Raising the errors would crash the current thread. As this thread
contains the Eventmachine, that would influence other connections
as well. Attaching the errors to the connection and reading them
after the connection was closed ensures that the thread stays
alive while handling the errors in the main thread of the current
request.
These tests were blocking because of the newly introduced
EventLoop. The messages sent to the EventLoop are now mocked
and the EventLoop isn't blocking anymore in the tests.
In order to provide an alternative to Poseidon, a strategy for the
DockerContainerPool is added that is used by the runner model.
Co-authored-by: Sebastian Serth <Sebastian.Serth@hpi.de>
Timeouts are now handled correctly and the Runner automatically
creates the execution environment if it could not be found in
Poseidon. The runner is deleted locally if Poseidon returns
a bad request error.
After removing the logic that stores the duration that has been
waited for a runner in the runner, this now also removes the column
from the database as it is not used anymore.
The runner model is only a class responsible for storing information
now. Based on the configuration it picks a strategy for the runner
management. The Poseidon strategy is already implemented and tested.
The Docker strategy will follow.
This adds policy tests to ensure only an admin can synchronize all
execution environments. It also adds controller tests that check that
all execution environments get synchronized.
This adds a button to the execution environment index page that, when
clicked, causes all execution environments to be synchronized to the
runner management (Poseidon) by creating or replacing them.
CodeOcean does not synchronize it's execution environments on startup or
when a new runner management configuration is used for the first time.
The administrator has to manually start this process by pressing this
button.
The equivalent for syncing just one execution environment is updating
it.
The errors are raised in the runner model and in the runner connection
class. In the submission controller the errors are rescued and,
depending on the error, the status timeout / container depleted is
sent to the client.
17 tests are always failing, due to changes introduced when adding the
Runner abstraction. To know only these fail, they now get skipped in
order to make it apparent if tests that should not fail do fail in the
pipeline.
