On November 4, 2026, between approximately 03:40 UTC and 10:01 UTC, some Copilot agent users experienced errors listing and creating agent tasks. On average, 1.22% of requests were failed, with a peak of 2.68% over a ten-minute interval.
The issue occurred when increased task-list traffic and expensive queries exhausted shared database capacity. We mitigated the broader impact by disabling a client experiment and temporarily restricting task-list requests from the affected client. Task-list requests remained restricted after the wider service recovered; we removed the blanket restriction on November 6 and continued a gradual rollout.
We’ve since improved task-list queries, moved rate limiting to separate infrastructure and added capacity monitoring. We’re continuing to reduce unnecessary client requests and validate capacity before expanding the rollout to reduce the chance of recurrence.
Posted Nov 04, 2026 - 10:01 UTC
Monitoring
The degradation has been mitigated. We are monitoring to ensure stability.
Posted Nov 04, 2026 - 10:01 UTC
Update
We're continuing to see recovery as rollout continues
Posted Nov 04, 2026 - 09:50 UTC
Update
We've identified the issue with a specific client version and are deploying a mitigation. We are already seeing partial recovery.
Posted Nov 04, 2026 - 09:11 UTC
Update
Copilot Mission Control is experiencing increased error rates (approx 2.5%) in listing and creating tasks due to upstream database throttling. We are investigating two potential mitigations.
Posted Nov 04, 2026 - 08:47 UTC
Investigating
We are investigating reports of impacted performance for some GitHub services.