Attackers used leaked credentials to get into parts of Rackspace Spot on November 17, and for roughly 22 hours they could read Kubernetes Secrets in some customer clusters. The Rackspace Spot breach matters because a Secret rarely stays inside the cluster that holds it. A stolen password, token or API key can open a database, a cloud account or another service entirely.
The account reached the public through a customer notice that BleepingComputer obtained and reported on November 21, the same day Rackspace restored full service. Rackspace has not published a count of affected customers, and a reporter who asked basic questions about scope got no response when the story ran. Treat the notice as the most detailed account available, and check Rackspace's status page and direct customer communications for the official version before acting on any single detail.
What the Rackspace Spot breach exposed
According to the notice, an unauthorized party "used leaked credentials to access parts of Rackspace Spot." Spot is the Kubernetes cloud platform Rackspace co-developed with Platform9. In some customer organizations, the intruder could act as a legitimate user, create pods, read Kubernetes Secrets, and list nodes and volumes.
Those permissions let the attacker deploy unauthorized workloads inside affected clusters. Pod creation and Secret access together are a strong combination, because an attacker who can run new workloads and read stored credentials does not need to break anything else in the cluster to reach what sits behind it.
The affected group is US companies and developers that run workloads on Spot, along with the systems their cluster Secrets can unlock. The notice describes exposure of databases, cloud services and other external systems tied to those credentials. That scope is why the response is bigger than a single platform reset.
How the Rackspace Spot breach unfolded
The timeline is short, but each date carries a different lesson.
| Date (2026) | What happened |
|---|---|
| November 17, about 00:20 to 22:00 UTC | Intruder access window, roughly 22 hours |
| November 18 | Rackspace identifies the intrusion after a customer report |
| November 21 | Full Spot service restored after several delays |
Rackspace says it then revoked affected sessions and tokens, removed unauthorized pods, rotated keys and hardened controls. The intrusion also pushed the Spot administration console offline for an extended period. Customers who needed to manage clusters or rotate credentials during the response had to work around that outage.
The timeline rests on a small set of outlets, so confirm exact dates with Rackspace before you cite them in a customer or compliance report.
Why Kubernetes Secrets turn one cluster into a wider problem
Kubernetes Secrets hold passwords, authentication tokens and API keys. Those credentials can reach databases, cloud services and other systems, which is why a single exposed cluster can cascade. A threat-research post from Aviatrix argues that attackers increasingly target secrets stores and orchestration-layer permissions, and this incident fits that pattern.
The practical consequence is that Spot tokens are only the first thing to rotate. Rackspace's guidance, as relayed in reporting, says any Secret in affected clusters, along with Spot API tokens and refresh tokens, should be treated as compromised. Anything those Secrets unlock should be treated the same way, including databases and third-party APIs.
What Rackspace has told customers to do
Rackspace's customer guidance, as reported, asks customers to take these steps:
- Reissue Spot API tokens.
- Replace refresh tokens used in automation.
- Download new Kubernetes config files.
- Rotate every credential stored in cluster Secrets.
- Review service logs from November 17.
Order matters. Start with the platform credentials, Spot API tokens and refresh tokens, because they control access to the cluster management layer. Then move to the credentials those clusters hold, since the Secrets are the path to outside systems. Plan for CI/CD pipelines and deployment scripts to break during the swap, and schedule the change so a failed build does not land during a release.
Teams with many clusters should keep a written inventory of every Secret, the systems it authenticates to, and the person who owns each one. That inventory turns a vague instruction into a checklist. It also creates a rotation record, which matters if a customer or regulator later asks what changed and when.
How to hunt for signs of compromise
Reporting has shared indicators that customers can check against their own environments. One is the IP address 23.224.79.106, which falls within the 23.224.0.0/15 range. The others are suspicious pod and object names that begin with the prefixes "i1", "glm-" and "probe".
Those indicators come from a secondary report, and Rackspace has not confirmed them in the sources reviewed. Use them as leads, not a complete list. Look for pods or objects you do not recognize in every cluster, not only the ones you think were touched, and search logs for traffic from the listed range during the November 17 window.
If you find a match, isolate the cluster, preserve logs before rotating anything, and open a ticket with Rackspace support. Rotating first can erase the evidence you need to determine what the intruder read.
What is still unknown
Several basic facts are missing. Rackspace has not said how many customers were affected. The sources also do not explain how the credentials leaked, and no downstream compromise has been established.
Platform9's co-founder and Chief Growth Officer, Sirish Raghuram, offered a short apology: "the team and I are incredibly disappointed to have let you down in this way." The statement expresses regret but does not answer the open questions about scope or stolen data.
Where this fits in a rough week for cloud providers
The incident landed during a week already crowded with outages and supply-chain breaches. A weekly cybersecurity roundup from The Layman Security grouped the Rackspace case with other supply-chain and third-party breaches from the same period. USA 24 has also covered a Microsoft Entra Connect Health breach that exposed admin emails and tenant data, and a Fireworks AI breach that reached Uber drivers and Cursor and Notion users.
The common thread is trust. When a cloud platform or identity system is compromised, attackers inherit the access customers have built around it. Providers gain when they disclose quickly and clearly. Customers lose time when they have to rebuild their own access from scratch.
What to watch next
Three things will show how serious this incident is. The first is an official Rackspace notice with a scope count and a clear answer on whether customer data left the environment. The second is whether Rackspace names the source of the leaked credentials, since that will show whether other services could be exposed in the same way. The third is whether the Spot console and cluster management stay stable in the weeks ahead.
Until those answers arrive, rotate first and assume the worst about any Secret that touched a cluster during the November 17 window.




