Service Health

ISSUE SUMMARY

On Wednesday 27 June 2018, Google Cloud Console and Google Cloud Resource Manager API experienced a significant performance degradation for a duration of 51 minutes. Although, this did not affect user resources running on the Google Cloud Platform, we understand that our customers rely heavily on Google Cloud Console to manage their resources and we sincerely apologize to everyone who was affected by the incident.

DETAILED DESCRIPTION OF IMPACT

From 14:21 to 15:11 PDT, users were unable to manage their folders, projects and organizations using Google Cloud Console, Google Cloud Resource Manager API and gcloud command line. The following APIs were affected:

Google Cloud Console: Impacted users were unable to list their projects, search for projects, folders and organizations or view their bill. Search box failed to return the above too. Google Cloud Resource Manager API: Impacted users were unable to list their projects, folders and organizations BigQuery: Impacted users were unable to list their bigquery projects using the API.

ROOT CAUSE

The incident was triggered by a configuration change in the search infrastructure powering Cloud resource metadata search. The search infrastructure sends ACL checks to a central ACL server to make sure the end user has access to the Cloud resource metadata it plans to return. The configuration change introduced a new field in the ACL check request, while the central ACL server had not whitelisted the search infrastructure to send that field, causing an outage in Cloud resource metadata search.

REMEDIATION AND PREVENTION

At 12:26 PDT, Google Engineers rolled out the configuration change. Our automated release validation system rejected the change due to a high rate of errors. Around 14:16 PDT, an unrelated change was made to the same search infrastructure which triggered a bug that disabled its automated release validation system. This change also inadvertently picked up the prior configuration change and due to the lack of automated release validation, the change was successfully propagated to production. Within a span of few minutes, several engineering teams were automatically alerted to the situation and began the mitigation process.

The issue was fully mitigated at 15:11 PDT when the configuration change was rolled back.

We apologize again for the inconvenience caused. We are taking immediate steps to prevent recurrence and improve reliability in the future, including:

Fixing the bug that inadvertently disabled the canary analysis system. Improving process around pushing changes that involve several dependencies. Improving testing and staging alerts to catch issues of this nature before they reach production.

The issue impacting the ability to list projects and organizations has been resolved for all affected users as of 15:16 US/Pacific. We will conduct an internal investigation of this issue and make appropriate improvements to our systems to help prevent or minimize future recurrence. We will provide a more detailed analysis of this incident once we have completed our internal investigation.

We are experiencing an issue impacting the ability of customers to list projects and organzations beginning at Wednesday, 2018-06-27 14:34 US/Pacific. Current data indicate(s) that approximately 75% of all GCP customers are affected by this issue. For everyone who is affected, we apologize for the disruption. We will provide an update by Wednesday, 2018-06-27 16:13 US/Pacific with current details.

We are investigating an issue with Identity & Access Management. We will provide more information by Wednesday, 2018-06-27 15:00 US/Pacific.