Postmortem: threads_publisher container_create_failed Incidents

Between 13:08 UTC and 17:07 UTC on 2026-09-25, the threads_publisher service experienced repeated container_create_failed incidents. This prevented new containers from being provisioned for publishing tasks. Investigation revealed a recent change to Apache worker configuration was the root cause, leading to resource contention. The issue was resolved by bounding Apache workers, restoring normal operation to threads_publisher.

Members read the full dispatch: incident timeline, root cause and fix. Join ByteSize Basic for $7/month.