GraphQL API请求事件

2026年7月26日21:34 UTC，我们开始观察到GitHub GraphQL API出现间歇性错误。部分GraphQL API请求在短时间内返回HTTP 502错误。在影响期间，受影响区域内平均有0.09%的GraphQL API请求失败，最严重时在7月27日03:02 UTC的两分钟内，失败请求峰值达到0.50%。失败的请求在重试后通常能成功，没有数据丢失或更改。其他GitHub服务未受影响。<br /><br />这些错误追溯到处理部分GraphQL API流量的一组服务器。该组上的应用程序进程在完成响应前间歇性关闭连接。影响于7月27日03:52 UTC结束，当时这些进程已被替换，我们在确认错误率恢复正常后，于7月27日04:09 UTC解决了该事件。<br /><br />我们仍在调查这些进程为何关闭连接，这项工作由拥有底层计算平台的团队负责进行。与此同时，我们正在增加检测和自动缓解措施，以应对单个服务器组行为与其同类不同的情况。

原文On July 26, 2026 at 21:34 UTC we began seeing intermittent errors on the GitHub GraphQL API. A subset of GraphQL API requests returned HTTP 502 errors in short bursts. During the impact window an average of 0.09% of GraphQL API requests in the affected region failed, with a peak of 0.50% of requests failing during the worst two-minute period at 03:02 UTC on July 27. Requests that failed generally succeeded when retried, and no data was lost or altered. Other GitHub services were not affected.<br /><br />The errors were traced to a single group of servers handling a share of GraphQL API traffic. Application processes on that group intermittently closed connections before completing responses. Impact ended at 03:52 UTC on July 27 when those processes were replaced, and we resolved the incident at 04:09 UTC on July 27 after confirming error rates had returned to normal.<br /><br />We are still investigating why those processes closed connections, and that work is being carried out by the team that owns the underlying compute platform. In the meantime we are adding detection and automated mitigation for when a single group of servers behaves differently from its peers.

影响API请求的服务降级问题已得到缓解。我们正在持续监控以确保稳定性。

原文The degradation affecting API Requests has been mitigated. We are monitoring to ensure stability.

我们正在调查有关API请求性能下降的报告。

原文We are investigating reports of degraded performance for API Requests