Incident Library
Applied filters OrganisationHello Inc.Clear allEvery case under the filters, with no sentence; newest first.
Suggested filters Search, and the values found in your words appear here.

Graph

  • Root cause
  • Component
  • Symptom
  • Contributing factor
  • Case

Cases1 case

1 to 1 of 1

  1. Symptoms

    A bug report said that some reservation notifications in the AutoReserve restaurant reservation service never arrived. Of the several kinds of notification sent as a reservation progresses, only one kind sometimes arrived and sometimes did not.

    Root cause
    Component
    Contributing factor
    Detection

    A bug report from AutoReserve said that some reservation notifications were not arriving.

    Remediation
    • Scaled the Memorystore for Redis instance from 8 GiB to 16 GiB, which did not stop the write rejections
    • Upgraded Redis from 3.2 to 7.2, which removed the fragmentation and stopped the write rejections
    • Moved the notification sent status from Redis to the relational database
    Action items
    • Consider enabling active defrag if physical memory use rises above 1.3 times logical memory use
    Lessons

    The logs and the instance page metrics did not show the cause; it took looking inside Redis with INFO memory to see physical memory far above logical memory. A cache is a place for values that can be rebuilt if lost and whose writes may fail, so the sent status, which had to persist and be decided once under concurrent processing, belonged in the relational database.

Check your cluster for this

This case involves In-memory cache. kubegap audits a Kubernetes cluster for gaps like these, read-only, from Claude Code or Codex; its checks cite public reports like the ones in this library.

By root cause1