CVE-2023-28452 is a weakness in CoreDNS through version 1.10.1 that can make a resolver ignore valid responses, causing a denial of service for normal name resolution. NVD scores it 7.5, High. I could not confirm a fixed version from the sources, so check the CoreDNS release notes.
What's affected
The NVD record says CoreDNS through 1.10.1 is affected. A research write-up linked from the record says the same, and covers a companion cache poisoning issue, CVE-2023-30464, that affects the same versions. It names a CoreDNS pull request, number 6014, titled Continue waiting after receiving malformed responses, but it does not say which release contains the fix.
Because I could not confirm a fixed version, check the CoreDNS release notes and security pages and upgrade to a release newer than 1.10.1. The write-up is a researcher's account and not an official advisory.
No second severity score appeared in the sources I read.
How to check if you're exposed
Find the CoreDNS version. On a host you can run coredns -version. In Kubernetes, read the image tag with:
kubectl -n kube-system get deployment coredns -o jsonpath='{.spec.template.spec.containers[0].image}'
Managed Kubernetes services often ship their own CoreDNS build, so check your provider's documentation as well. A version of 1.10.1 or lower is in the affected range.
What to do now
- Upgrade CoreDNS to a release newer than 1.10.1, after confirming in the release notes that the issue is addressed.
- On managed clusters, follow your provider's guidance for CoreDNS updates, since the add-on version may be set by the platform.
- Limit which networks can send DNS responses to your resolvers, and keep cluster DNS off untrusted networks.
- Monitor resolution failures and timeouts, since the effect is a denial of service.
- Check the Gold page for this CVE for later changes.
Why it matters
The record classes the flaw as CWE-290 (Authentication Bypass by Spoofing), authentication bypass by spoofing. The resolver sends a query from a source port and waits for the answer. According to the record, an attacker can forge a response aimed at that source port, without needing to guess the transaction ID, and the resolver then ignores valid answers.
DNS sits underneath every other service. If name resolution fails, everything that relies on names fails with it, including service discovery inside a cluster. For that reason a denial of service in a resolver can look like an application outage rather than a security event.
The related poisoning issue is the reason to take it seriously even though this record is about availability. Both come from the same research on how resolvers handle unexpected responses. Treat the pair as a reason to upgrade, and read the release notes for both identifiers.
Try it yourself
Click through it at your own pace, or open it full screen.
Quick check
Is a fixed version confirmed?
Not in the sources I could read. The record says CoreDNS through 1.10.1 is affected. Check the release notes for the fix.
Sources
Last reviewed: 19 October 2026.