Breaches and incidents Developing story

Rackspace Spot breach let intruders read customer Kubernetes secrets

Rackspace told customers that an unauthorized party used leaked credentials to act inside Spot Kubernetes clusters on 17 November, reading Secrets and creating pods.

By Junglewise News DeskPublished 5 min read
A dark access card reader mounted on a concrete wall in an empty industrial-style corridor
Photo by Jakub Żerdzicki on Unsplash

Rackspace has told customers of its Spot cloud platform that an unauthorized party used leaked credentials to get inside parts of the service and, in some customer organisations, act as if it were a legitimate user. According to a customer notification reported by BleepingComputer, that access allowed the intruder to start new pods, read the Secrets held in a cluster, and enumerate its nodes and storage volumes.

The reading of Secrets is the most consequential part. Kubernetes Secrets hold passwords, authentication tokens and API keys used to reach databases, cloud services and other systems, so the exposure does not stop at the cluster boundary: anything a Secret unlocks is potentially reachable by whoever read it. Rackspace warned that every Secret in an affected user's clusters was within reach, as were that user's Spot API tokens and refresh tokens. The company has not said how many customers were affected or whether data was taken, and BleepingComputer reported that it asked both questions and received no response.

The incident has not been confirmed by a second independent outlet or by a public primary advisory, and the details below come from the customer notification and related statements as reported by BleepingComputer. Rackspace Spot is a platform for deploying and managing containerised applications on Kubernetes, co-developed with Platform9.

What to do now

If your organisation runs workloads on Rackspace Spot, treat every credential stored in those clusters as potentially exposed. Rackspace's instructions to affected customers, as reported, were:

  1. Reissue your Spot API tokens, and replace any refresh tokens embedded in automation tools, pipelines or scripts.
  2. Download fresh Kubernetes configuration files rather than reusing existing ones.
  3. Rotate every credential stored in cluster Secrets, including database passwords, cloud service keys and third-party API keys. (General practice, not part of the notification as reported: change the credential where it is issued, then update the Secret.)
  4. Review logs from databases, container registries, Git repositories, cloud storage and third-party services going back to 17 November 2026, looking for sign-ins or API calls that used the exposed credentials.
  5. Check your clusters for workloads you did not create. Rackspace flagged pod and object name prefixes i1, glm- and probe as suspicious, and the IP address 23.224.79.106 within the range 23.224.0.0/15.

Prioritise credentials with broad reach: cloud provider keys, registry push tokens and anything that can write to source control. A read of a Secret is cheap; the expensive part is what the key opens later. If the rotation list is long, work outward from the systems that hold customer data.

Who is affected

Only organisations with workloads on Rackspace Spot are in scope, and within that group Rackspace said the intruder was able to act as a legitimate user in some customer organisations rather than all of them. The company has not published a count, which means customers cannot currently tell from public information alone whether they were among those directly touched. The notification went to affected customers, so receipt of one is the clearest available signal.

The clean-up had a complication of its own. The breach was accompanied by an extended outage of the Spot administration console, which left customers temporarily unable to manage their clusters or rotate the very credentials they had just been told to rotate. Sirish Raghuram, co-founder and Chief Growth Officer of Platform9, said in a Reddit discussion that the shutdown was deliberate while the infrastructure was secured. "I know you disagree, but we made the decision that shutting all access down and sanitizing the entire environment is what we can do to best protect everyone using Spot," he wrote. He added: "I've been instrumental in creating Spot and in its operations to date, and I can tell you that the team and I are incredibly disappointed to have let you down in this way."

Technical details

  • Platform: Rackspace Spot, a cloud platform for deploying and managing containerised applications using Kubernetes, co-developed with Platform9.
  • Initial access: compromised or leaked credentials. Rackspace has not described how those credentials were obtained.
  • Window of access: 17 November 2026, roughly 00:20 to 22:00 UTC, according to the customer notification.
  • Capabilities observed: acting as a legitimate user in some customer organisations, starting pods, reading Secrets, and listing cluster nodes and volumes.
  • Potential exposure: all Secrets held in an affected user's clusters, together with that user's Spot API tokens and refresh tokens.
  • Indicators of compromise: IP address 23.224.79.106; IP range 23.224.0.0/15; Kubernetes pod and object name prefixes i1, glm- and probe.

Timeline as reported: the unauthorized access occurred on 17 November 2026. Rackspace learned of the intrusion on 18 November 2026 after a customer reported it, and confirmed the same day that the reporting customer's organisation was affected. Rackspace says it has since revoked affected tokens and sessions, removed unauthorized pods, rotated keys, and tightened access controls and monitoring. After several delays, the company announced that Spot service was fully restored on 21 November 2026.

Several things remain unknown. Rackspace has not said how many customers were affected, whether any data was exfiltrated, how the credentials were leaked, or whether the unauthorized pods were used for anything beyond establishing a presence. No attribution has been offered and none should be assumed. Customers who have not received a notification cannot presently confirm their status from public information, and the best available check is the indicator list above applied to their own cluster and downstream service logs.

Share this article

Related stories

Scan your domain free: see what attackers can see from outside, in about 30 seconds.

This article as Markdown. Atom feed.