Yandex says a drone attack damaged its data center in Vladimir on October 11, forcing the facility to stop operating. Its Yandex Cloud incident log reports that some cloud zones became unavailable and that the platform was still operating with limited functionality at 15:01 UTC. This is a third reported facility incident in four days, following the Sasovo shutdown and partial Kaluga outage.

The incident log gives customers more detail than the earlier facility notices: it separates the shutdown of one building from the changing status of cloud services. It does not establish the condition of individual servers, customer data or Yandex's AI training systems.

The big change

  • What changed: A third Yandex data center has stopped operating, and Yandex now describes the remaining cloud configuration as unstable. Its updates show some services available while others are impaired.
  • Why it matters: A customer deciding where to restore a workload needs service-level information, not just a facility status. Yandex told customers to use alternative recovery plans and later recommended considering other platforms for faster deployment and redundancy.
  • What to watch: The next Yandex Cloud update is due October 12 at 11:00 Moscow time. It should clarify which functions have returned and whether Yandex still advises customers to use external capacity. The incident log gives no restoration date for the Vladimir facility.

What Yandex reported on October 11

Yandex Cloud first reported that some zones were unavailable at 01:11 UTC. At 03:08 UTC it said a UAV attack had damaged infrastructure at its Vladimir data center and that operations there were completely suspended. The company reported no casualties and said emergency services were on site. It described the platform as operating in emergency mode, called its remaining resource configuration unstable and recommended alternative disaster recovery plans. These are Yandex's accounts of the incident and its service status; the status page does not identify who launched the drone.

The service picture changed during the day. At 03:11 UTC, Yandex said its cloud console and API were available, compute was partially available, and most high-availability managed database clusters had switched over, although some clusters could be read-only or unavailable. Object storage, labeled S3 in the update, was available for reading, with possible writing problems. The company was still assessing regional Kubernetes clusters and warned that support communications could be unstable.

By 08:14 UTC, Yandex said S3 was available for reading and writing, though operations could be slower. It listed Kubernetes as available with possible problems in individual clusters; compute remained partially available. Yandex AI Studio and security services were listed as partially available, while Yandex Monium was unavailable. The company also said it had temporarily suspended increases in quotas for additional cloud resources. At 10:12 UTC it reported no change in service availability and said support channels had been partly restored, with significant response delays still possible.

The 15:01 UTC update said the cloud platform continued to operate with limited functionality. Yandex again recommended considering alternative platforms to deploy resources more quickly and provide redundancy. Its next planned update was October 12 at 11:00 Moscow time, or 08:00 UTC. The incident remained a live recovery process at that point; the update did not say the facility or every cloud service had recovered.

A facility outage and a wider service disruption

Yandex's October 9 statement had described a full shutdown at Sasovo after damage overnight on October 7 to 8 and infrastructure out of service at Kaluga after an October 9 attack. It gave no restoration estimate for those sites. The Vladimir incident is a dated continuation of that sequence, with the added detail that Yandex Cloud publicly named affected components and advised customers to consider recovery outside the platform.

Cloud status does not capture every consumer experience. AFP reported that users described trouble with Yandex ride-sharing and payment apps on October 11. Those interviews show disruption to some users; they do not measure how many people were affected or establish that every reported app problem came from the Vladimir site. Al Jazeera's report, drawing on AFP and Reuters, also describes the Vladimir shutdown and the preceding facility attacks.

For cloud customers, the status log is a recovery guide with limits. It identifies services that could be read, written to or used in a reduced state at specific times, while warning of delayed support and suspended quota increases. It does not say whether any particular customer's backup is reachable or whether data was lost. The earlier reporting that two AI-development supercomputers were housed at Sasovo does not establish their present condition, and the October 11 listing of partial AI Studio availability does not show whether model training was affected. Those questions need a specific update from Yandex.

Sources & further reading