An Exception happens in a check will make the Runner not available for the rest of the analysis about macaron

[tromai]

Looking into the scan method in registry.py. In this block, if an Exception happens, we return the check results without putting the runner which was handling the Error check back into runner_queue. That is why during the next analysis, our runner is not available.
Suggested fix:

if current_future.exception(timeout=self.runner_timeout):
    logger.error("Exception in check %s: %s.", current_check_id, current_future.exception())
    info("Check %s has failed.", current_check_id)
    current_future.cancel()
    # Fix here.
    self.runner_queue.put(current_runner)
    return results

Another solution is to put the exception handling to the Runner itself (so if an exception happens, each runner instance will catch it and put itself back into the queue.

[behnazh-w]

If an exception happens in a check in the analysis of software component, the corresponding analysis should fail. However, the other analysis instances that will run dependencies for instance should keep running. I'm not sure if your suggested fix is helpful in this case.

[tromai]

Yes, it should fail for the current software component we are analyzing. However, because if we don't put the runner back to the runner queue (we only have one runner at the moment), in the next analysis of the next software component, the runner queue is empty and no runner will be able to run the analysis for the rest of the software component.
The reason we need to manually put the runner back is because if an exception happens before this line (which is potentially inside check.run the runner cannot put itself back to the runner queue).

[behnazh-w]

OK, please go ahead with the PR to continue the discussion.