Intermittent Outage - EU
» View Event Details | Created
Post-Mortem
Summary
A background worker exhausted memory and stopped, which paused some tasks. Automatic scaling then reduced server numbers and falsely judged the last server healthy, leaving no capacity. Users saw brief outages.
Actions taken
• Restarted affected worker and restored normal server capacity.
• Changed memory rules so the search service is terminated before background tasks, avoiding repeat overloads.
• Tightened health-check settings in automatic scaling to replace unhealthy servers immediately.
Posted: