Tocsin Payments platform · Incidents · INC-2291 You are on call
SEV 2

INC-2291 · checkout-api · eu-west · opened —

checkout-api is slow in eu-west.

Not acknowledged. Level 2 is paged in —.

  1. Triggered
  2. Acknowledged
  3. Under the line
  4. Resolved
Open for 07:36 4.1% of checkouts over 1 s

Instruments

Why it is slow, what it has cost, and where.

Suspected cause

Correlated

Deploy 4f2a91c reached eu-west at —, two minutes before p95 crossed. It doubled the queries per checkout; the orders-db pool went from 52% to 94%.

Error budget · checkout

30 days
Objective
p95 under 300 ms, 99.9%
Left this month
11 min 40 s
Under 30%
Deploys frozen

checkout-api · every region

Live

People

Who holds the pager, and who is on this incident. Press a card to page them.

Rota board · Payments platform

12 h shifts

On this incident

1

    Log

    What has happened, newest first, and the rest of the checkout path.

    Timeline

    —

      Service board · the checkout path

      3 of 10 need attention
      Choose a header to sort · readings every 2 s

      History

      Twelve weeks of incidents on this platform, by the hour they fired.

      Incidents—
      Out of hours—
      Median to acknowledge4min 12s
      Median to resolve41min
      Acknowledge
      Escalate to the next level
      Page a responder
      Roll back checkout-apito 7c1e04b
      Mark resolved
      Instruments
      Rota board
      Timeline and services
      Incident history
      ↑↓ move↵ runEsc closeTocsin

      Escalate · INC-2291

      Page level 2 now?

      Ilse Brandt and Omar Haddad, the platform leads, get a phone call now. You stay primary.

      Page · INC-2291

      Who do you need?

      They are paged until they acknowledge, and join the people on this incident.

      How

      Resolve · INC-2291

      Mark INC-2291 resolved?

      Paging stops for everyone on it, and Tocsin opens a postmortem draft with this timeline.