Resolved
All error metrics are back to the baseline, closing this incident.
Monitoring
The hotfix was successful, and the resource overload is resorbed. We are monitoring the situation but expect all systems should be functioning perfectly normally from now on.
Identified
An internal system on the critical path of the handling of most requests to Nabla is operating above its desired load, leading to an increased error rate. The number of errors overall remains very small, so in all likelihood the user impact, if any, is minimal at this point.
We're rolling out a hotfix to resolve the root cause of the resource overload, which should bring things back to normal. ETA: 30' to 1h.