From 11:40am to 12:20pm PT, the us001 database was under high load due to data processing jobs.
During this time, the p95 latency for API requests were elevated, peaking at 1.6seconds. We did not see any API errors during this time period. There were around eleven total 500s due to timeouts from our web UI in the same time period.
We are currently making changes to our data processing jobs to tackle the root cause.