Cloudflare Developers•3y ago

I mean no heavy compute or long requests

CCyb3r-Jak3 I mean no heavy compute or long requests

sathoro•5/24/23, 3:39 PM

we do several minute long requests it works fine actually

sathoro•5/24/23, 3:39 PM

the client just needs to stay connected lol

sathoro•5/24/23, 3:40 PM

using SSE

kian•5/24/23, 3:42 PM

What's your Worker code?

NNeo Is Cloudflare Workers good for making an API server?

dave•5/24/23, 3:48 PM

all of our public facing API services are on Cloudflare Workers, and even most of the private APIs are also on Workers.

KKartik ```app.get("*", async (ctx) => { const cacheurl = new URL(ctx.req.url); // c...

kian•5/24/23, 3:49 PM

headers

headers

is always going to be blank other than pix-apipix-api in that scenario.

kian•5/24/23, 3:50 PM

You're not using the request headers, you define

headers

headers

with let headers = new Headers();let headers = new Headers(); which is an empty object and then only contains the pix-apipix-api header which you set.

kian•5/24/23, 3:50 PM

If you want to use the request headers, that'd be ctx.req.headersctx.req.headers

tobi•5/24/23, 4:07 PM

with the upcoming workers snippets, in theory i should be able to make a "free" load balancer for an api; and the only real limit should be the dns resolution fees, is that correct (which only start at 1m/month)?

kian•5/24/23, 4:11 PM

Snippets can only make 1 subrequest so you wouldn't be able to do that.

kian•5/24/23, 4:11 PM

Well, you can do Math.random and select a subdomain based on that

kian•5/24/23, 4:11 PM

but you wouldn't be able to hit one and then decide to fallback if it failed

Neiki•5/24/23, 4:44 PM

can i somehow send me an email or sending with an webhook if that happens again? like let me notify

NNeiki can i somehow send me an email or sending with an webhook if that happens again?...

Neiki•5/24/23, 4:56 PM

i cant find anything to workers or crons in the notifications tab, only about pages

Kkian Well, you can do Math.random and select a subdomain based on that

tobi•5/24/23, 5:04 PM

yeah, fallback would not be a possibility. but i could do either random, or based on some "hash" of the request ip + maybe some user cookie or sth.

tobi•5/24/23, 5:07 PM

@kiannh then, if 1 request can be split to multiple subdomains, one could assign only functioning servers to those subdomains, i.e. if i know one of my servers is down, i change its subdomain to point to a known good server

tobi•5/24/23, 5:08 PM

one thing i had in mind was sending ppl to "local" servers; i.e. near their closest cf-POP. i hope this will be possible with snippets

Ttobi hi 😄 with the upcoming workers snippets, in theory i should be able to make a ...

dave•5/24/23, 5:28 PM

Is your API HTTP already?

tobi•5/24/23, 5:31 PM

yea

Ttobi yea

dave•5/24/23, 5:35 PM

you can already do multiple fetches in parallel and just race to the first one that completes.

tobi•5/24/23, 5:35 PM

hmm... doesn't seem like an elegant solution.

dave•5/24/23, 5:35 PM

sure it is

tobi•5/24/23, 5:36 PM

if i have lets say 5 servers then i get 5x the traffic with this method

tobi•5/24/23, 5:37 PM

yea, basically just "forwarding" the request somewhere, right?

Ttobi if i have lets say 5 servers then i get 5x the traffic with this method

dave•5/24/23, 5:37 PM

Sure, but you can cache the result too if it's problem.

tobi•5/24/23, 5:38 PM

depends on what one wants to accomplish.

dave•5/24/23, 5:38 PM

It's really nice since you can do "advanced" ish load balancing too if want.

e.g. can see the round-trip latency by doing:

curl -s "https://untrusted.files.ai.moda/api/v1.0.0-alpha/ping" | jq

curl -s "https://untrusted.files.ai.moda/api/v1.0.0-alpha/ping" | jq

tobi•5/24/23, 5:39 PM

if i, lets say want to build some kind of "low-latency" storage for small files, that is kinda guaranteed to have sub <50-60 ms latency, i could make that with less (or no) js on the user side, while still being able to serve the data quickly.

Ddave It's really nice since you can do "advanced" ish load balancing too if want. e...

tobi•5/24/23, 5:40 PM

what do you mean with that?

Ttobi if i, lets say want to build some kind of "low-latency" storage for small files,...

dave•5/24/23, 5:40 PM

https://untrusted.files.ai.moda/api/v1.0.0-alpha/url/generate_one_time

https://untrusted.files.ai.moda/api/v1.0.0-alpha/url/generate_one_time

^ we have a similar ish use-case, and made this to return the endpoint with the estimated lowest latency for the user.

tobi•5/24/23, 5:41 PM

ahhh

tobi•5/24/23, 5:41 PM

basically just a request to where the client should then upload to. so same thing accomplished, just not intransparent to the user

dave•5/24/23, 5:41 PM

sure it can be

tobi•5/24/23, 5:42 PM

uh, nearest datacenter hit is london?

Ttobi uh, nearest datacenter hit is london?

dave•5/24/23, 5:43 PM

that sounds right, since that uses an S3 backend with buckets in Frankfurt, Ireland, London, Montreal, San Jose, and Virginia.

tobi•5/24/23, 5:43 PM

well, then from austria it should have hit frankfurt

Ttobi basically just a request to where the client should then upload to. so same thin...

dave•5/24/23, 5:44 PM

if it's <100MB, you can just accept the request data body directly in the Worker and POST it to an endpoint elsewhere without needing to make the client wait.

Ttobi well, then from austria it should have hit frankfurt 😛 😉

dave•5/24/23, 5:45 PM

ah so this crappy S3 provider has terrible uptime, and their Frankfurt server has crazy reliability issues (hours of downtime per month). That's why this Worker is really nice, because it actually checks if the S3 endpoint is working or not.

dave•5/24/23, 5:46 PM

(off-topic but are only using this unreliable S3 provider temporarily mainly because of the lack of per-bucket API tokens for R2.)

I mean no heavy compute or long requests

Similar Threads

Similar Threads

Similar Threads