Platform engineering ·
How CI/CD and Production Troubleshooting Fit Together
By Manu Mathew, Technical Lead Engineer · Published 3 October 2026 · Last updated 5 October 2026

A release goes out on a quiet afternoon. An hour later, something in production looks wrong. What happens next depends on two things: how the change was shipped, and how calmly the team can work out what broke.
That is why I think of CI/CD and production troubleshooting as one job rather than two. CI/CD stands for continuous integration and continuous delivery: building, testing and releasing every change through the same automated path.
Quick answer
CI/CD makes every change go out the same tested way, and production troubleshooting shows whether that change was actually safe. A technical lead treats them as one loop: ship through a pipeline people trust, watch production closely, find the specific cause when something breaks, and feed the fix back into the pipeline as a test, alert or safer default.
What do CI/CD and production troubleshooting have in common?
Both are about learning what a change really did, as early and as cheaply as possible.
A pipeline catches some problems before release. Production shows the rest. If the two are owned separately, lessons from incidents rarely reach the pipeline, and the same kind of failure returns.
What should a pipeline do for a team?
It should make the safe way to release the easy way, so nothing depends on someone remembering the steps.
- Build the same way every time, from version control, with pinned dependencies.
- Run the tests that matter for each change, quickly enough that people don’t skip them.
- Release in small steps, so each change is easy to understand and to roll back.
- Record what went out and when, so troubleshooting starts from facts.
- Use the same path for a one-line fix and a large rollout; emergency shortcuts are where mistakes hide.
How do you troubleshoot a production problem calmly?
Stabilise first, then find the specific cause, and only then fix it properly.
- Limit the impact. Roll back or switch off the change if that is safe; explaining comes later.
- Gather evidence: what changed, when the symptoms started, what logs and metrics show.
- Form one hypothesis at a time and test it, rather than changing several things at once.
- Make the smallest fix that addresses the cause, and confirm it in production.
- Write it up while the details are fresh.
The useful outcome is a cause, a fix, and a change that makes the same failure harder next time: a new test, a clearer alert, a safer default or a smaller interface.
What makes a useful post-incident review?
It changes the system, not just the story. A review that ends without a change to the platform is only a record.
Google’s SRE book has a clear chapter on blameless postmortem culture. The idea is simple: people share what really happened when they aren’t worried about blame, and that honesty is what makes the fix accurate.
How does this help engineers learn?
Engineers learn a platform by shipping through it and by working through real incidents with someone experienced.
Part of a lead’s job is to keep both paths open: a pipeline people trust enough to use for everything, and a habit of looking at production until the cause is specific. Pairing a newer engineer with an experienced one during an incident review often teaches more than any document.
Does this apply to a small WordPress site?
Yes, at a smaller scale. A small site benefits from the same loop: a staging copy, a repeatable way to deploy, backups you have tested and a quick check after every update.
Why a WordPress site is slow walks through measuring before and after a change, which is the same habit applied to speed.
Frequently asked questions
What is the difference between continuous delivery and continuous deployment?
With continuous delivery, every change that passes the pipeline is ready to release, and a person decides when it goes live. With continuous deployment, every change that passes goes live automatically. Many teams start with delivery and move to deployment once their tests and monitoring are trusted.
What should be checked before a change reaches production?
At least an automatic build, the tests that matter for the change, and a check that configuration is valid. For risky changes, add a staging environment that matches production, and release to a small share of traffic first, so a problem affects few users and is easy to roll back.
What is a blameless post-incident review?
A review that looks for the conditions that allowed a failure, not for someone to blame. People describe what they saw and did openly, and the outcome is a short list of changes to systems and processes. Blame tends to hide information, which makes the next incident harder to understand.
How can a small team start with CI/CD?
Start by automating what already happens by hand: run the tests on every push, build the same way every time, and deploy with one scripted command instead of copying files. Add a staging copy next. Each step removes a chance for the release to depend on one person’s memory.
Related posts: what platform engineering owns; what highly available means for hosting; my journey to technical lead.
Want to talk?
If you’d like to compare notes on platform work, or you need help with a WordPress site or Perl code, I’m happy to hear from you. You can reach me through the contact section or email me at manu@manu.co.in.