local 5ee4042f4095 none null local

Steps To Reproduce
  1. Add any erroneous (one that does not start) container to a docker-compose.yml
  2. Run a test
  3. After a failure to start the test container run docker network ls
  4. Manually delete the network (docker network rm [id]), or repeat step 2 to see multiple network added by cube.
Additional Information

There should be shutdown hooks added to the JVM to ensure resources are cleaned up.

AndyGee commented 5 years ago

It is starting to look like networks are not cleaned up in general, which leads to duplicate network errors on a second run.

AndyGee commented 5 years ago

It is lines CubeSuiteLifecycleController#104 and #107 that throws an exception and breaks all cleanups - Thsese lines need to be replaced with log outputs. The @Observes methods must not throw exceptions - This needs to be a golden rule.