V
Visiting Puffin
Problem
  • Calling the log download API immediately after a pipeline finishes can return a ZIP with missing/empty step logs. The first request generates and caches the ZIP even if not all logs have reached storage yet, so later downloads (API or UI) keep returning the same incomplete archive.
Impact
  • Users receive incomplete log bundles and must retry.
  • Confusion and wasted time when UI/API keep serving the cached partial ZIP.
  • Automation hits timeouts after multiple retries.
Suggested enhancement
  • Defer ZIP generation until all step logs are available, or
  • Regenerate the ZIP automatically if a prior archive was incomplete, and/or
  • Provide an API readiness flag (e.g., status: "logs_pending") so clients know when it’s safe to download.
Workaround (today)
  • Wait/poll before downloading, but this is fragile and adds latency.
Created by Pedro Mastelaro
·