Securing self-hosted Actions runners inside private data centers against lateral movement #213426
Replies: 2 comments
|
💬 Your Product Feedback Has Been Submitted 🎉 Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users. Here's what you can expect moving forward ⏩
Where to look to see what's shipping 👀
What you can do in the meantime 💻
As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities. Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐ |
|
Two different problems tend to get mixed together in this conversation: keeping a runner from being compromised, and limiting the damage after it is. Worth treating them separately. For the first one, the runner itself isn't a security boundary. GitHub's own secure-use reference is fairly blunt about this: don't use self-hosted runners with public repos, and be careful even with private/internal ones, because anyone who can fork and open a PR can potentially run code on the machine. The For isolation, the closest thing to a real answer is just-in-time ephemeral runners via the REST API, or Runner groups need to be part of it, not an afterthought: restrict the group to selected private repos, leave Egress is where I'd push hardest, because nothing on the runner can be trusted to enforce it. Do it at the network layer — dedicated VLAN/subnet, firewall allowlist, block link-local metadata ( For benchmarks, the CIS Docker Benchmark is useful if runners are containerized; GitHub's secure-use reference and the self-hosted runner reference are the closer fits otherwise. One last thing from the docs that surprises people: a job waiting for a missing label stays queued for 24 hours, not minutes — so capacity gaps look like hangs unless you alert on queue depth. |
Uh oh!
There was an error while loading. Please reload this page.
🏷️ Discussion Type
Bug
Body
Hi community,
We operate private data center infrastructure and host our own self-hosted GitHub Actions runner fleets for sensitive build workflows.
We want to benchmark our data center security controls against community standards, specifically regarding:
Ephemeral runner isolation (preventing rogue PR workflows from escaping container/VM sandboxes into the local data center network).
Egress filtering / internal DNS restrictions to ensure runners cannot reach sensitive internal corporate services.
Secrets handling and credential rotation between data center HSM/vaults and GitHub environments.
For context, here is our current implementation checklist covering physical & virtual runner safeguards:
🔗 Data Center CI/CD Runner Hardening Checklist
How are other teams approaching runner isolation in their private infrastructure? Any recommended CIS benchmarks or network policies you follow?
All reactions