Impact
On 16 January 2026, customers experienced intermittent and occasionally failed search responses on the Searches API. The issue was first identified at 13:05 GMT and occurred in short bursts through the afternoon, with the service stabilising after fixes were applied.
Cause & Resolution
A sustained increase in traffic caused our task processing system to reach a threshold where an existing configuration issue affected how search-related processes were queued and handled. In simple terms, this system acts like a dispatch office, receiving tasks and handing them to available workers. Under higher load, the dispatch rules were not balancing work as effectively as expected, which led to intermittent search failures.
We investigated from 13:05 GMT, applied fixes to restore search results, and continued monitoring. A fix was deployed at 15:50 GMT to reconfigure how tasks were delegated, which improved stability. We observed a brief recurrence shortly afterwards, carried out additional tuning, and continued monitoring until the platform returned to normal operation.
Prevention Measures
To reduce the risk of a repeat:
We have improved the task delegation configuration so the task processing system behaves predictably during higher traffic volumes.