Test debugging

Jan Safranek jsafrane at redhat.com
Fri Oct 11 07:40:54 UTC 2013


Hey,

Our automated tests in our BuildBot[1] were extended lately:

1: http://openlmi-rnovacek.rhcloud.com

* All builds and tests have available repository with latest nightly
builds of our providers. That means e.g. openlmi-storage tests get
automatically the nightly openlmi-python-providers from git. Previously,
openlmi-storage tests were running with openlmi-python-providers from
distro.

* All tests are running on clean virtual machine. After tests, the
  machine is reverted to stable snapshot (loosing all log messages etc,
  see below to prevent this).

* When starting new build manually, you can fill the 'Force build' form
  in web GUI and set Name='sleep' with Value='1d'.
  * BuildBot will then build packages from git, install them and run
    all tests as usual, but before reverting to snapshot it will sleep
    for given time, one day in this case.
  * You can ssh to the machine, look at state of the system after
    tests, manually restart any tests and use any means to debug your
    providers.
  * Just kill the running sleep afterwards and the machine will be
    reverted to the stable state so other tests can run. All your
    changes will be lost!
  * Ask me or Radek Novacek to update the master snapshot if some
    adjustment is needed.
  * Of course, you need BuildBot account to do all this :)

Big thanks to Radek who made all this possible.

Jan


More information about the openlmi-devel mailing list