CVE-2026-63136 Overview
CVE-2026-63136 is an uncontrolled resource consumption vulnerability [CWE-400] affecting Elasticsearch. An authenticated user holding search privileges can submit a specially crafted search request that forces a data node to exhaust its available Java Virtual Machine (JVM) heap memory. The result is node unavailability and cluster degradation, mapped to the Excessive Allocation attack pattern (CAPEC-130).
Restoring service after exploitation requires manual operator intervention, extending outage duration. Elastic addressed the issue in advisory ESA-2026-60 covering Elasticsearch versions 8.19.15, 9.2.9, and 9.3.4.
Critical Impact
A low-privileged authenticated user can trigger heap exhaustion on Elasticsearch data nodes, causing cluster downtime that requires manual recovery.
Affected Products
- Elasticsearch versions prior to 8.19.15
- Elasticsearch 9.x versions prior to 9.2.9
- Elasticsearch 9.3.x versions prior to 9.3.4
Discovery Timeline
- 2026-07-21 - CVE-2026-63136 published to the National Vulnerability Database
- 2026-07-22 - Last updated in NVD database
Technical Details for CVE-2026-63136
Vulnerability Analysis
The flaw stems from insufficient bounds enforcement on memory allocated during specific search request handling paths. When Elasticsearch processes the crafted query, the data node allocates unbounded structures on the JVM heap. The allocation continues until the heap is exhausted, at which point the node stops responding to cluster coordination and query traffic.
Because Elasticsearch clusters replicate shard state across data nodes, degradation on one node forces the master to reroute shards. If multiple nodes are targeted in sequence, cluster-wide search availability collapses. The vulnerability does not expose or modify data, limiting impact to availability.
Exploitation requires only search privileges, which are commonly granted to application service accounts and analyst roles. This lowers the barrier to abuse in shared or multi-tenant deployments.
Root Cause
The root cause is a missing or inadequate resource cap on an internal allocation path used during search request parsing or execution. Elastic has not published exploit-facing detail, but the CWE-400 classification and CAPEC-130 mapping indicate the code path allocates memory proportional to attacker-controlled request fields without enforcing a ceiling.
Attack Vector
The attack vector is network-based against the Elasticsearch REST API. An authenticated user issues a search request containing fields or parameters that trigger the excessive allocation code path. No user interaction is required, and the attacker does not need elevated cluster privileges beyond the ability to submit searches.
No verified proof-of-concept code has been published. Refer to the Elastic Security Update ESA-2026-60 advisory for vendor guidance.
Detection Methods for CVE-2026-63136
Indicators of Compromise
- Sudden JVM heap saturation on Elasticsearch data nodes without a corresponding increase in indexing or legitimate query load
- OutOfMemoryError entries in Elasticsearch logs originating from search thread pools
- Data nodes dropping out of the cluster and requiring restart to rejoin
- Repeated search requests from a single principal preceding node instability
Detection Strategies
- Correlate Elasticsearch slow-log entries with heap usage metrics from the _nodes/stats API to identify queries that trigger disproportionate allocation
- Alert on circuit_breaking_exception events tied to the parent or request circuit breakers exceeding configured limits
- Baseline search request patterns per authenticated principal and flag deviations in payload size or query structure
Monitoring Recommendations
- Ship Elasticsearch cluster and node logs to a centralized analytics platform for correlation with authentication events
- Track jvm.mem.heap_used_percent and node departure events as high-priority telemetry
- Enable audit logging for search requests and retain the requesting principal, source IP, and request body signature
How to Mitigate CVE-2026-63136
Immediate Actions Required
- Upgrade Elasticsearch to 8.19.15, 9.2.9, or 9.3.4 or later as documented in ESA-2026-60
- Inventory service accounts and users granted search privileges and revoke access that is not required for business function
- Ensure JVM circuit breaker settings are configured with conservative parent and request limits
Patch Information
Elastic released fixed versions on the schedule documented in Elastic Security Update ESA-2026-60. Apply the patch corresponding to your deployed major version. Elastic Cloud tenants receive the update automatically per Elastic's managed service policy.
Workarounds
- Reduce the indices.breaker.request.limit and indices.breaker.total.limit circuit breaker thresholds to fail crafted queries earlier
- Place a reverse proxy or API gateway in front of Elasticsearch to inspect and rate-limit search requests per principal
- Restrict search privileges to the minimum set of roles required and audit role-mapping configurations
# Tighten Elasticsearch circuit breakers as an interim control
curl -X PUT "https://es.example.local:9200/_cluster/settings" \
-H "Content-Type: application/json" \
-u "$ES_ADMIN_USER:$ES_ADMIN_PASS" \
-d '{
"persistent": {
"indices.breaker.request.limit": "40%",
"indices.breaker.total.limit": "70%",
"search.max_buckets": 10000
}
}'
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

