AbortError on Github Action

[cursive fractal]

I am getting an error after 5 minutes of execution of my Dagger CI pipeline. I am a bit confused about where this comes from and if I can configure it. I could not find anything in the documentation about timeouts. Surely it is possible to have pipelines running for more than 5 minutes?

AbortError: The operation timed out
Error: Encountered an unknown error while requesting data via graphql

https://github.com/demarches-simplifiees/demarches-simplifiees.fr/actions/runs/13704859634/job/38336924878

[tropic geyser]

@cursive fractal do you have a Dagger Cloud trace URL to share? it's easier for us to understand the error there

additionally cc @trail gulch could it be related to a bun sdk thing? I'm seeing that they're using bun here and I recall we had some typescript timeout issues in the past

not sure if/how they might affect bun

[cursive fractal]

Oh. Bun. Yeah https://github.com/oven-sh/bun/pull/6217

5 min is default bun fetch timeout

[trail gulch]

Yeeeeah... I think that's the error
However I added a fix that should also works for Bun 😮

https://github.com/dagger/dagger/pull/8576/files

I remember that I also extended timeout in all of our units tests and it seems we never encountered that problem anymore

[cursive fractal]

maybe related https://github.com/oven-sh/bun/issues/16682

[trail gulch]

Ohhh yeah, that could definitely be a problem

[cursive fractal]

I tried to switch runtime to node in dagger.json but it seems bun is still used. Is it because of the presence of bun.lockb ?

[cursive fractal]

I was thinking more about the fetch issue: if you're using node-fetch instead of native fetch, the bun issue should not affect dagger...

[trail gulch]

node-fetch maybe reuse fetch underneath?

[trail gulch]

The runtime must be configured in the package.json, not the dagger.json: https://docs.dagger.io/configuration/modules#alternative-runtimes

[cursive fractal]

I ended up switching to the node runtime, and it solved the problem

[tropic geyser]

I guess we could still open an issue and track this in Dagger? WDYT @trail gulch ?

[limber lion]

jep had the same issue and had to switch back to node. although bun seems to be faster

[trail gulch]

I tried with bun last week and it was doing fine, if you can create a repro in an issue then yeah for sure

[limber lion]

Should be pretty easy. I think you just need a function that sleeps for >5 minutes

but maybe i need to update the bun runtime

[limber lion]

@trail gulch

Error logs:

✘ .bunTimeoutRepro: String! 5m1s
AbortError: The operation timed out
Error: Encountered an unknown error while requesting data via graphql
! process "bun /src/.dagger/src/__dagger.entrypoint.ts" did not complete successfully: exit code: 1
Error: input: test.bunTimeoutRepro process "bun /src/.dagger/src/__dagger.entrypoint.ts" did not complete successfully: exit code: 1

  /**
   * Bun timeout
   */
  @func()
  async bunTimeoutRepro(): Promise<string> {
    return dag
      .container()
      .from("alpine:latest")
      .withExec(["sleep", "10m"])
      .stdout();
  }

created an issue: https://github.com/dagger/dagger/issues/10091

[trail gulch]

Is it only happening if you sleep for 10 minutes?
Edit: okay I looked at the issue.

I’ll see if there’s a way to fix it on our side

[trail gulch]

Thanks for the repro btw 🙂

[limber lion]

Anything that runs for >5 minutes

[trail gulch]

Okay, I'm wondering if the bundling (see: #typescript) may be helpful to fix that issue because it seems bun can run the node bundle, so maybe the timeout issue is fixed that way

I'll try

Ohhh yep, seems like using the bundle library fixes the problem when bundle in target=node

[trail gulch]

Even if bun doesn't fix that problem, we can workaround it with the bundler, so I'll keep you updated when this new capability is introduced 😄
I'm working on hit full throttle

[trail gulch]

PR open: https://github.com/dagger/dagger/pull/10094 😄

I think it will take couple of days to be merged but it's on the way 😄

[limber lion]

sweet!