Feature #9803
closed
Storing multiple job records to support job reuse
Description
Correct me if I'm wrong, but I think this is how job reuse works (only in experience, I don't have the proof to back it up)
Pipeline = Job A (Complete) > Job B (Complete) > Job C (Complete)
User A changes something in Job C...
Pipeline = Job A (re-use-complete) > Job B (re-use-complete) > Job C (Running to Complete)
User A runs --no-reuse, because they want to see an end-to-end run
Pipeline = Job A (Running) > Job B (Waiting) > Job C (Waiting)
User B realizes that Job C needs to another change
Pipeline now starts...
Pipeline = Job A (Running) > Job B (Waiting) > Job C (Waiting)
User B would like to test Job C alone without waiting for Job A and Job B to finish again
My idea is to store an additional job record for jobs that are currently running so re-use would work while the job is running with --no-reuse, so that User B can test Job C with Job A and B reused.
Maybe that record is still stored, and that it chooses to use the running job instead of re-using, if that is the case, re-using would be better if --no-reuse isn't flagged.
Updated by Bryan Cosca
Updated by Tom Morris over 9 years ago
- Subject changed from Storing multiple job records to Storing multiple job records to support job reuse
Updated by Tom Morris about 7 years ago
- Status changed from New to Closed
This is about the Jobs API which is obsolete, but I think this use case is covered by the current Containers API (ie Crunch 2) implementation, if I'm reading it correctly. ie Container Requests can be satisfied by running Containers without creating new ones. cf #9623