Oh wait… it probably expects a global. What happens if you initialize it as it asks, in the construc
Oh wait… it probably expects a global. What happens if you initialize it as it asks, in the constructor?
Lbtw, this constructor would otherwise read the same env var if I didn't pass this:
HAfaik, process.env is only defined within Next
HAt least on Workers
Lwhat do you mean by that?
Lthis project uses next.js
HWorkers don’t have global env variables. Next gets around this somehow, but afaik that only applies within the Next scope
HSo a 3rd-party package wouldn’t be able to read any variables
Lwhat's the Next scope?
HThe code you write with Next
Lbut the way I'm reading from process.env and passing it to the constructor should work, right?
HYes. Just to be sure, check whether they are defined, by console-logging them
HOh wait
HDo deployments actually show logs other than errors…
HActually, when are you calling that?
HIn a server route? Globally?
LI define
const apiKeyThat’s probably the issue then. Environment Variables aren’t populated on startup
LHSo you can’t start the API client globally
Loh. so when are they populated?
HOnce the Worker/Function is actually invoked
Lso I can't cache the return value of the OpenAI constructor between invocations?
L(in a global var or otherwise)
HYou could, but you can’t actually construct it until you have received a request
Lok, so what would you recommend for doing that? some kind of nullable memoization type
HJust have a variable that can be an OpenAI API client, but by default is null. Then, whenever you need it, use the variable if present, otherwise initialize it first
Lah ok, thanks :)
�@Helpflare cloudflare keeps on saying failure
Fhi, I code a worker and it can solve POST request but I get 403
FWhy the respond is 403?
F
FAnd 421
FBut my program haven't return 403 or 421
IThose kind of error pages usually appear if your browser is making a request that isn't valid according to the HTTP specification.
IFor example, if you send a body in a GET request then you might get that. Or if you send a header name that isn't valid. Or any number of other things really.
ITake a look and make sure that the POST request you're making is valid and looks fine syntactically
SThere is any good project example for workers and D1 to start from ? Like with multiples APIs / files
DQuestion, for those of you who have had their Workers DDoS'd, how was the experience? Does CF do a decent job of blocking requests before they hit the Workers?
Dnice. I was debating if I should use Google Cloud Run without Workers, but I think it makes sense to always use Workers in front.
Dit does increased the cost from $0.20/million requests to $0.50/million requests when stacking a Worker on top, but the benefit of DDoS protection makes it nice.
ZYou can even block by header value on, so if they did have a valid authentication token somehow they would still get blocked.
MIs middleware only available for Cloudflare Pages and not Workers?
JMiddleware is a thing for Functions since they have file-based routing, and the concept of Middleware makes a lot of sense there for handling auth, etc.
JMany popular routers and frameworks for Workers also support Middleware though, such as https://hono.dev/
FMy cron worker job got killed for exceeding time (32,000 ms). Is there a way to increase that number?
FThe docs, under CPU time say
When the schedule interval is more than 1 hour, a scheduled Worker may run for up to 15 minutes.
S
wrangler 3.17.1-------------------
⎔ Starting local server...
✘ [ERROR] service core:user:search: Uncaught TypeError: Incorrect type for map entry 'DEFAULT_PAGE_SIZE': the provided value is not of type 'function or ExportedHandler'.
Is there a way to get a more detailed error here? I removed default_page_size from my code entirely and I'm still getting the error for some reason
Sodd, no error when I turn off local mode
Stried deleting node modules and .wrangler and reinstalling with yarn, no luck there either, something gotta be cached somewhere