Root cause: An internal component responsible for retrieving content from client feeds stopped processing on October 10 and was not automatically restarted. This reduced the amount of content reaching the publishing system, and automated posts to connected social networks were delayed or not published. Posts scheduled directly by users were not affected.
Resolution: The component was restarted on October 12 at 07:20 UTC, and publishing returned to normal. Content that had been queued was published shortly afterwards.
Next steps: We are improving our monitoring and automated recovery so that this type of issue is detected and resolved faster. We apologize for the disruption.