> Since Cloudflare’s Workers can run before, and after the cache I'm a little confused by this page, it seems like all of the examples below only mention ways to run the worker before the Cloudflare cache. It mentions the browser cache, so is that what it means by after? Or can you actually run workers behind a CDN cache? (Meaning the worker only runs if the request isn't cached)

the

fetch()

funciton is what goes to the cache (or origin if it's no in there)

so before the function executes is before the cache, after it executes, is after the cache

basically a worker is a middleware

I'm confused by that answer, I don't think it's really answering my question?

I guess I don't understand your question

> It mentions the browser cache, so is that what it means by after? No

> Or can you actually run workers behind a CDN cache? Yes.

-ish

it's infront and behind the cache simulatniously

because it's a middleware

The essence of the question is if it's possible to only invoke a worker if a cached response from that worker is not available

no

That paragraph is misleading, we'll need to adjust the language. Workers always run before cache. They can interact with the Cache API, but it still runs before cache.

the worker always executes no matter way

*no matter what

Okay, that's why I think it's confusing! Because the docs say "Since Cloudflare’s Workers can run before, and after the cache", but it seems like they can only run before the cache.

well they run after in the sense that you can modify a response from the cache

your question is really: can I prevent a worker from executing? and the answer is: no

Workaround: Other CDN => Cloudflare Workers

I use BunnyCDN to cache the answers from my workers.

why do you need to cache it? like I'm struggling to understand what value it brings?

like let's say it takes a long time to generate a response, you can use https://developers.cloudflare.com/workers/runtime-apis/cache to store that in the cache and try and match it on the next reqeust...

less invocations!

why is the number of invocations a problem? they are.... really cheap

My workers always produce the same results with the same inputs. And I have several million queries a day. A cache simply makes sense. This way I save 95% of the costs.

$5 instead of $100 makes a difference

My analytics are currently broken so im not really sure how many I have lol

it's either 519k or 287k per day. maybe the first number is last 24 hours and the second number is today so far? not sure

I thought i saw something in the docs about KV automatic caching but now I can't find it