Some tests modules require pre-configured storage in a well-known location. If the storage is not configured, these tests can skip, xfail or fail (depending on the test).
To setup storage for these tests, run:
make storage
This can be run once when setting up development environment, and must be run again after rebooting the host.
If you want to tear down the user storage, run:
make clean-storage
There is no need to tear down the storage normally. The loop devices are backed up by sparse files and do not consume much resources.
The storage is configured in the file storage.py. Tests that use user storage, load this configuration file and access the storage via the BACKENDS dict. These are the available storage configurations:
-
file-512 - a file on a file system backed by loop device with 512 bytes sector size.
-
file-4k - a file on a file system backed by loop device with 4k sector size. This configuration is not available on CentOS 7 and is known to be fail randomly on oVirt CI.
-
mount-512 - mounted file system backed by loop device with 512 bytes sector size.
-
mount-4k - mounted file system backed by loop device with 4k sector size. This configuration is not available on CentOS 7 and is known to be fail randomly on oVirt CI.
Storage configurations which are not supported on the current environment are skipped automatically. The tests must deal with the missing storage configuration by detecting that the storage does not exist.
To add new storage configurations edit the BACKENDS list in storage.py.
For more info on userstorage, including details of how it works, see https://github.com/nirs/userstorage.