@Skye know of any weirdness with D1's shim taking a ton of CPU time? The operations it does isn't complex. Perhaps the JSON parse from the SQL response?

(God I can't wait to get rid of this shim)

I mean the only high cpu thing would be parsing a big json blob

Yeah that's what I'm thinking... I hate that this manifests for users due to the shim

I mean, wouldn't that happen for native bindings too?

I don't think we count binding compute time

(Also ideally we don't use JSON for the real binding)

I'd love to see how that works when I have a gchat

Don't think you want to read C++

It's not easy to parse

true, ew

almost as unreadable as rust

Sadly it's far far worse

In short, I can't do anything to bring the CPU time down on my side?

Hey, I'm trying to set up postgres, but I'm just accessing a public db so I don't want to set up tunneling. Is it possible to do this? I tried following this (https://developers.cloudflare.com/workers/tutorials/query-postgres-from-workers-using-database-connectors/), but when I fetch my API and it tries to connect to postgres I receive:

(log) Fetch API cannot load: mainnet.db.explorer.indexer.near.dev

my client connection looks like this:

ts
      const client = new Client({
        user: 'public_readonly',
        password: 'nearprotocol',
        database: 'mainnet_explorer',
        hostname: 'mainnet.db.explorer.indexer.near.dev',
        port: '5432',
      })

The database connector connects to a WebSocket, which would be provided by the Tunnel. If you are connecting to a public DB that doesn't provide a WebSocket API, then it will not work.

Are there any other libraries I could use instead, so that I don't need the tunnel?

Will go to sleep now so checking back later

No. At the moment, unless you have access to the TCP Worker beta, then you cannot connect to anything that doesn't support HTTP(S) or WebSockets, whether that be a Database, a Live Stream, or something else.

Note too that directly opening a connection to your DB in a Worker may not be the best approach anyway, as most databases are not optimized for extremely short-lived connections running only 1-2 queries.

Hi, I am very confused, I have made a websocket server which allows connections from any IPv4, and I am trying to use a worker as a proxy for a client. I can successfully to the server from a local client, but when connecting through the proxy I am getting a 403 Error

This is the response that the worker gets when trying to connect to the server:

Response { body: Stream(ReadableStream { obj: Object { obj: JsValue(Object({"locked":false})) } }), headers: Headers {
cache-control = private, max-age=0, no-store, no-cache, must-revalidate, post-check=0, pre-check=0
cf-connection-close = close
cf-ray = 7b6e9f69a2e34141-LHR
connection = keep-alive
content-length = 16
content-type = text/plain; charset=UTF-8
date = Wed, 12 Apr 2023 21:39:34 GMT
expires = Thu, 01 Jan 1970 00:00:01 GMT
referrer-policy = same-origin
server = cloudflare
x-frame-options = SAMEORIGIN
}
, status_code: 403, websocket: None }

The weird thing is that this proxy thing is working well when connecting to an external server, but not to the barebone server I have just created for testing, which suggests to tweak something in the server side?

Does anyone know if it is possible to run workers on only one region?

Not without being enterprise. You can just make a firewall that blocks access to the worker apart from certain areas depending on your needs

Thank you so much for your quick response, that’s what I though, sad to hear.

If I need to cache posts a user loads from the db should I be using workers kv for this?

Does anyone host a discord bot on pages/functions? If so could you tell me do you use a scheduledfetchevent to update your commands and intent or is there a better way to do it

You should probably do it during the deployment process