DevOps Last-Minute Revision

30 quick revision notes for DevOps

1. chmod 755 gives the owner read, write and execute, and gives group and others read and execute permissions.
2. SIGTERM (15) asks a process to stop gracefully, while SIGKILL (9) cannot be caught and kills it immediately.
3. Load average counts running and waiting processes; compare it with the number of CPU cores to judge whether a server is overloaded.
4. If df shows a full disk but du does not, a deleted file is probably still held open by a process; check with lsof +L1.
5. set -euo pipefail makes bash scripts stop on errors, unset variables and failures inside pipelines.
6. git fetch only downloads remote changes, while git pull fetches and then merges or rebases them into your branch.
7. Use git revert on shared branches because it adds an undo commit; git reset rewrites history and suits local work only.
8. git reflog records where HEAD has been and can recover lost commits or deleted branches on the same machine.
9. Continuous delivery keeps every build releasable with a manual approval; continuous deployment releases automatically.
10. Build an artefact once and promote the same version through dev, staging and production for reliable releases.
11. Canary releases send a small share of traffic to a new version first, while blue-green switches all traffic at once.
12. The four DORA metrics are deployment frequency, lead time for changes, change failure rate and time to restore service.
13. Containers share the host kernel and use namespaces for isolation and cgroups for resource limits.
14. Order Dockerfile instructions from least to most frequently changing so dependency layers stay cached.
15. Multi-stage builds keep compilers and source code out of the final image, making it smaller and more secure.
16. Use exec form for CMD and ENTRYPOINT so the application runs as PID 1 and receives SIGTERM on docker stop.
17. Named volumes keep container data such as databases safe when the container is removed or recreated.
18. The Kubernetes control plane has the API server, etcd, scheduler and controller manager; nodes run kubelet, kube-proxy and a runtime.
19. A Deployment manages ReplicaSets, which keep the desired number of identical pods running and enable rolling updates.
20. A failing liveness probe restarts a container, while a failing readiness probe only removes the pod from service endpoints.
21. Exceeding a CPU limit throttles a container, but exceeding a memory limit gets it OOM killed with exit code 137.
22. CrashLoopBackOff is debugged with kubectl describe pod and kubectl logs --previous to see why the container keeps exiting.
23. Terraform state maps code to real resources; store it in a remote encrypted backend with locking, never in Git.
24. Prefer for_each over count for named collections so removing one item does not recreate the others.
25. Ansible is agentless over SSH and its modules are idempotent, so rerunning a playbook makes no unwanted changes.
26. Security groups are stateful and allow-only; network ACLs are stateless, work at subnet level and support deny rules.
27. Private subnets reach the internet through a NAT Gateway in a public subnet, never directly through the Internet Gateway.
28. Give EC2, ECS and Lambda access to AWS services through IAM roles with temporary credentials, not access keys.
29. An SLO of 99.9 percent allows about 43 minutes of downtime per 30-day month, which is the error budget.
30. Use rate() on Prometheus counters and histogram_quantile on histogram buckets to calculate p95 or p99 latency.

Night-before checklist

• Revise your saved questions on this page once, aloud.
• Read the quick notes and mark any you cannot explain in one line.
• Rehearse a two-minute self-introduction and two project stories.
• Keep CV, ID, laptop charger and interview link ready; sleep on time.