Every step green, queue still growing: 7 cross-posts, 72 hours, and the step that was a browser tab
Our cross-posting pipeline has five steps. Four of them are scheduled jobs, and all four have been green for a week. The fifth is a person pasting into a browser. This morning the queue was seven items deep, the oldest 72 hours old, and not one thing in the system had failed. The owner's question was reasonable and, in a way we did not expect, instructive: "the cross-posts look stopped — is the…
The cross-posting pipeline has five steps: publishing to dev.to, building a mirror copy, notifying, pasting into a browser tab, and recording. Four steps are scheduled jobs, while the fifth requires a human to paste text into a browser. The queue grew to seven items, with the oldest being 72 hours old, and no failures were reported.
The owner's reasonable question led to the realization that the step requiring human intervention was the issue. The platform moved its GraphQL API to a paid tier, and the owner refused to pay, resulting in a manual step being automated. The monitoring system detected the issue but couldn't distinguish between a temporary lag and a permanent failure.
The root cause was the lack of a clear distinction between the two types of alarms. The team automated around the manual step, which removed a failure mode and reduced minutes per article. However, the value of the step 4 (human pasting) step decays if the step 5 (recording) never runs. The team noticed this after writing the report and re-verified that the platform now provides an official agent skill for its API, which still holds true.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
