[bug] Request failures in v4 on ECONNRESET and 403 Forbidden.

[judson-firebolt]

What happened?

Sometimes actions/upload-artifact@v4 gives errors like this one:

Wed, 10 Apr 2024 18:38:00 GMT
Run actions/upload-artifact@v4
Wed, 10 Apr 2024 18:38:01 GMT With the provided path, there will be 1 file uploaded
Wed, 10 Apr 2024 18:38:01 GMT Artifact name is valid!
Wed, 10 Apr 2024 18:38:01 GMT Root directory input is valid!
Wed, 10 Apr 2024 18:38:01 GMT Beginning upload of artifact content to blob storage
Wed, 10 Apr 2024 18:38:01 GMT Uploaded bytes 360403
Wed, 10 Apr 2024 18:38:01 GMT Finished uploading artifact content to blob storage!
Wed, 10 Apr 2024 18:38:01 GMT SHA256 hash of uploaded artifact zip is <redacted>
Wed, 10 Apr 2024 18:38:01 GMT Finalizing artifact upload
Wed, 10 Apr 2024 18:39:53 GMT Error: Failed to FinalizeArtifact: Unable to make request: ECONNRESET

It feels like this is happening more since upgrading from v3.

Is the backend for v4 less stable than the one for v3?

Also, at approximately the same time, I got one of Failed request: (403) Forbidden: Error from intermediary with HTTP status code 403 "Forbidden" :

Wed, 10 Apr 2024 18:45:52 GMT Run actions/upload-artifact@v4
Wed, 10 Apr 2024 18:45:52 GMT With the provided path, there will be 1 file uploaded
Wed, 10 Apr 2024 18:45:52 GMT Artifact name is valid!
Wed, 10 Apr 2024 18:45:52 GMT Root directory input is valid!
Wed, 10 Apr 2024 18:45:52 GMT Beginning upload of artifact content to blob storage
Wed, 10 Apr 2024 18:45:53 GMT Uploaded bytes 8388608
Wed, 10 Apr 2024 18:45:53 GMT Uploaded bytes 10016359
Wed, 10 Apr 2024 18:45:53 GMT Finished uploading artifact content to blob storage!
Wed, 10 Apr 2024 18:45:53 GMT SHA256 hash of uploaded artifact zip is <redacted>
Wed, 10 Apr 2024 18:45:53 GMT Finalizing artifact upload
Wed, 10 Apr 2024 18:45:53 GMT Error: Failed to FinalizeArtifact: Received non-retryable error: Failed request: (403) Forbidden: Error from intermediary with HTTP status code 403 "Forbidden"

Which is strange. Supposedly we don't have any outbound restrictions on our runner environment so I don't think this was on our side.

What did you expect to happen?

The upload should have succeeded

How can we reproduce it?

Try uploads throughout the day? It's a flaky problem.

Anything else we need to know?

No response

What version of the action are you using?

v4

What are your runner environments?

self-hosted

Are you on GitHub Enterprise Server? If so, what version?

No response

[milesgranger]

Can second this, have a workflow w/ ~100 jobs and it seems to fail when two or more try uploading around the same time, but that could be in my head. However the workflow never completes successfully without needing to retry failed jobs solely due to this action failing intermittently.

[aarr0n]

Also experiencing this issue with v4. We have a workflow that runs multiple jobs which use this action and intermittently one or more of these fails because of the Failed to FinalizeArtifact: Unable to make request error.

[carlcsaposs-canonical]

+1

Example failure (on GitHub-hosted ubuntu-latest runner): https://github.com/canonical/pgbouncer-k8s-operator/actions/runs/9557337019/job/26344437669

Raw logs: job-logs.txt

[carlcsaposs-canonical]

Potentially related: https://github.com/actions/upload-artifact/issues/569 https://github.com/actions/download-artifact/issues/338

[cgturner1]

I am also seeing this issue on my self hosted ubuntu runners (sporadically) that then pass on re-run no issues after upgrading to v4