mirror of
https://github.com/OpenRouterTeam/spawn.git
synced 2026-05-22 03:14:57 +00:00
The _run_with_restart wrapper in all 8 DigitalOcean agent scripts catches SIGTERM/SIGKILL exit codes (143/137) and retries the orchestration process. In headless mode (E2E tests), when the provision timeout kills the process, this restart loop would re-run main.ts, creating duplicate droplets and exhausting the account's droplet quota — causing ALL subsequent DO agents to fail provisioning. Skip the restart loop entirely when SPAWN_HEADLESS=1 (set by runScriptHeadless in the CLI). The restart behavior is only useful for interactive sessions where the user's SSH connection drops. Fixes #2794 Agent: code-health Co-authored-by: B <6723574+louisgv@users.noreply.github.com> Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com> |
||
|---|---|---|
| .. | ||
| aws | ||
| cli | ||
| digitalocean | ||
| docker | ||
| e2e | ||
| gcp | ||
| hetzner | ||
| local | ||
| shared | ||
| sprite | ||
| test | ||