Some users receive an old HTML shell with new assets, or a new shell that requests deleted chunks after deployment. How would you diagnose and prevent it?

Diagnoses incompatible cached HTML and immutable assets across CDN, browser, service-worker, and rollout boundaries.

1Say itAnswer out loud first

2CompareCompare with Core answer

3RefineRefine with Stronger and Advanced

TL;DR

I capture the failing document URL, requested chunk URL, response status, cache headers, age, CDN result, service-worker state, and deployment version. Hashed assets can be cached for a long time because their URLs change with content, but HTML should revalidate frequently. I keep old hashed assets available long enough that cached HTML can still load them.

Full answer below

Practice focus

Production judgment about cache policy, atomic compatibility, evidence, rollback, and user recovery.

Answer first

Answer path

Core answer first, then checkpoints toward a stronger answer.

Start
Core answer
Start here. This is the direct spoken answer to practice first.
Overview
The failure is a version-skew problem. The investigation must identify which layer served each artifact and which versions are compatible.
I capture the failing document URL, requested chunk URL, response status, cache headers, age, CDN result, service-worker state, and deployment version. Hashed assets can be cached for a long time because their URLs change with content, but HTML should revalidate frequently. I keep old hashed assets available long enough that cached HTML can still load them.
Checkpoint
Stronger answer
Optional upgrade with examples, trade-offs, and sharper phrasing.
I reproduce with the affected browser and CDN region, then compare the HTML's asset manifest with what the origin and edge currently serve. A service worker may also return an older shell, so I inspect its cache and update lifecycle. The deployment should publish new assets before exposing new HTML and avoid deleting the previous generation immediately. A chunk-load failure can offer one controlled refresh rather than trapping the user in a reload loop.

Review signals

Use these to catch weak phrasing, shallow reasoning, and risky claims.

Common mistakesCandidateCommon mistakesMistakes that weaken the answer or signal shallow understanding.Open details

Details

Purging every cache repeatedly without identifying the stale layer.
Caching HTML as immutable alongside hashed assets.
Deleting the previous asset generation as soon as deployment finishes.
Fixing chunk failures with an unlimited reload loop.

Some users receive an old HTML shell with new assets, or a new shell that requests deleted chunks after deployment. How would you diagnose and prevent it?

Practice focus

Answer path

Overview

Review signals

Follow-up prompts

Continue practice

Try a variant

Skill areas

Human editorial direction

Was this drill useful?

Found this answer useful?