Skip to content

run-workflow: long runs, run records, and trace reading - #5

Merged
mzxchandra merged 1 commit into
mainfrom
run-workflow-long-runs
Sep 15, 2026
Merged

mzxchandra merged 1 commit into
mainfrom
run-workflow-long-runs

Conversation

@mzxchandra

Copy link
Copy Markdown
Contributor

Manual runs hold the request connection for the entire execution, so on hosted deployments any run longer than the fronting load-balancer timeout (roughly a minute, observed) fails as a transport error or is cancelled when the connection drops - and --manual refuses --async, so there is no synchronous workaround. This PR documents the deploy-and-run-async path for long runs, the run-record lifecycle (logs get is NOT_FOUND for in-flight and cancelled runs, and completed runs pass through a redacting status), the deploy-then-run race where an immediately fired run can still execute the previous deployment, nested child-workflow spans in run traces, and the credits-vs-dollars cost unit split between the in-workflow logs block and the CLI.

🤖 Generated with Claude Code

https://claude.ai/code/session_01MYsyypbYR5eT4rxM2vaeEF

@mzxchandra
mzxchandra merged commit b2563e4 into main Sep 15, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant