# 사고 보고: 2월 20일 gcp 인프라 장애 (복구 완료) **URL:** \ **Category:** Kaia\ **Created:** [2월 21, 2026, 6:30오전 UTC](https://devforum.kaia.io/t/2-20-gcp/9089 "2026-02-21T06:30:12Z")\ **Posts on this page:** 4\ **Page:** 1
**Author:** ![erickchandra](https://yyz1.discourse-cdn.com/flex035/user_avatar/devforum.kaia.io/erickchandra/32/3722_2.png) [@erickchandra](https://devforum.kaia.io/u/erickchandra)\ **Post date:** [2월 21, 2026, 6:30오전 UTC](https://devforum.kaia.io/t/2-20-gcp/9089/1 "2026-02-21T06:30:12Z")
2월 20일 05:17 KST (2월 19일 20:17 UTC)에 Google Cloud Platform(GCP) 이슈로 인해 일부 인프라 및 서비스에서 일시적인 중단이 발생했습니다. 이번 문제는 메인넷(Mainnet)과 카이로스(Kairos)의 일부 노드 및 엔드포인트, 그리고 Kaia Safe, Kaiascan, FeeDelegator 서비스의 일부 기능에 영향을 주었습니다. 중요한 점은, 이번 사건은 **네트워크 장애가 아니었다는 것** 입니다. Kaia는 탈중앙화 네트워크로서, 사고 발생 동안에도 네트워크 자체는 정상적으로 운영되었습니다. 영향을 받은 모든 구성 요소는 2월 20일 09:12 KST (00:12 UTC)에 완전히 복구되었습니다. 자세한 내용은 아래 블로그 공지를 참고해 주세요: > **[Incident Report: GCP Infrastructure Disruption on Feb 20](https://blog.kaia.io/incident-report-gcp-infrastructure-disruption-on-feb-20/)** > > On February 20, at 05:17 KST (20:17 UTC, Feb 19), some of our infrastructure and services experienced an interruption due to an infrastructure issue on Google Cloud Platform (GCP). > > Impacted Services: > > \* Minor subset of Mainnet and Kairos... ---
**Author:** ![yui\_r](https://yyz1.discourse-cdn.com/flex035/user_avatar/devforum.kaia.io/yui_r/32/4177_2.png) [@yui\_r](https://devforum.kaia.io/u/yui_r)\ **Post date:** [4월 7, 2026, 4:37오전 UTC](https://devforum.kaia.io/t/2-20-gcp/9089/2 "2026-04-07T04:37:11Z")
This incident highlights the importance of decoupling critical infrastructure from centralized cloud providers to improve resiliency. Implementing a more robust multi-cloud or hybrid deployment strategy would effectively mitigate these single points of failure for services like FeeDelegator and Kaiascan in the future. ---
**Author:** ![yui\_r](https://yyz1.discourse-cdn.com/flex035/user_avatar/devforum.kaia.io/yui_r/32/4177_2.png) [@yui\_r](https://devforum.kaia.io/u/yui_r)\ **Post date:** [4월 7, 2026, 7:42오전 UTC](https://devforum.kaia.io/t/2-20-gcp/9089/3 "2026-04-07T07:42:43Z")
The resilience of the network during this GCP outage confirms the effectiveness of the decentralized architecture, even if the surrounding tooling experienced temporary downtime. It might be worth exploring a multi-region or multi-cloud failover strategy for the RPC endpoints to mitigate these single-provider dependency risks in the future. ---
**Author:** ![yui\_r](https://yyz1.discourse-cdn.com/flex035/user_avatar/devforum.kaia.io/yui_r/32/4177_2.png) [@yui\_r](https://devforum.kaia.io/u/yui_r)\ **Post date:** [4월 7, 2026, 7:55오전 UTC](https://devforum.kaia.io/t/2-20-gcp/9089/4 "2026-04-07T07:55:50Z")
This incident highlights the inherent risks of relying on centralized cloud providers for RPC and explorer infrastructure. It’s a strong reminder that we need to prioritize more robust multi-cloud or hybrid node deployments to ensure these critical access points remain resilient even when a single provider hits an availability snag.