If a batch fails, the Miseq monitor resubmits all the samples. Kive then sifts through them to see which ones can reuse old results and which ones need to run. That can make Kive slow as the Miseq monitor starts up.
The Miseq monitor should write the run ids into a text file as it starts the Kive runs for a batch of samples. Then if it gets restarted, it can treat those runs the way it currently treats active runs - reuse them if the results are still available.
This is more complicated than I thought, and I think it will be irrelevant once Kive supports searching for a run by its inputs as part of cfe-lab/Kive#739.
Closing the issue.
If a batch fails, the Miseq monitor resubmits all the samples. Kive then sifts through them to see which ones can reuse old results and which ones need to run. That can make Kive slow as the Miseq monitor starts up.
The Miseq monitor should write the run ids into a text file as it starts the Kive runs for a batch of samples. Then if it gets restarted, it can treat those runs the way it currently treats active runs - reuse them if the results are still available.