System status
A queue worker in our notifications pipeline lagged behind during a regional traffic spike. Notifications continued to deliver but with a 6–9 minute delay for ~12% of accounts. Worker scaling adjusted; backlog cleared by 14:54.
Database read replicas re-elected after a routine maintenance window. Dashboard pages took 3–6 seconds to load for ~7 minutes. No data loss; sessions remained signed in.
Routine Postgres minor-version upgrade. The service entered read-only mode during the cutover. Notice was sent to account owners 7 days in advance.