Closed RickKessler closed 1 year ago
Current workflow is:
hash_old.txt
so that if ALL.DONE is set to SUCCESS, you can simply mv hash_old.txt hash.txt
to force Pippin into accepting the change. Note that if the Pippin job has changed, the hash will still be different and pippin will want to rerun. In that case you can just use -i N
where N is the task number, to ignore hash checks and not rerun up to and included the Nth task
If a pippin task fails, user should be able to manually fix it and set SUCCESS in ALL.DONE. Example case is LC fit on ~10^5 jobs, and only 1 job fails.