Live
Cloudflare Images Analytics Dashboard Adds 30‑Day Delivery VisibilityAmazon ECS Adds Native Blue/Green, Canary, and Linear Deployments for Safer Container RolloutsGemini agentic tools add LLM‑driven automation to Partner Marketing StudioGitHub Actions makes RHEL 9/10 runner images generally available for larger workloadsIntegrating Decider Models to Reduce Latency and Cost in AI Agent WorkflowsAWS Emulator Enables Local Testing of 200+ Services, but Lacks IAM ValidationSageMaker AI monitoring adds automatic latency drift, idle GPU, and configuration gap detection for inference endpointsEmbedding Proactive Governance into Enterprise‑Scale Agentic AICloudflare Images Analytics Dashboard Adds 30‑Day Delivery VisibilityAmazon ECS Adds Native Blue/Green, Canary, and Linear Deployments for Safer Container RolloutsGemini agentic tools add LLM‑driven automation to Partner Marketing StudioGitHub Actions makes RHEL 9/10 runner images generally available for larger workloadsIntegrating Decider Models to Reduce Latency and Cost in AI Agent WorkflowsAWS Emulator Enables Local Testing of 200+ Services, but Lacks IAM ValidationSageMaker AI monitoring adds automatic latency drift, idle GPU, and configuration gap detection for inference endpointsEmbedding Proactive Governance into Enterprise‑Scale Agentic AI
Docker

Docker Hardened Images now ship Java4Shell and NGINX HTTP/2 fixes

AI SummaryPowered by AI

Docker released updated Hardened Images that include patched Java runtimes and NGINX builds addressing the Java4Shell memory‑corruption bug and two HTTP/2 flaws. Practitioners must adopt the new images or apply runtime work‑arounds to keep Java‑based services and NGINX front‑ends secure on vulnerable hosts.

Docker’s Hardened Images catalog was refreshed on November 19 to incorporate fixed builds for the Java4Shell vulnerability (CVE‑2026‑114938) and the recently disclosed NGINX HTTP/2 issues. The update means that the official Docker images for Amazon Corretto, SapMachine, Azul Zulu, and NGINX now contain the upstream patches, while the Eclipse Temurin variant will follow once Adoptium publishes its rebuilt layers. Teams that rely on these runtimes should pull the new tags or re‑tag their pinned references to the patched versions.

What Changed in Docker Hardened Images

The Java4Shell bug is a memory‑corruption flaw in the HotSpot Base64 decoder that triggers when the JVM runs on x86‑64 CPUs supporting AVX‑512 VBMI. The decoder writes up to 62 bytes beyond a heap buffer, enabling heap disclosure and remote code execution. Docker responded by rebuilding its Hardened Images with Java 17.0.21.1, 21.0.13.1, 25.0.5.1, and 27.0.1.1 (or vendor‑specific suffixes) for Corretto, SapMachine, and Zulu. The NGINX images were rebuilt on versions 1.30.7 and 1.31.9, which contain the upstream HTTP/2 patches.

Why It Matters for Engineers

Any container that bundles a Java runtime is potentially exposed, even if the application itself is not written in Java. The vulnerability only manifests on hosts with the AVX‑512 VBMI instruction set – typically Intel Ice Lake Xeon, AMD EPYC, or Ryzen Zen 4 CPUs – but the same image can be safe on one node and vulnerable on another. For AI/ML pipelines, CI runners, and platform services that frequently spin up Java‑based containers, an unpatched image could become an attack vector without any credential compromise.

Operational and Security Implications

Because Docker’s Hardened Images are immutable, the fix is only present after a rebuild. Practitioners must verify the actual Java build inside a running container; a tag name or rebuild timestamp is insufficient. The reliable check is:

docker run --rm --entrypoint java <image> -version

to confirm the version string contains one of the fixed numbers. If the host is AVX‑512 capable, the JVM will enable the accelerated Base64 routine unless explicitly disabled. The temporary mitigation is to start Java with the flag -XX:-UseBASE64Intrinsics and verify the flag via:

docker run --rm --entrypoint java <image> -XX:+PrintFlagsFinal -version | grep -w UseBASE64Intrinsics

All JVM instances in the container must receive the flag, not just the first process. For Kubernetes clusters, operators can query node capabilities with:

kubectl get nodes -L kubernetes.io/arch,node.kubernetes.io/instance-type

or rely on Node Feature Discovery labels such as feature.node.kubernetes.io/cpu-cpuid.AVX512VBMI=true. CI/CD workers that compile or test Java code also need the same verification because they often run on the same hardware pool.

Related CloudNinjas coverage: hands-on guides.

What This Means For Practitioners

Immediate actions:

  • Pull the latest Docker Hardened Image tags for Corretto, SapMachine, Zulu, and NGINX; update any pinned digests.
  • Run java -version inside each container to confirm the fixed build number.
  • If the host reports AVX‑512 VBMI, add -XX:-UseBASE64Intrinsics to the JVM startup until a fixed runtime is available (e.g., Temurin).
  • Audit node inventories for the vulnerable CPU flag set and consider isolating or upgrading those nodes.
  • Monitor Docker’s catalog for the upcoming Temurin update and replace the image as soon as it appears.

By treating the host as potentially vulnerable until proven otherwise, teams can avoid accidental exposure of Java‑based services and keep NGINX front‑ends protected from the HTTP/2 bugs. Ongoing vigilance—checking image versions, CPU feature flags, and applying the recommended JVM flag—will close the window of exposure while the ecosystem transitions to the patched builds.

Originally published atDocker Blog