Here are the areas that should be documented (see gittrac
#2815). I re-ordered them somewhat to make the list more logical in terms of related functionality.
- condor_submit_dag/scheduler universe startup
- DAG file parsing
- Rescue DAG/condor_rm of DAGMan
- recovery mode/condor_hold/condor_release
- proper ordering of jobs/scripts (ready queue, etc.)
- implementation of PRE/POST script semantics
- reading the user logs
- deciding when the DAG is finished
- ABORT-DAG-ON
- final node
- node variables
- nested DAGs
- Multiple disjoint DAGs
- splices
- Node retries
- submit retries
- DAG-wide throttles (maxjobs, maxidle, maxpre, maxpost)
- category throttles
- depth-first DAG traversal
- node priorities
- halt file
- NOOP nodes (at DAG level)
- multi-job clusters
- strict warnings
- DIR on nodes
- node status file
- jobstate.log file
- dot file
- Stork support
- Miscellaneous notes