Preparing to read tracked jobs from EJR instead of ZK.
635 will currently fetch a lot of jobs from EJR, including those that were never actually started. This mimics the ZK implementation in a way (it will just get those in the /ongoing branch) but that one employs a user_limit to keep requests down; this uses a different strategy and if this thing works in ZK, we can just port it to EJR.
The rationale is that JobTracker requires an application ID in order to do something useful (= use this application ID to fetch the state of the underlying application and update the job status).
Preparing to read tracked jobs from EJR instead of ZK.
635 will currently fetch a lot of jobs from EJR, including those that were never actually started. This mimics the ZK implementation in a way (it will just get those in the
/ongoing
branch) but that one employs auser_limit
to keep requests down; this uses a different strategy and if this thing works in ZK, we can just port it to EJR.The rationale is that
JobTracker
requires an application ID in order to do something useful (= use this application ID to fetch the state of the underlying application and update the job status).