Incidents log

Public incidents and resolutions for upivia.com. Add newest first.

2026-05-26 — Postgres out of disk on swarm node

• Detected: 03:40 UTC, every request returning 500 with PrismaClientInitializationError.
• Root cause: CapRover was retaining all old image builds (~150 images on /dev/sda1) until the partition hit 100%; Postgres couldn't write its lock file.
• Resolution: docker image prune -af reclaimed 108.6 GB. Postgres container came back up automatically.
• Duration: ~25 minutes.
• Follow-up: weekly housekeeping cron (in repo as scripts/server-housekeep.sh).