Great question. Free can definitely purge by cache-tags now, so I would think yes, but I've not test

Great question. Free can definitely purge by cache-tags now, so I would think yes, but I've not tested

JamesOP•4/4/25, 3:37 PM

Or maybe not, since this is for

fetch

fetch

cacheKeycacheKey. It's still documented as ent only at https://developers.cloudflare.com/workers/examples/cache-using-fetch/#custom-cache-keys

JJames Great question. Free can definitely purge by cache-tags now, so I would think ye...

dave•4/4/25, 4:01 PM

cacheTagcacheTag says enterprise only too though. Is there another way to create cache tags that I'm unaware of?

Walshy•4/4/25, 4:01 PM

i still need to verify if this works or not yet on non-ent

Walshy•4/4/25, 4:02 PM

but the cache-tags header does work on non-ent, someone in support already tested that

WWalshy i still need to verify if this works or not yet on non-ent

dave•4/4/25, 4:02 PM

you talking about cf.cacheTagscf.cacheTags or cf.cacheKeycf.cacheKey?

Walshy•4/4/25, 4:03 PM

cacheTags specifically ~~but i assume cacheKey would be in the same boat~~

WWalshy but the cache-tags header does work on non-ent, someone in support already teste...

dave•4/4/25, 4:03 PM

Is there a difference between using cf.cacheTagscf.cacheTags vs. Cache-TagCache-Tag as a header in a Worker subrequest?

Walshy•4/4/25, 4:05 PM

just convience

dave•4/4/25, 4:06 PM

Do cache tags impact cache hits, or is it only for purging?

dave•4/4/25, 7:53 PM

Does storing things compressed (with gzip) vs uncompressed with the Cache API make any real difference?

WWalshy cacheTags specifically ~~but i assume cacheKey would be in the same boat~~

Zack0322•4/4/25, 8:07 PM

to be clear cachetags is now available to everyone. but cacheKey i.e custom cache keys is still ENT only

Ddave Do cache tags impact cache hits, or is it only for purging?

Zack0322•4/4/25, 8:07 PM

it's only for purging, it won't affect caching decisions

ZZack0322 to be clear cachetags is now available to everyone. but cacheKey i.e custom cach...

Walshy•4/4/25, 8:08 PM

ah cool thanks

Ddave `cacheTag` says enterprise only too though. Is there another way to create cache...

Zack0322•4/4/25, 8:12 PM

can you link me to the page that says cachetags is for ENT only? we've been updating various pages to remove the "ENT only" part but we may have missed some

ZZack0322 can you link me to the page that says cachetags is for ENT only? we've been upda...

dave•4/4/25, 8:12 PM

that's from within VS Code

Zack0322•4/4/25, 8:13 PM

guess it's in some code/library?

dave•4/4/25, 8:13 PM

https://github.com/search?q=repo%3Acloudflare%2Fworkerd%20%22Only%20available%20for%20Enterprise%20customers.%22&type=code

GitHub

Build software better, together

GitHub is where people build software. More than 150 million people use GitHub to discover, fork, and contribute to over 420 million projects.

Zack0322•4/4/25, 8:13 PM

thank you, i'll chat with some people to get that updated

dave•4/4/25, 8:22 PM

Do cache rules apply when trying to insert the cache with cache.putcache.put in a Worker?

dave•4/4/25, 9:09 PM

Will the Cache API miss if there's a difference in user-agent? https://developers.cloudflare.com/cache/how-to/cache-keys/#headers

Cloudflare Docs

Cache Keys · Cloudflare Cache (CDN) docs

A Cache Key is an identifier that Cloudflare uses for a file in our cache, and the Cache Key Template defines the identifier for a given HTTP request.

Kkamama how to calculate the Execution duration? will it cost? or worker only charge for...

kamama•4/5/25, 1:22 AM

Any info about this ?

JamesOP•4/5/25, 1:28 AM

Workers only charge based on requests and CPU time these days. The only exception to this is Durable Objects, which are still wall-time based.

kamama•4/5/25, 2:39 AM

So worker only charge by request and CPU time in metric now? The rest indicator for reference only ?

Kkamama So worker only charge by request and CPU time in metric now? The rest indicator ...

Isaac McFadyen•4/5/25, 2:45 AM

Correct, if you are not using Durable Objects then you will only be charged based on requests and CPU time, not wall time.

kamama•4/5/25, 2:45 AM

Wow, then SSE is a good solution now

andifined.dev•4/5/25, 6:23 AM

Hello!

I‘m looking for a way to process Videos using workers or workflows. In Detail my goal is to cut an image out of a video that’s stored in R2 every 1 second and persist it in R2 again. I thought maybe media transformations would work, but I don‘t Thank it fits my needs quite well because of caching and requiring an URL to work. Are there any other approches for that?

く

くもりにくもったクラウドサービスlocaler•4/5/25, 2:35 PM

I'm trying to use Argon2 with SvelteKit (on Cloudflare Pages) and Argon2 WASM (on Cloudflare Workers) via service bindings,
but I keep getting the error: "Worker exceeded CPU time limit."
Is there any way to fix or work around this issue?

く

くもりにくもったクラウドサービスlocaler•4/5/25, 2:37 PM

Do I need to upgrade to a paid plan to make it work?

seflanka•4/5/25, 6:54 PM

hello

Avi•4/6/25, 6:37 AM

is pages officially deprecated yet?

JamesOP•4/6/25, 12:11 PM

Not officially yet, and I’d be surprised if they ever do, but I would recommend Assets for new projects. Pages is already lagging behind in terms of new bindings its supports, and you’ll have a better experience with Assets.

Silvan•4/6/25, 2:13 PM

I still wish we had a better way of binding services like DOs to meta frameworks without a monorepo :/

Often you just dont have “access” to the server entrypoint due to it being generated

Silvan•4/6/25, 2:18 PM

unless someone has a cool solution. in case of sveltekit you have this

main = ".svelte-kit/cloudflare/_worker.jsmain = ".svelte-kit/cloudflare/_worker.js

and i really haven’t found a way to make it include a DO class inside that file

SSilvan I still wish we had a better way of binding services like DOs to meta frameworks...

DaniFoldi•4/6/25, 2:48 PM

I'd actually recommend deploying them separately, wherever it makes sense:

if you use websockets, a new deployment will disconnect all clients, so you'll get a spike of reconnects - and it could be unnecessary if you didn't actually change your websocket handling DO code

DDaniFoldi I'd actually recommend deploying them separately, wherever it makes sense: if y...

Silvan•4/6/25, 2:51 PM

oh thats actually a very valid point :thinkies:

Even do having a DO repo seems silly but i dont think you can have a monorepo at all then if you use git for deployments due to it triggering both if they both look at the same repo

Silvan•4/6/25, 2:51 PM

but good input thanks!

Silvan•4/6/25, 2:52 PM

(also dev experience suffers a bit from it but cant have it all)

SSilvan (also dev experience suffers a bit from it but cant have it all)

DaniFoldi•4/6/25, 2:54 PM

If it's in the same repo, you can define a script that starts

wrangler dev

wrangler dev

with all of the workers

DDaniFoldi If it's in the same repo, you can define a script that starts `wrangler dev` wit...

Silvan•4/6/25, 2:57 PM

yea im aware but that would enforce deploying the DO over the cli if im not completely off here? else i dont see how they could be deployed separate from each other

DaniFoldi•4/6/25, 2:59 PM

you can think of Durable Objects as separate workers, and as long as your code remains compatible (since you can't deploy at the exact same second, building in some compatibility is always best), you can deploy a new version to only one or the other
yes, workers CI would trigger on both of them if you push to the repo by default, but you can check https://developers.cloudflare.com/workers/ci-cd/builds/build-watch-paths/ on how to set up watch paths so they only run if the relevant code is changed

Cloudflare Docs

Build watch paths · Cloudflare Workers docs

Reduce compute for your monorepo by specifying paths for Workers Builds to skip

Silvan•4/6/25, 3:01 PM

oh wow thats the point of watch paths omg learning something new everyday! thanks a lot

Kotkoroid•4/6/25, 3:43 PM

Hi, should I be putting worker-configuration.d.tsworker-configuration.d.ts into compilerOptions.typescompilerOptions.types or

include

include

? Docs recommend the first option but I'm getting:
error TS2688: Cannot find type definition file for 'proxy/gateway/worker-configuration.d.ts'.
│ The file is in the program because:
│ Entry point of type library 'proxy/gateway/worker-configuration.d.ts' specified in compilerOptions

The file exists.

tsconfig:
{
"compilerOptions": {
"target": "ESNext",
"module": "ESNext",
"moduleResolution": "Bundler",
"strict": true,
"skipLibCheck": true,
"lib": ["ESNext"],
"types": [
"@cloudflare/workers-types",
"./proxy/gateway/worker-configuration.d.ts"
]
}
}

No problem when added to

include

include

but I think that's not where d.tsd.ts should be?

Deltachaos•4/7/25, 12:00 PM

reagrding the free tier limits:
it says 10ms cpu time for workers. Some of my requests have something around 77ms cpu time (and worked). Can i assume that this will not work reliably? Or is the average cpu time relevant for the limits? And if so, how do crons play into this?

For example i have a cron that accidentally was spending 1,329ms cpu time before it got interupted. Now following crons are failing with something like Exception in callback <_asyncio.TaskStepMethWrapper object at 0x1a38940>()Exception in callback <_asyncio.TaskStepMethWrapper object at 0x1a38940>() with 0ms cpu time. Can i assume that they are throttled? How long do such throttles last?

hffmnn_•4/7/25, 12:37 PM

Hi. I am running into the "I need multiple fallback origin" problem. That means that I have multiple apps aka. workers that need to be available under different custom, external hostnames. So I would try out the proxy worker workaround aka. the fallback origin will point to a worker that routes to other workers internally. Does anyone know about a write-up that sums up problems with that approach?

dave•4/7/25, 6:14 PM

Is there something like