OCSP responder unavailable

On-call incidentHard35 min

Your task

Your TLS certs are valid for months, but at 09:30 a slice of strict clients (some banks' API gateways, hardened browsers) start failing handshakes to your endpoint with 'OCSP response expired' / 'revocation status unavailable', while most clients connect fine. Dashboards: your origin and certs are healthy; the failing clients are the ones that hard-require a fresh stapled OCSP response. Your TLS terminator staples OCSP responses it fetches from the CA's OCSP responder and caches them. The CA's OCSP responder has been returning errors / timing out since ~08:45. How do you triage and mitigate?

It covers

  • Ocsp
  • Stapling
  • Revocation responder