# Rackspace Spot Customers&#39; Kubernetes Secrets Were Read: Rotate Yours Now

Rackspace says attackers used stolen credentials to read Kubernetes secrets on its Spot platform. Here is how to rotate keys, check for rogue workloads and patch four KEV products.

Author: J.A. Watte
Published: November 22, 2026
Source: https://jwatte.com/blog/rackspace-spot-kubernetes-secrets-rotate-credentials/

---

A stolen login can be enough to read everything you keep secret in a rented cloud. Rackspace has warned customers that attackers used compromised credentials to deploy unauthorized workloads on its Spot platform and to read Kubernetes secrets in customer clusters. Secrets in this context are the API keys, database passwords and tokens your apps use to talk to payment processors, email services and databases. If an intruder can read those, they don't need to break anything else.

You may think this doesn't apply because you've never heard of Rackspace Spot or Kubernetes. That's the point of writing it down. If a developer, agency or hosting company runs your store, portal or internal tools, there's a decent chance something in your stack sits on a rented container platform, and you may not know which one. The question to ask this week is who holds the logins to it and what secrets live behind them.

The same week brought a short list of newly exploited admin and edge products with a November 23 federal fix-by date. If you run any of them, or pay someone who does, it's a five-minute question with a real answer.

## What happened

The analysis here comes from Apprised.news desk briefs, mainly [the tech and cyber desk's November 22 brief](https://apprised.news/desk/tech/2026-11-22) and [the November 21 brief](https://apprised.news/desk/tech/2026-11-21). They summarise public reporting, and I've linked the original sources below so you can check them.

- **Rackspace Spot.** [BleepingComputer reported](https://www.bleepingcomputer.com/news/security/rackspace-spot-breach-let-hackers-access-customer-kubernetes-secrets/) that Rackspace warned customers that attackers used compromised credentials to deploy unauthorized workloads on the Spot platform and read Kubernetes secrets in customer clusters. How many customers were affected, and which secrets were taken, hasn't been published. The tech desk lists a Rackspace follow-up as the thing to watch.
- **Four more exploited flaws.** According to [the desk's November 22 brief](https://apprised.news/desk/tech/2026-11-22), four known exploited vulnerability (KEV) entries were added on November 20 for Apache Roller, VMware vCenter, SAP NetWeaver and Cisco Expressway, with a November 23 federal fix-by date. The KEV catalog is CISA's list of flaws that are being used in real attacks. The fix-by date binds federal agencies, but it's a good deadline for everyone else too.
- **A model-routing layer.** [The November 21 brief](https://apprised.news/desk/tech/2026-11-21) notes CISA added five exploited flaws the day before, including BerriAI LiteLLM, which many companies use to route AI model calls. Synology, REDCap and Apache Axis entries in that batch carried a November 21 fix-by date.
- **The common thread.** The tech desk's analysts agree that stolen credentials matter more than exotic exploits in this week's incidents. That's the desk's read, and it fits the Rackspace account as reported.
- **A caution on labels.** [BleepingComputer reported](https://www.bleepingcomputer.com/news/security/bell-canada-says-agentic-hacker-stole-enterprise-customer-data/) that Bell Canada confirmed an "agentic" intruder stole enterprise customer data. The desk treats the "agentic" label as unverified until the company says more, and so do I.
- **The wider trend.** The desk notes that [Wired](https://www.wired.com/story/brace-yourself-for-years-of-outage-chaos/) and the NYT both frame faster AI-assisted patching and exploitation as a driver of coming outages.

I can't tell you from the reporting how the Rackspace credentials were compromised, or whether your accounts were touched. Nobody outside Rackspace can, yet.

## What it means for a small business

The exposure is indirect for most owners. You rarely hold the cloud login yourself. A freelancer, an agency or an IT contractor does, and your secrets sit behind it. When one person's login falls, everything that login could read falls with it.

### A 12-person online retailer with a freelancer-run cluster

The store backend runs on a rented Kubernetes cluster managed by a freelance developer. The cluster holds three secrets: a payment-processor API key, a database password and an email-service token. If the freelancer's cloud login was phished, an intruder could read all three in minutes. Rotating them is a 30-minute job when someone has the list ready.

The arithmetic: about 2 hours of freelancer time at $150 an hour is $300. Compare that with one fraud incident. Say the intruder uses the email token to send a phishing blast to your 8,000 customers. Even a modest response (a week of your time, a notification letter, a handful of chargebacks at $25 to $100 each) runs well past $300 before you count lost trust. You can't price that last part, but you can price the $300.

### A 40-employee engineering firm with one vCenter server

The firm runs a single VMware vCenter server for its virtual machines and hasn't patched since spring. vCenter is on the November 20 KEV list. The owner asks the IT contractor three things: which version are we on, can you patch it this week, and is the management interface reachable from the internet?

The arithmetic: a patch window of about 3 hours at $150 an hour is $450. Three days of ransomware downtime for 40 staff at an average loaded cost of $45 an hour is 40 x 8 x 3 x $45 = $43,200 in idle payroll alone, before recovery work or lost client deadlines. Even if the odds of getting hit are small, the $450 is the better bet. For background on an earlier vCenter flaw, see [this post on vCenter and NGINX patches](/blog/vcenter-nginx-critical-patches-small-business/).

### A small marketing agency with an AI gateway

The agency pays for a LiteLLM gateway so staff can reach several AI models through one key. The gateway stores the upstream provider keys. The owner checks three things: is the gateway reachable from the internet, has the update been applied, and have the stored provider keys been rotated?

The arithmetic: a leaked provider key running automated requests at, say, $40 an hour overnight for 14 hours is $560, and it can run longer on a weekend. A spending cap with each provider limits the damage to whatever you set, say $200 a day. Rotation takes about an hour. Setting the caps takes another 15 minutes per provider.

## What to do this week

Most of this is questions for whoever runs your infrastructure. You don't need to understand Kubernetes to ask them.

1. **Ask who runs what.** Ask your developer, agency or hosting contact: do we run anything on Rackspace Spot or on Kubernetes, and who has admin logins? Get a written list of every person and service account with access.
2. **If you were on Rackspace Spot, rotate everything.** Treat every secret in your clusters (API keys, database passwords, tokens) as exposed. Rotate them all now, in order of risk: payment keys, database passwords, then email and other tokens. Then watch for Rackspace's follow-up on which customers and secrets were affected.
3. **Look for workloads nobody launched.** In your cloud console, check for compute, containers or servers your team didn't start (odd names, unusual regions). Check the last 30 days of billing for surprise spikes.
4. **Turn on multi-factor authentication** for every cloud, hosting and admin account. Remove accounts for people who have left and contractors who are finished.
5. **Move secrets out of plain config files** into a secrets manager, and give each app its own least-privilege key so one theft doesn't open everything. For the web-facing version of this problem, see [your API keys are in your HTML](/blog/blog-tool-api-secret-leakage/) and the [agentic attackers and leaked keys checklist](/blog/agentic-attackers-exposed-secrets-scan-checklist/).
6. **Confirm patch status for the four KEV products.** Ask your IT provider about VMware vCenter, SAP NetWeaver, Cisco Expressway and Apache Roller (fix-by was November 23), and ask them to confirm none of these admin interfaces is exposed to the open internet. If the date has passed and nobody can answer, treat that as the answer.
7. **Check any AI gateway.** If you use LiteLLM or similar, confirm it's patched and rotate the model-provider keys stored in it. Set spending caps with each provider.
8. **Write a one-page "if a login is stolen" runbook.** Who to call, which keys to rotate first, who decides whether to notify customers. Check whether your cyber insurance requires notice within a set number of days. Staff accounts are the other half of this, so run your own work addresses through the [email breach check](/tools/email-breach-check/).

If you want to see how containers and Kubernetes fit together before talking to your contractor, [this explainer on CI/CD and Kubernetes](/blog/modern-cicd-kubernetes-jenkins/) and the [container visualizer](/tools/container-visualizer/) are free and plain.

## Copy-paste prompts

Paste these into an AI assistant. Each one tells it to use your own documents and to flag what it can't determine.

Use this when you want to pressure-test your contractor's setup before a call.

```text
I run a small business that hosts apps on a rented cloud/Kubernetes platform managed by a contractor. I'm pasting below our hosting invoices, the list of people with admin access, and any notes the contractor has sent me. Using only those documents, write me a plain-English checklist of 10 questions to ask my contractor to confirm whether stolen credentials could let an attacker read our secrets, and say what a good answer looks like for each. If the documents don't tell you something you need (which platform, who has access, where secrets are stored), say so instead of guessing.
[paste invoices, access list, notes]
```

Use this the day you decide to rotate secrets.

```text
A hosting provider says attackers accessed customer Kubernetes secrets. I'm pasting below an inventory of the secrets our web app uses (API keys, database passwords, third-party tokens) and where each is used. Build a step-by-step rotation plan ordered by risk, include how to avoid downtime, and give me a log template with columns: secret, where used, rotated by, date, verified. Work only from my inventory. If something looks missing from it, or you are unsure how a particular service handles rotation, say so and tell me what to check in that vendor's documentation.
[paste inventory]
```

Use this when you or an admin are about to review the cloud account.

```text
I need to review our cloud account for unauthorized activity. I'm pasting below the last 30 days of billing line items and a list of our users and running resources, exported from our console. Tell me what looks out of place (unfamiliar workloads, odd regions, cost spikes, accounts that should have been removed) and write the exact questions I should ask my admin to verify each item. Don't assume anything is safe or unsafe unless the data shows it; mark each finding as "clear", "suspicious" or "can't tell from this data".
[paste billing and resource export]
```

Use this to get your runbook drafted in one sitting.

```text
Draft a one-page incident runbook for our company titled "A cloud or admin login was stolen." I'm pasting our staff list with roles, our cloud and hosting vendors, and our cyber insurance policy summary. Using those, say who to call first, what to rotate in the first hour, what to preserve as evidence, and when to contact the insurer and customers. If the policy summary doesn't state a notice deadline, tell me to look it up rather than inventing one.
[paste staff list, vendors, policy summary]
```

Use this to turn the KEV batch into a message you can send today.

```text
CISA added VMware vCenter, SAP NetWeaver, Cisco Expressway and Apache Roller to its Known Exploited Vulnerabilities list. I'm pasting our IT asset list below. Tell me, in plain terms, which of these products appear in it. Then write a short message to my IT provider asking which versions we run, whether the management interfaces are reachable from the internet, and when each will be patched. If the asset list doesn't name versions, say so and add that to the questions.
[paste asset list]
```

## Download the kit

The kit is a printable markdown pack: an access inventory worksheet, a secrets rotation checklist with an order-of-risk table and log, a 30-minute unauthorized-activity review, a patch tracker for the five products in this story, an email template to your host or contractor, a one-page incident runbook with a notification checklist, and the five prompts above. [Download the Stolen-Login Response Kit](/downloads/rackspace-spot-kubernetes-secrets-rotate-credentials-kit.md).

## Fact-check notes and sources

- **Rackspace Spot breach**: Supports the claim that attackers used compromised credentials to deploy unauthorized workloads and read Kubernetes secrets in customer clusters. [BleepingComputer](https://www.bleepingcomputer.com/news/security/rackspace-spot-breach-let-hackers-access-customer-kubernetes-secrets/)
- **Apprised.news tech and cyber desk, November 22**: Supports the four KEV entries added November 20 with a November 23 fix-by date, the analysts' agreement that stolen credentials matter more than exotic exploits, and the pending Rackspace follow-up on scope. [Desk brief](https://apprised.news/desk/tech/2026-11-22)
- **Apprised.news tech and cyber desk, November 21**: Supports the five-entry KEV batch including BerriAI LiteLLM, the November 21 fix-by date for Synology, REDCap and Apache Axis, and the unverified status of the "agentic" label. [Desk brief](https://apprised.news/desk/tech/2026-11-21)
- **Bell Canada incident**: Supports that Bell Canada confirmed an intruder described as "agentic" stole enterprise customer data. The label is unverified. [BleepingComputer](https://www.bleepingcomputer.com/news/security/bell-canada-says-agentic-hacker-stole-enterprise-customer-data/)
- **Outage and AI-assisted exploitation framing**: Supports the trend claim as framed by the desk. [Wired](https://www.wired.com/story/brace-yourself-for-years-of-outage-chaos/)
- **Worked examples**: The three business scenarios are hypothetical. Hourly rates, headcounts and cost figures are illustrative assumptions, with arithmetic shown in the text.

## Related reading

- [A 9.8 vCenter flaw was exploited since September](/blog/vcenter-nginx-critical-patches-small-business/): the earlier vCenter story, useful if that server is on your network.
- [Agentic attackers hunt your leaked API keys and OAuth secrets](/blog/agentic-attackers-exposed-secrets-scan-checklist/): a scan-first checklist for the keys that matter most.
- [Your API keys are in your HTML and anyone can read them](/blog/blog-tool-api-secret-leakage/): the other way secrets leak, from the website itself.
- [Your vendor is the breach path](/blog/vendor-breach-path-supply-chain-kev-deadlines/): why a provider's mistake becomes your incident.
- [Modern integration environments: CI/CD with Jenkins, Kubernetes](/blog/modern-cicd-kubernetes-jenkins/): plain background on where Kubernetes secrets live.

*This post is informational, not legal, financial or security-consulting advice. It draws on AI-written Apprised.news analysis of public reporting; check the linked sources before acting. Mentions of third parties are nominative fair use. No affiliation is implied.*


---

Canonical HTML: https://jwatte.com/blog/rackspace-spot-kubernetes-secrets-rotate-credentials/
RSS: https://jwatte.com/feed.xml
JSON Feed: https://jwatte.com/feed.json
Hero image: https://jwatte.com/images/rackspace-spot-kubernetes-secrets-rotate-credentials.webp
