// troubleshooting guide · elasticsearch
Elasticsearch Java4Shell (CVE-2026-114938) Vulnerability: How to Check and Fix Your Cluster
- Status
- Workaround available
- platforms:
- Self-managed Elasticsearch and Logstash on Linux, Docker, ECK, ECE
- updated:
- November 22, 2026
- reports:
- 3
- Method
- how we research
RUNBOOK · elasticsearch/java4shell-cve-2026-1149384 FIXES
// quick answer
Java4Shell (CVE-2026-114938, disclosed 17 Nov 2026) affects Elasticsearch 7.17.3 and later on x86-64 CPUs with AVX-512 VBMI. Elastic says RCE needs an authenticated user with read access, and unauthenticated users can cause denial of service. Elastic's fix for self-managed nodes: add -XX:-UseBASE64Intrinsics in jvm.options.d and restart.
Most likely fix: Check whether your nodes are affected
How to fix it
// try these in order, the most common fix is first
FIX 1Check whether your nodes are affected
- 01On each node run grep -c avx512vbmi /proc/cpuinfo. A result above 0 means the host is affected, according to Elastic
- 02Remember pods and VMs can move between hosts, so check every x86-64 node in the cluster
Did this fix it?FIX 2Apply Elastic's mitigation file
- 01Create config/jvm.options.d/cve-2026-114938.options (or /etc/elasticsearch/jvm.options.d/ for package installs) containing -XX:-UseBASE64Intrinsics
- 02Restart the node; a rolling restart avoids downtime
- 03Verify with GET _nodes/jvm and look for -XX:-UseBASE64Intrinsics in jvm.input_arguments
- 04Never set -XX:+UseBASE64Intrinsics, because it turns the affected optimization back on
- 05Logstash: add the flag to config/jvm.options (or LS_JAVA_OPTS for Docker and ECK) and confirm it in the startup log
Did this fix it?FIX 3Other deployment types
- 01ECK: add -XX:-UseBASE64Intrinsics to ES_JAVA_OPTS in the podTemplate of each affected nodeSet; ECK rolls the restart automatically
- 02ECE: Elastic says to upgrade to 4.2.2 or 3.8.7 once released and restart deployments; until then contact Elastic support
- 03Elastic Cloud Serverless got a hotfix on 18 Nov 2026 and Elastic Cloud Hosted a configuration change on 20 Nov 2026, so no customer action is needed there
- 04Bring-your-own JDK with the latest fixed Java build also works, but not with the Docker image
Did this fix it?FIX 4Upgrade when the fixed releases ship
- 01Watch Elastic's Java4Shell post for the release of 8.19.26, 9.4.10, 9.5.7 or 9.6.0
- 02After upgrading, you can remove the mitigation file; keep it until every node is on a fixed release
- 03Restrict network access and review which accounts have read access in the meantime
Did this fix it?
SANDBOX · elasticsearch RUNS IN YOUR BROWSER
Try it: import a catalog, search with typos and facets, tune ranking, search by meaning
A hands-on Elasticsearch-style sandbox, including the spots where users get stuck. See where problems like this one start. 1-2 minutes, nothing is saved.
Symptoms
No visible symptom: this is a security exposure, not an outage. Scanners flag CVE-2026-114938 on Elasticsearch or Logstash hosts, or nodes crash unexpectedly with a JVM error. Per Elastic, Kibana, Beats, Elastic Agent, APM Server, ECK Operator and Elastic Maps Server are not affected. Logstash 8.6.2 and later is affected only through the elastic_integration filter (denial of service), and 7.17.x and versions before 8.6.2 are not. Elastic lists fixed releases 8.19.26, 9.4.10, 9.5.7 and 9.6.0 (bundled JDK 27.0.1.1+4) as upcoming, with no release date in its post.
Why it happens
- C-01Heap out-of-bounds write in the HotSpot AVX-512 VBMI Base64 decoding intrinsic in OpenJDK (CVE-2026-114938)
- C-02Elasticsearch 7.17.3 and later bundle a JDK in the affected range
- C-03Node CPU reports avx512vbmi (Intel Ice Lake or newer, AMD Zen 4 or newer)
- C-04Cluster reachable by users or services with at least read access, or by anonymous clients for the denial of service case
// this guide also covers
- elasticsearch java4shell
- CVE-2026-114938 elasticsearch
- elasticsearch UseBASE64Intrinsics
- is elasticsearch affected by java4shell
Still not working?
If Elasticsearch keeps failing, you're not alone. See every Elasticsearch problem users report or the alternatives people switch to.
Have an app idea? Our small team builds any software: web and mobile apps, games, desktop, AI tools. Price agreed together.
Talk to usOther Elasticsearch errors
- Elasticsearch Cluster Health Yellow: Causes and Fixes
- Elasticsearch Cluster Health Red: How to Fix It
- Elasticsearch Unassigned Shards: Why and How to Fix
- Elasticsearch circuit_breaking_exception: Data Too Large
- Elasticsearch High JVM Memory Pressure and Out of Memory
- Elasticsearch Disk Watermark: Index Read-Only (FORBIDDEN/12)
- Elasticsearch Limit of Total Fields [1000] Exceeded
- Why Is Elasticsearch So Slow? Fixing Slow Searches