Docker backend backoff: maximum elapsed time exceeded

[BeckYeh]

Component

agent

Describe the bug

after upgrade from 3.13.0 to 3.16.0
some pipeline like terraform plan or phpunit test will failed with blow:

{"level":"error","workflow_id":"8269","repo":"some repo","pipeline_number":"281","error":"error during connect: Post \"http://%2Fvar%2Frun%2Fdocker.sock/v1.44/containers/wp_01KWDTMTX3AK8DAZ3BWSS1KEYK/wait\": backoff: maximum elapsed time exceeded (last error: rpc error: code = Unavailable desc = unavailable)","step":"postgres-woodpecker","time":"2026-07-01T03:23:03Z","message":"failed to trace detached step result"}

{"level":"error","workflow_id":"8269","repo":"some-repo","pipeline_number":"281","error":"error during connect: Post \"http://%2Fvar%2Frun%2Fdocker.sock/v1.44/containers/wp_01KWDTMTX45WTVPCXJQ71ZM2VT/wait\": backoff: maximum elapsed time exceeded (last error: rpc error: code = Unavailable desc = unavailable)","step":"redis","time":"2026-07-01T03:23:03Z","message":"failed to trace detached step result"}

{"level":"error","error":"backoff: maximum elapsed time exceeded (last error: rpc error: code = Unavailable desc = unavailable)","time":"2026-07-01T03:23:53Z","message":"grpc error: next(): code: Unavailable"}

{"level":"error","error":"backoff: maximum elapsed time exceeded (last error: rpc error: code = Unavailable desc = unavailable)","time":"2026-07-01T03:23:53Z","message":"runner error, retrying..."}

Steps to reproduce

1. upgrade amazonlinux 2023 with newest docker version and woodpecker 3.16.0
2. run origin terraform or phpunit pipeline

Expected behavior

still work

System Info

amaoznelinux 2023 with woodpecker 3.16.0

Additional context

No response

Validations

• Read the docs.
• Check that there isn't already an issue that reports the same bug to avoid creating a duplicate.
• Checked that the bug isn't fixed in the next version already [https://woodpecker-ci.org/versions]

[hiSandog]

The fix would be easier to review if it came with a narrow test for Docker backend backoff: maximum elapsed time exceeded. I would assert both the current failure mode and the intended successful behavior, especially around the reported case.