- Start Small: Begin with small test jobs to verify your setup
- Monitor Resources: Use
squeue and sacct to track jobs
- Set Time Limits: Always specify reasonable time limits
- Log Output: Use
--output and --error to capture logs
- Use Job Arrays: For multiple similar tasks, use job arrays instead of submitting many individual jobs
- Resource Estimation: Request slightly more resources than you think you need
- Check Output Files: Always verify output logs for errors
- Use Job Dependencies: For workflows that need sequential execution
- Clean Up: Remove unnecessary output files and cancel unused jobs