Chapter 00Free
What Slurm is
A complete free foundation chapter. Learn what Slurm is, why shared clusters need it, how jobs enter the system, and how to prove what happened.
10 focused lessons + 1 advanced lessonTrue zero to confident Slurm user
Slurm Core Operations
Operate Slurm from first request to evidence-backed diagnosis.
Learn Slurm work from first request to verified history: inspect resources, submit jobs, diagnose waits and failures, build workflows, and recover safely.
No credit card for the free chapter. Preview what the chapter covers.
By the end
Observable skills you will practice.
- Explain what Slurm is and how scheduling decisions flow
- Submit, monitor, cancel, and verify batch and interactive work
- Right-size CPU, memory, time, partition, array, dependency, MPI, and GPU requests
- Diagnose failures from queue state, reason codes, accounting, exit codes, and node evidence
- Recognize policy, QOS, fairshare, reservation, and operator-level causes
Performance-assessed credential
Earn the todo.academy Skills Credential: Slurm Core Operations.
Complete the focused path, then pass a hands-on final assessment whose required actions are replayed and checked before eligibility is created.
- 01Complete the focused path
Clear every required course checkpoint. Advanced practice remains optional.
- 02Pass the final assessment
Produce the required results without using optional readiness scenarios as a shortcut.
- 03Share only when you choose
Download the artifact, add it to LinkedIn, or share a public status page with verifiable evidence.
Independent skills credential issued solely by todo.academy. Not affiliated with, sponsored by, endorsed by, or issued by SchedMD LLC. Slurm is a registered trademark of SchedMD LLC.
Syllabus
26 chapters from the published course structure.
The focused path contains 42 required lessons. The advanced library remains available when your role needs more depth.
Focused path chapters11 chapters contain the required 42-lesson path.
Chapter 01Full course
Batch scripts as contracts
Learn how a shell file becomes a Slurm job request, how directives shape the record, and how to prove output, errors, paths, and accounting.
5 focused lessons + 6 advanced lessonsChapter 02Full course
Live queue triage
Read live queue state, decode pending reasons, filter noisy output, inspect detailed job records, and choose the next action from evidence.
5 focused lessons + 6 advanced lessonsChapter 03Full course
Resource requests and fit
Understand partitions, time, CPUs, tasks, memory, GPUs, and why a request fits, waits, or needs repair.
6 focused lessons + 5 advanced lessonsChapter 04Full course
Interactive work and job steps
Use srun, salloc, job steps, short interactive shells, queue proof, and cleanup without bypassing the scheduler.
4 focused lessons + 7 advanced lessonsChapter 05Full course
Software environment and reproducibility
Use modules, runtime version proof, Slurm job variables, clean logs, and accounting to make jobs reproducible.
4 focused lessons + 7 advanced lessonsChapter 06Full course
Accounting and postmortem evidence
Use sacct fields, step rows, MaxRSS, DerivedExitCode, sstat, seff, stderr, and job detail to diagnose completed or failed work.
3 focused lessons + 9 advanced lessonsChapter 07Full course
Failure diagnosis and recovery decisions
Classify cancellation, signals, node failure, preemption, requeue, timeout, OOM, and application failure before choosing the next action.
1 focused lesson + 11 advanced lessonsChapter 08Full course
CPU and memory troubleshooting
Diagnose CPU waste, threaded shape, memory pressure, memory per CPU, live stats, efficiency summaries, and right-sized fixes from evidence.
1 focused lesson + 11 advanced lessonsChapter 09Full course
The incident gauntlet
Classify Slurm incidents, gather proof, rank hypotheses, choose safe action, verify outcomes, and escalate with exact evidence.
2 focused lessons + 14 advanced lessonsChapter 10Full course
Final professional capstone
Own a deadline workflow from cluster snapshot through preprocess, array repair, GPU, MPI, hidden failure diagnosis, fixed report, and final evidence packet.
1 focused lesson + 12 advanced lessonsExplore 15 advanced library chapters
These chapters provide optional drills and reference depth. They are not required to complete the focused path.
Library 07Full course
Chapter 7: Job arrays and polite scaling
Scale repeated work with array indexes, task variables, safe per-task logs, throttles, targeted reruns, and accounting proof.
11 advanced lessonsLibrary 08Full course
Chapter 8: Dependencies and workflow graphs
Build workflow edges with afterok, afterany, afternotok, fan-in, OR release, never-satisfied diagnosis, singleton guards, and accounting proof.
12 advanced lessonsLibrary 12Full course
Chapter 12: Accounts, QOS, and priority
Read account associations, QOS limits, policy holds, priority factors, fairshare context, and repaired debug QOS requests from evidence.
12 advanced lessonsLibrary 13Full course
Chapter 13: MPI, GPU, and placement
Prove rank layout, multi-node distribution, CPU binding, exclusive allocations, GPU GRES, GPU visibility, and TRES accounting.
12 advanced lessonsLibrary 14Full course
Chapter 14: Operator evidence and safe state changes
Read daemon health, config intent, partitions, nodes, reservations, GRES, cgroups, SlurmDBD policy data, and safe scontrol update behavior.
12 advanced lessonsLibrary 15Full course
Chapter 15: Federation, heterogeneity, and site-scale policy
Read federation scope, heterogeneous components, submit plugins, file staging, usage rollups, site variance, and production runbook discipline.
12 advanced lessonsLibrary 16Full course
Chapter 16: Scheduler internals and backfill diagnosis
Explain scheduler behavior from config, diagnostics, start estimates, priority, preemption, topology, and final diagnosis packets.
12 advanced lessonsLibrary 17Full course
Chapter 17: Reservations, maintenance windows, and planned capacity
Read reservation inventory, access scope, reservation submissions, pending reasons, maintenance flags, safe updates, deletion guards, and planned down evidence.
12 advanced lessonsLibrary 18Full course
Chapter 18: Preemption, requeue, and restart-safe work
Prove preemption policy, read requeued and suspended states, handle signals, protect output, checkpoint progress, and repair long jobs for safe restart.
12 advanced lessonsLibrary 19Full course
Chapter 19: Containers and reproducible environments
Use Slurm container requests with site-aware OCI config, clean environment boundaries, bind paths, GPU exposure, policy diagnosis, and proof-backed reproducibility.
12 advanced lessonsLibrary 20Full course
Chapter 20: Operator view, daemons and control plane
Read controller health, node daemon evidence, accounting delay, authentication trust, logs, diagnostics, and incident classification without unsafe guesses.
12 advanced lessonsLibrary 21Full course
Chapter 21: Configuration as policy code
Read slurm.conf, node and partition definitions, includes, configless distribution, GRES mapping, protected SlurmDBD config, reconfigure safety, and rollback evidence.
12 advanced lessonsLibrary 22Full course
Chapter 22: Accounts, associations, limits, and billing
Read accounting policy data, association tuples, users, accounts, QOS allowed lists, TRES-minute limits, fairshare context, billing evidence, onboarding, and access failures.
12 advanced lessonsLibrary 23Full course
Chapter 23: Throughput, scheduler load, and good citizenship
Refactor noisy repeated work into polite high-throughput arrays with throttles, targeted polling, clean output, responsible chunking, and controller diagnostics.
12 advanced lessonsLibrary 24Full course
Chapter 24: Advanced placement, dynamic capacity, and federation
Reason about topology-aware placement, node features, dynamic and cloud-style capacity, heterogeneous components, federation scope, and final advanced-placement proof.
12 advanced lessonsFull access
Choose access for Slurm Core Operations
Every available offer is shown with its exact CAD price and billing model. Checkout opens only after you choose an offer and enter the receipt email.
Questions
Know what to expect before you start.
What can I try for free?
Chapter 1: What Slurm is is free and contains 11 lessons. No credit card is requested before the free workspace opens.
What experience do I need?
No Slurm experience is required. Basic command-line familiarity is helpful.
What does the focused path include?
42 required lessons, with 10 free and 32 included in paid access.
Is there material beyond the focused path?
Yes. The course also includes 268 advanced practice and reference resources. They are available when you need more depth, but they do not lengthen the required path.
Which paid options are available?
Slurm Founder Pass: CA$59 one-time. Academy Founder Annual: CA$139 per year. Founder Vault: CA$279 one-time.
How is paid access billed?
Slurm Founder Pass: No recurring charge. Access remains available unless the purchase is refunded or reversed. Academy Founder Annual: Access continues while the annual subscription remains active. Founder Vault: No recurring charge. Coverage includes courses launched during the first 24 months.
How long does it take?
The published estimate is 6 to 10 active hours for Slurm Core Operations. Your pace will depend on how much you repeat the practice.
Do I need to install anything?
No installation is required to start the guided free chapter. Later lessons explain the real tools, files, and operating boundaries relevant to the skill.
What technology is covered?
The syllabus and these stated outcomes are the source of truth: Explain what Slurm is and how scheduling decisions flow; Submit, monitor, cancel, and verify batch and interactive work; Right-size CPU, memory, time, partition, array, dependency, MPI, and GPU requests; Diagnose failures from queue state, reason codes, accounting, exit codes, and node evidence; Recognize policy, QOS, fairshare, reservation, and operator-level causes.
Which browsers are supported?
Use a current browser with JavaScript enabled. The free chapter is the quickest compatibility check for your device.
Does the workspace work on mobile?
The reading pages reflow for small screens. Command-heavy practice is more comfortable with a physical keyboard and a larger display.
What happens when I make a mistake?
The practice state is isolated from production systems. Read the resulting evidence, revise the action, and try again.
How do I restore access?
Use the receipt email on the restore-access page. The sign-in link verifies the account before paid entitlements are loaded.