strapi-community/plugin-rest-cache/configure-rest-cache

Configuring REST Cache

Do not write a configuration before establishing the four things below. Most bad cache configurations are bad because one of them was assumed.

Establish first

1. How many Strapi instances run in production? More than one means the

memory provider is wrong — each process would hold its own cache and a purge would only reach one of them. That decision is not tunable later without changing infrastructure.

1. Is the traffic public, authenticated, or both? Authenticated caching is

safe only with per-caller keys, and is off by default for that reason.

1. Which routes actually hurt? Caching everything is rarely the goal.

A few expensive endpoints often account for most of the load.

1. How stale may content be? Usually the answer is "it may not be" — which

is fine, because invalidation is automatic. maxAge is a backstop for changes nothing observed, not the primary mechanism.

Ask the user these if the answers are not already evident from their project.

Then pick a starting point

Each of these is a complete, copy-pasteable configuration with the trade-offs written out. Prefer sending the user to the matching one over inventing a config from scratch.

Full option reference: https://strapi-community.github.io/plugin-rest-cache/guide/reference/config.html

Rules that are not negotiable

Durations are milliseconds. maxAge: 3600 is 3.6 seconds. One hour is 3600000. Write the value with a comment saying what it is in human units.

Never set `hitpass: false` without `keys.useAuth: true`. Together they cache authenticated traffic keyed per caller. hitpass: false alone means two callers authorised for the same route share one entry, and whoever misses first decides what everybody else sees. The server logs a warning at boot if you do this, and the admin panel flags the content type — but by then it is already serving.

Custom route paths must include the API prefix, exactly as Strapi registered them: /api/articles/slug/:slug.

Match `keysPrefix` to your Redis `keyPrefix` if the connection sets one. A mismatch means the plugin enumerates nothing, and every purge silently does nothing.

Sanity-check the result

Have the user turn on enableXCacheHeaders, then:

curl -sI http://localhost:1337/api/articles | grep -i x-cache   # MISS
curl -sI http://localhost:1337/api/articles | grep -i x-cache   # HIT

Then edit that content in the admin panel and request again — it should be MISS, proving invalidation works.

HITPASS means the request bypassed the cache; with the default hitpass a browser session will always do that. Test with curl.

If any of that does not behave, switch to the diagnose-rest-cache skill.

Writing the config

Both ./config/plugins.js and ./config/plugins.ts are supported. Match whichever the project already uses, and keep the plugin's block under the rest-cache key with everything inside config.