What
It is possible to define a run target under run_targets called fallback, which interferes with the run target generated by the fallback_run_target option.
|
// the fallback state doesn't have a name in the config, so we use |
|
// "fallback", it doesn't actually matter... |
It is not clear what should happen in this case.
This mistake actually appears in our own integration tests:
Acceptance Criteria (DoD)
- Behaviour is clearly defined
How
A few different ideas:
- Delete the
fallback_run_target option, tell users to configure a run target called fallback directly.
- Make
fallback_run_target take only a string, which names an existing run target to be used.
- Generate an error if a run target called
fallback appears under run_targets.
- Replace the hardcoded name with a dynamically chosen one, to avoid the conflict.
What
It is possible to define a run target under
run_targetscalledfallback, which interferes with the run target generated by thefallback_run_targetoption.lifecycle/score/launch_manager/src/daemon/src/process_group_manager/process_group_manager.cpp
Lines 344 to 345 in 2822719
It is not clear what should happen in this case.
This mistake actually appears in our own integration tests:
lifecycle/tests/integration/process_wrong_binary_failure/process_wrong_binary_failure.json
Line 116 in 2822719
Acceptance Criteria (DoD)
How
A few different ideas:
fallback_run_targetoption, tell users to configure a run target calledfallbackdirectly.fallback_run_targettake only a string, which names an existing run target to be used.fallbackappears underrun_targets.