do worker -> tunnel connections work for non-http? eg. postgres
do worker -> tunnel connections work for non-http? eg. postgres
CYou can do it with Hyperdrive
https://developers.cloudflare.com/hyperdrive/configuration/connect-to-private-database/
https://developers.cloudflare.com/hyperdrive/configuration/connect-to-private-database/
Cloudflare Docs
Hyperdrive can securely connect to your private databases using Cloudflare Tunnel and Cloudflare Access.
Mthanks cyber, I played around with that yesterday but i've implemented multi-tenancy in my DB using a schema+role per-tenant... and since those are included in the hyperdrive config, i'd quickly breach the limits for number of configs allowed per account
im still researching but hyperdrive seems like the best option, if i do go ahead with it i'll have to switch back to a single schema with row-level security and update drizzle to apply RLS in migrations & then also ensure every transaction sets local settings to apply the auth
im still researching but hyperdrive seems like the best option, if i do go ahead with it i'll have to switch back to a single schema with row-level security and update drizzle to apply RLS in migrations & then also ensure every transaction sets local settings to apply the auth
Mim also doing per-tenant worker script deploys so its a shame that if i did move to hyperdrive, every worker would technically have access to the full db -- previously I had a nice setup where i provisioned the worker with just the schema name + user/pass needed to access its own data
ideally i'd switch to hainvg a D1 db per worker, but I need GIN indexes for fast filtering on a Record<string, string> column where the keys are not known ahead of time
ideally i'd switch to hainvg a D1 db per worker, but I need GIN indexes for fast filtering on a Record<string, string> column where the keys are not known ahead of time
Khas anyone here gotten sequelize to work on cloudflare workers? For me it just indefinitely hangs in the new Sequelize() method without any errors
SDo i understand correctly that the caches api overwrites the response with the data in the cache if its a hit?
In metaframeworks you often have a load function that runs before your page renders. i tried it there and on cache hit i only get my cache data back (not the website) even without writing New Response..
Just making sure that i have to extract the logic i want to cache inside an api route
In metaframeworks you often have a load function that runs before your page renders. i tried it there and on cache hit i only get my cache data back (not the website) even without writing New Response..
Just making sure that i have to extract the logic i want to cache inside an api route
Eawait cache.match(request) returns a ResponseHm in that case i should try again. Also weirdly enough when i check on github i also see some people do
platform.env.CACHES instead of just caches (with sveltekit) those should both be the same right?Sin my head the env is for bindings so its weird to see
Ei think (but could be wrong) thats just the way svelte chooses to expose it
SYea both work but platform is easier to type. even tho i still have the same issue but im assuming this is a sveltekit issue im having here
HJust to confirm, you are doing
platform.caches, not platform.env.CACHES, right?
Syes i missed a column i indeed use platform.caches
SBest place to post workers build issues is in #workers-help ? or somewhere else, I also have enterprise plan if I should post it in a ticket somewhere?
SHas anyone found a nice solution on typing
ResponseResponse
Chey everybody...getting a little frustrated not getting the functions folder in my svelte pages project to be recognized by wrangler and upload...seems so basic. It's at the top level, i used the cloudflare create to start the project, haven't done anything too fancy. the pages project loads and shows pages but the functions folder (and a basic script to show its working) is just not being uploaded and in the console when deploying it doesn't show being uploaded. Any ideas or some small thing I am not doing that I a missing? banged my head on this for hours, employed the help of grok 3...no go. Thank you to anybody that might know or might be aware of some flaw. Latest version of wrangler being used. New project that was created yesterday.
Cand now..with fog clearning..i see this:
https://developers.cloudflare.com/pages/framework-guides/deploy-a-svelte-kit-site/#functions-setup
So i am guessing this has something to do with it...
https://developers.cloudflare.com/pages/framework-guides/deploy-a-svelte-kit-site/#functions-setup
So i am guessing this has something to do with it...
Cloudflare Docs
Svelte is an increasingly popular, open-source framework for building user interfaces and web applications. Unlike most frameworks, Svelte is primarily a compiler that converts your component code into efficient JavaScript that surgically updates the DOM when your application state changes.
SFunctions contained in the /functions directory at the project's root will not be included in the deploymentSwhats stopping you from writing +server.ts endpoints?
Cjust learning that now, i (and all the AI's) were under the impression it followed the pattern of the functions folder set up. Nothing is stopping me, just new to pages and wasn't aware as this is the first mention of this difference on that link. searching the internet everywhere else I couldn't find that note.
SPages is a bit a hacky concept imo. Maybe why this exists now https://developers.cloudflare.com/workers/frameworks/framework-guides/svelte/
Cloudflare Docs
Create a Svelte application and deploy it to Cloudflare Workers with Workers Assets.
Cyeah that's where i found it. In my sunday experiment i had set up svelte pages without functions and then created a worker backend. But i was running in to CORS errors over and over again, so i thought maybe just using the functions feature (that i thought was available consistently on all pages frameworks) would be the way to go which then...destroyed me sunday
Scloudflare is very different to many other services i agree its a hard learning curve at the start
Cbut the irony is 2 minutes after i posted the question, i found that link. But still...it should be more front and center and wrangle IMHO should perhaps have a console error that says "we noticed a functions folder, please note svelte wont recognize this folder and yo must do it like ...., please refer to this example...."
Cthat would be easy...save developers HOURS.....if any cloudflare employees are watching this...
Spages really is meant to just use the framework i feel like
Swith worker with assets you could easily bind a functions folder
Ccan you explain what you mean by that?
C(i might be less advanced in my framework/worker expeirence)
Ahttps://developers.cloudflare.com/workers/static-assets/ Pages will eventually merge into Workers (not for a long time probably, at least Pages is unlikely to “stop working” for a long time would be my guess), and you can serve static assets from Workers now (in beta). You can use frameworks too, same as pages. See https://developers.cloudflare.com/workers/frameworks/framework-guides/svelte/ for Svelte. Either way you won’t need the functions folder, as if you’re using Svelte, just use its approach for server-side endpoints.
Cloudflare Docs
Create full-stack applications deployed to Cloudflare Workers.
Cloudflare Docs
Create a Svelte application and deploy it to Cloudflare Workers with Workers Assets.
Cnoted. I used svelte because grok suggested it for my small idea. I wanted something to practice with because i wanted to eventually convert a full django project i created in 2017 into a cloudflare workers/pages project instead of having to maintain this server and upgrade pythong and django etc.
Cbut maybe svelte was a bad choice to learn on
Cyou have way more experience with workers/pages, do you think moving django social bookmarking site built via in 2016 from pinax/django to a cloudflare site is a good idea, or will it cost a lot more to host in the long run and i should just update the django and python code and move it to a new server on say digital ocean
K
Cloudflare Docs
Learn about the limits associated with Browser Rendering.
SWhen i generate my types using
is this expected when just using
wrangler types --experimental-include-runtimeis this expected when just using
nodejs_als?AI have no experience of Django I’m afraid 
Svetle is a pretty easy framework to pick up though, I would say. And when using AI, just treat it as a standard Svetle app, for the most part that will work as the Cloudflare Pages adapter handles translation from Svetle -> Cloudflare. Only caveat will be if you use bindings to connect to databases and stuff on Cloudflare,
Svetle is a pretty easy framework to pick up though, I would say. And when using AI, just treat it as a standard Svetle app, for the most part that will work as the Cloudflare Pages adapter handles translation from Svetle -> Cloudflare. Only caveat will be if you use bindings to connect to databases and stuff on Cloudflare,Ci have a good amount of expeirence on a past project with workers and D1/KV but haven;t been keeping up with the new stuff you just linked me too. In the past workers project it was using kv-asset-handler , is https://developers.cloudflare.com/workers/static-assets/ replacing that for some things?
Cloudflare Docs
Create full-stack applications deployed to Cloudflare Workers.
AI believe so, yes. Static Asset Workers effectively store the assets in KV under the hood. I think using Pages is fine for what it’s worth, considering Static Asset Workers are in beta, and there’s probably a bajillion sites using Pages, so they aren’t just going to suddenly kill it (well, I would be very shocked )
)Cshould i update my current project using kv-asseet-handler to static asset workers? Do you know of a tutorial or how to do to that?
AIt’s pretty simple to setup, docs are linked above - effectively you designate a folder as containing your static assets, and some config in wrangler.toml, and any request to a path matching a static asset will be served
Cand that is safe to say will be the future moving forward?
AYes, Pages & Workers are merging gradually, and Static Asset Workers are effectively going to replace Pages in the long run, in short
Mare there any options to connect from workers into a private network? getting tired of seeing bots trying to brute force an API from my k8s cluster that i only access from workers
HUse a tunnel, put it on a subdomain, and then add a firewall that blocks all requests not from a Worker on your zone: https://developers.cloudflare.com/ruleset-engine/rules-language/fields/reference/cf.worker.upstream_zone/
Cloudflare Docs
Identifies whether a request comes from a worker or not.
Soh yea i wondered about that too.
i have 1 public api endpoint in my sveltekit app. is there an easy way to block access not from the same worker? just to reduce possible spam
(preferably in code, not a custom firewall rule)
i have 1 public api endpoint in my sveltekit app. is there an easy way to block access not from the same worker? just to reduce possible spam
(preferably in code, not a custom firewall rule)
Pbelieve you could use an Transform Rule to add the value of
cf.worker.upstream_zone
SHm i did look into that but i dont see any way of "selecting"
cf.worker.upstream_zonethe rule would be something like this
S'cf.worker.upstream_zone' is not a valid value for expression because the use of field cf.worker.upstream_zone is not allowed:bibicat:
Pah right, I believe there should already be an
CF-Workerwhich has the same value as the upstream_zone one
