Support querying the worker for the currently executing job. #269
Conversation
|
I think this is a reasonable feature request and will be really useful for monitoring purposes. What do you think @nvie? |
| @@ -215,15 +216,41 @@ def register_death(self): | |||
| p.expire(self.key, 60) | |||
| p.execute() | |||
|
|
|||
| def set_state(self, new_state): | |||
| def set_state(self, new_state, pipeline=None): | |||
There was a problem hiding this comment.
I think having the pipeline kwarg in these setters is a Good Thing™, but they aren't used when accessed through the .state property. Actually, I don't think it was good design (I'm fully to blame here!) to have implicit IO when setting a property, because both property access and assignment should have side effects.
I'm fine adding this .set_state(..., pipeline=None) change when we stop using it through the .state property, and I think we should remove it (via proper deprecation, that is, for those who rely on it now). Would you feel comfortable adding that, perhaps?
Same goes for the job_id, which I think is a fine addition. So thanks for that!
Use docstring instead of comment. Sorry.
|
Superseded by #319 |
Adds the possibility for pipelining (similar to Job.save and Job.delete) by adding optional pipeline=None parameters. Upon starting a job, right after setting the 'busy' state, saves the id of the job to be executed. A convenience method for the probably most common use case is also available (
Worker.get_current_job()), which returns aJobinstance.I did not implement pipelining at this time, mainly because I am not 100% of the implications. It would also require reordering of the following code:
Someone who knows the codebase better could easily add it though.