Cross-Topic Study Paths

These paths are designed for repeated practice. Each path crosses multiple topic boundaries because real incidents rarely respect the guide’s navigation tree.

Use the page completion control on individual topic pages as you finish each step. The path cards below calculate progress locally in the browser so you can repeat the same path without changing the repository.

Linux Foundations to Operations

Build from boot, processes, filesystems, users, networking, and observability into practical host triage.

0 of 6 complete
  1. Boot and Userspace Not started
  2. Processes and Threads Not started
  3. Filesystems and IO Not started
  4. Network Stack Not started
  5. Logs and Observability Not started
  6. Performance Triage Runbooks Not started

Kubernetes Networking Incident Path

Follow a request through Pod networking, Services, DNS, policy, ingress, and packet capture.

0 of 6 complete
  1. Kubernetes Networking Not started
  2. Pod Networking and CNI Not started
  3. Services and EndpointSlices Not started
  4. Kubernetes DNS and CoreDNS Not started
  5. NetworkPolicy Not started
  6. Packet Capture and Analysis Not started

PostgreSQL Reliability and Zero Downtime

Connect PostgreSQL HA, PgBouncer, CloudNativePG, backups, and upgrade mechanics into one operational path.

0 of 5 complete
  1. PostgreSQL Not started
  2. PostgreSQL Operations and HA Not started
  3. PgBouncer Not started
  4. CloudNativePG Not started
  5. PostgreSQL Zero-Downtime Upgrades Not started

Production ML from 101 to Advanced Systems

Move from ML basics through transformers, RAG, agents, serving, observability, evaluation, and security.

0 of 11 complete
  1. ML 101 Foundations Not started
  2. Transformer Internals Not started
  3. Retrieval-Augmented Generation Not started
  4. Agents and Tool Use Not started
  5. Serving, Inference, and vLLM Not started
  6. LLM Inference Systems Not started
  7. Model Memory Math Not started
  8. Inference Benchmarking Not started
  9. Inference Runbooks Not started
  10. Evaluation and CI/CD Not started
  11. ML Security Threats Not started

Cross-Layer Incident Response

Practice request-path debugging across DNS, proxies, load balancers, Kubernetes, Linux, and databases.

0 of 6 complete
  1. Incident Entry Points Not started
  2. Request Path Not started
  3. Cross-Layer Incident Runbooks Not started
  4. Resilience, Timeouts, and Draining Not started
  5. Load Balancers and Proxies Not started
  6. Kubernetes Troubleshooting Not started

Study Cards

Question

Why use cross-topic study paths?

Answer

They practice the same boundary crossings that production incidents require.

Question

Which path should start a 504 investigation?

Answer

Request Path, then Cross-Layer Incident Runbooks, load balancer behavior, timeout budget, and Kubernetes Services.

Question

Why does cloud egress need DNS and NAT together?

Answer

DNS selects the target address, and that address determines whether traffic uses private routing, NAT, or public egress.

Question

Why start distributed storage debugging with placement?

Answer

Placement and failure domains explain which OSDs should hold data and why recovery or fullness affects clients.

Question

Why separate RAG retrieval from generation?

Answer

A model cannot ground an answer in a source the retriever failed to return.

References