Skip to content
CasesNew Learn 30 Career · co-op History Leaderboard Log in
EN
English✓ Русский

Access required

CRITICAL Weekly on-call challenge Case 1 · 15 min

API exhausted PostgreSQL connections

Long connections overload PostgreSQL; fixing the timeout exposes a retry storm.

What the on-call engineer sees

  • API requests first slow down, then start failing
  • netstat shows many connections in SYN_SENT
  • PostgreSQL connections, CPU, and RAM usage grow together
  • The API logs errors in /var/log/api.log, and free space decreases
  • After two minutes of simulation, pg_lsclusters will show PostgreSQL down
  • Reducing the timeout restores PostgreSQL, but immediate API retries overload it again

You need an account to take this incident. Registration is free and takes a minute: just a username and password, with no email or phone number. The case you completed as a guest will carry over.

Sign up Log in

Without an account, you can complete case 0, “Store unavailable” —a short introduction in a real console.

Prepare for this caseLearn ↗

Explore the relevant skills before practice. Lessons open in a new tab.

  • PostgreSQL: startup, configuration and connections12 min ↗
  • Timeouts and retries: avoid amplifying an outage14 min ↗
  • Final incident: diagnosis, recovery and handoff20 min ↗
All preparation for this case →
Site terms · Privacy policy · Cookie settings · privacy.prodisdown@yandex.ru · Find us on VK · Find us on TikTok

We use cookies and Yandex Metrica to analyze visits. Learning analytics for signed-in users are linked to their account. By selecting “Accept”, you agree to the terms of data processing.