the finally block would never run

it does in fact execute if it's synchronous... is that the difference?

that's not true, though.

finally always runs.

Oh🤔

I admittedly haven't used it, assumed it worked similarly to a catch

But yeah in the workers runtime, I'm pretty sure it'd end as soon as the response returns unless you use waituntil

Thanks, though how would I go about setting up KV namespaces and Ids that the app depends on?

I don't know of anyone receiving this yet

but again, this is not true.

it'll log after returning a response.

Have you done asynchronous work with it?

not yet, i'm wondering if that's the difference.

Because waitUntil is specifically for async work - the types even only say it takes a promise

From the dashboard

they can create the kvs and all from the dashboard and then use those credentials in the config file

export default {
  async fetch(request, env) {
    try {
      console.log('try')
      return new Response("Hello world") 
    }

    finally {
      await new Promise(x => setTimeout(x, 1000))
      console.log('async finally')
    }
  }
}

this still logs in the finally block.

That's just going to block your Response

Set it to 5 seconds and you're not getting

Hello world

until 5 seconds passes

waitUntil

is fully non-blocking

oh interesting.

that answers my question.

so from a billing perspective they're the same, but the

finally

version blocks the response.

thanks, @kian!

Yeah, duration is duration so you'd see no difference in billing

All

waitUntil

does is let you return a Response and gives you 30 seconds to have the passed Promises resolve

When we get a surprise bill, how can we track down which Worker caused the damage?

Spot checking each Worker individually is both tedious and ultimately fruitless (as we can only see the prev 7 days in the dash)... would love the equivalent of DaisyDisk to explore/dig deeper, or even just a flat table of read/write requests/mo for each active Worker.

Copying the GraphQL calls that the dashboard makes & just changing the date range lets you go beyond 7 days

As to why the graphs only go back 7 days if the data is retained for longer,