Skip to content

TROUBLESHOOTING.md

Latest commit

 

History

History
54 lines (36 loc) · 1.61 KB

TROUBLESHOOTING.md

File metadata and controls

54 lines (36 loc) · 1.61 KB

Troubleshooting

Another run appears active

Message: error: Another run appears active (PID ...)

Cause: The state file is locked by another cgr apply process.

Fix:

  • Wait for the active run to finish.
  • If you intentionally need isolated concurrent runs, use cgr apply FILE --run-id ID.
  • If the previous process crashed and left a stale lock, run cgr state reset FILE.

State from one apply run is affecting another

Cause: By default, applies for the same graph share one state journal.

Fix:

  • Use cgr apply FILE --run-id canary.
  • Or pin the exact journal path with cgr apply FILE --state /path/to/custom.state.

A race branch kept running after another branch succeeded

Current behavior: Race branches are terminated when another branch wins. If the losing branch was already inside a remote SSH command, local cleanup happens immediately but the remote host may still need to reap its own process depending on SSH/curl behavior.

Fix:

  • Prefer idempotent race branches.
  • Use race into DEST: when branches compete to write the same artifact.

Inline config content contains #

Current behavior: Literal # characters are preserved inside inline content > PATH: and block in FILE, marker "X": bodies.

If output still looks wrong:

  • Check shell commands separately; comment stripping rules for shell lines are different from inline content bodies.

Apply needs a different timeout for one slow step

Fix: Set a step-local timeout:

[backup]:
  run $ pg_dump ...
  timeout 10m

The step timeout overrides the default per-step execution timeout without changing the global cgr apply --timeout.