EUVD-2026-27091
GHSA-8rm2-7qqf-34qm
Attack ChainAIDerived
Hypothetical attack flow derived from CVE metadata – not a confirmed exploit chain
-
AccessIdentify exposed Prometheus instance
-
DeliverySend crafted snappy payload to /api/v1/read
-
ExploitTrigger unchecked heap allocation
-
ExecutionRepeat concurrently
-
PersistExhaust process memory
-
ImpactCrash Prometheus server
Vulnerability Assessment
| Weakness |
CWE
OWASP Top 10:2025
A06:2025 - Insecure Design
ASVS 5.0
V2 - Validation and Business Logic Resource consumption limits
MITRE ATT&CK
1 technique across 1 tactic: Impact The technique list with ATT&CK links and the kill-chain view is part of the Basic plan and up. |
| Exploitation Conditions | Exploitation requires the Prometheus remote read endpoint (/api/v1/read) to be enabled and network-accessible to the attacker. Default Prometheus installations have this endpoint enabled when the --web.enable-remote-read flag is set or when remote_read configurations are present. Network-level exploitation (AV:N) succeeds against any exposed instance-internet-facing, DMZ, or internal network if attacker has access. No special configuration vulnerability-default snappy decompression handling is affected. No authentication required (PR:N), no user interaction needed (UI:N). Attack complexity is low (AC:L)-standard HTTP POST with crafted binary payload. Limiting factors: if /api/v1/read is disabled, endpoint does not exist; if network access is restricted to localhost or specific trusted IPs via firewall/ACL, remote attackers cannot reach it; if reverse proxy enforces request body size limits (e.g., nginx client_max_body_size), some protection exists but may not prevent the attack since the malicious payload is tiny (decompression bomb, not large upload). |
| Impact | Real-world risk is MODERATE to HIGH for exposed Prometheus instances. CVSS 7.5 (High) reflects network accessibility without authentication (AV:N/AC:L/PR:N/UI:N) and high availability impact (A:H). The attack is trivially simple-no exploitation complexity, no privilege requirements, no user interaction-making it automatable. However, impact is limited to denial of service (no confidentiality or integrity impact). Key risk factors: 1) Prometheus is ubiquitous in modern infrastructure, often exposed internally or via service meshes; 2) the remote read endpoint may be internet-exposed in multi-cluster monitoring setups; 3) the attack requires minimal bandwidth (concurrent 5-byte requests) to exhaust memory; 4) no authentication barrier. Mitigating factors: 1) No evidence of active exploitation per CISA KEV absence; 2) remote read is typically used for federated monitoring, not always enabled or externally exposed; 3) many deployments restrict /api/v1/read to internal networks. EPSS score not provided-without it, cannot assess weaponization likelihood, but the technical simplicity suggests high potential if endpoints are exposed. Organizations using Prometheus for federated queries or with internet-exposed metrics endpoints face higher risk. Those restricting remote read to localhost or internal VPNs face lower practical risk despite high CVSS score. |
| Exploit ScenarioDerived | An unauthenticated attacker identifies an internet-exposed or internal Prometheus instance via port scanning (default 9090) or Shodan/Censys. They craft a minimal 5-byte snappy-compressed payload with a header claiming 256 MiB decoded length and send it to /api/v1/read via HTTP POST. The Prometheus server allocates 256 MiB heap memory per request. The attacker launches 10-50 concurrent requests from a single machine, exhausting available memory within seconds and triggering an out-of-memory crash or severe performance degradation. Legitimate monitoring data collection ceases, disrupting observability for the entire infrastructure. No authentication or special permissions are required, and the attack leaves minimal forensic evidence beyond memory allocation spikes. |
Get an e-mail when this CVE is exploited, patched or re-scored. Free account.
Alert meAnalysisAIv1 – May 04, 2026
Memory exhaustion in Prometheus remote read endpoint allows unauthenticated attackers to crash the monitoring server via maliciously crafted snappy-compressed payloads. The /api/v1/read endpoint in versions prior to 3.5.3 and 3.11.3 accepts compressed request bodies without validating the declared decoded length, enabling a 5-byte payload claiming 256 MiB decoded size to trigger massive heap allocations. Concurrent requests can exhaust memory and crash the Prometheus process. EPSS data not provided; no evidence of active exploitation (not in CISA KEV). Vendor-released patches available in versions 3.5.3 and 3.11.3.
RemediationAI
Upgrade immediately to Prometheus version 3.5.3 or 3.11.3 depending on your release track (see https://github.com/prometheus/prometheus/releases/tag/v3.5.3 and https://github.com/prometheus/prometheus/releases/tag/v3.11.3). These versions implement decoded-length validation with a 32 MiB limit before memory allocation. If immediate patching is not feasible, implement network-level access controls: restrict /api/v1/read endpoint access to trusted internal networks only using firewall rules or reverse proxy ACLs; disable remote read functionality entirely if federated queries are not required (remove --web.enable-remote-read flag or equivalent configuration). Rate limiting at the reverse proxy/load balancer can reduce concurrent request impact but does not eliminate the vulnerability. Monitor memory usage closely for anomalies. Each mitigation has trade-offs: network restrictions may break legitimate federated monitoring; disabling remote read eliminates long-term storage query functionality; rate limiting adds operational complexity and may impact legitimate high-throughput queries.
Affected VersionsENISA
Affected Packages
Source: OSV (Open Source Vulnerabilities)
Affects your stack
Affects 3 packages in go, euvd
Checking this CVE against your own stacks and servers is part of the Basic plan and up.
Technical ContextAI · show details
Prometheus is a widely-deployed open-source monitoring and time series database system, commonly used in cloud-native and Kubernetes environments. The vulnerability resides in the remote read API endpoint (/api/v1/read), which accepts snappy-compressed protocol buffer requests for querying time series data. Snappy compression uses a stream format with a header declaring the decoded length before the actual compressed data. The flaw is a classic CWE-400 (Uncontrolled Resource Consumption) issue where the server trusts the client-supplied decoded length header without validation, allowing an attacker to send a tiny compressed payload (as small as 5 bytes) claiming a massive decoded size (e.g., 256 MiB). The vulnerable code path in storage/remote/codec.go's DecodeReadRequest() function allocates memory based on this untrusted header before decompression occurs. The fix implements a 32 MiB limit check (decodeReadLimit) on the declared decoded length using snappy.DecodedLen() before any memory allocation, rejecting payloads exceeding the threshold.
Know when Prometheus gets its next vulnerability
- Watch products and get an email when something new is published
- Check this CVE against your own stack and servers
- Exploit and PoC references, threat intelligence from 22 sources
No credit card · 7-day full trial
Severity by source
AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
Primary rating from Vendor (GitHub_M).
CVSS VectorVendor: GitHub_M
Lifecycle Timeline
1DescriptionCVE.org
Prometheus is an open-source monitoring system and time series database. Prior to versions 3.5.3 and 3.11.3, the remote read endpoint (/api/v1/read) does not validate the declared decoded length in a snappy-compressed request body before allocating memory. An unauthenticated attacker can send a small payload that causes a huge heap allocation per request. Under concurrent load this can exhaust available memory and crash the Prometheus process. This issue has been patched in versions 3.5.3 and 3.11.3.
Executive ViewAI
Prometheus versions before 3.5.3 and 3.11.3 are vulnerable to unauthenticated denial-of-service attacks through the remote read endpoint, where attackers can crash the monitoring server by sending specially crafted compressed payloads. This impacts observability infrastructure across your organization, potentially blinding security and operations teams to real incidents during an active attack.
If immediate patching is not feasible: (1) Implement network-level access controls to restrict /api/v1/read endpoint to trusted internal sources only; (2) Deploy rate limiting on the remote read endpoint to 10 requests per second per source IP; (3) Configure memory limits on Prometheus process to trigger graceful shutdown before system-wide exhaustion; (4) Enable request size validation in reverse proxy (nginx/HAProxy) to reject payloads claiming decompressed sizes exceeding 50 MiB. Apply vendor patch as priority remediation.
Threat Intelligence Overview
NVD
CVE.org
EPSS
Vulnrichment
VulDB
How this decision was reached
The engine reads each signal below and applies a fixed list of rules from top to bottom. The first rule that matches sets the recommended action – like a statement, every line is shown.
What every action level means
SSVC AssessmentENISA
Priority Score
Attack Technique
Denial of Service attacks render applications or systems unavailable by overwhelming resources or triggering failure conditions.
More in Prometheus
View allPrometheus Exporter Toolkit is a utility package to build exporters. Rated high severity (CVSS 8.8), this vulnerability
Unsafe PHP deserialization in openITCOCKPIT Community Edition 5.3.1 and earlier allows authenticated attackers to inject
Remote code execution in openITCOCKPIT 5.3.1 and earlier via unsafe deserialization in the Gearman worker component, whi
Prometheus monitoring system exposes Azure AD OAuth client secrets in plaintext via its /-/config HTTP API endpoint. Ver
Prometheus is an open-source monitoring system and time series database. Rated medium severity (CVSS 6.1), this vulnerab
Prometheus Blackbox Exporter through 0.17.0 allows /probe?target= SSRF. Rated medium severity (CVSS 5.8), this vulnerabi
blackbox_exporter v0.23.0 was discovered to contain an access control issue in its probe interface. Rated high severity
Scraparr versions 3.0.0-beta through 3.0.1 expose Readarr API keys in plaintext through the /metrics endpoint when the R
Denial of Service in Prometheus and Kibana metricsets can be triggered by sending specially crafted malformed payloads t
Prometheus through 3.15.1 can expose Azure AD or Google IAM bearer tokens when remote_write follows an HTTP 3xx redirect
A stored, DOM based, cross-site scripting (XSS) flaw was found in Prometheus before version 2.7.1. Rated medium severity
Alertmanager handles alerts sent by client applications such as the Prometheus server. Rated medium severity (CVSS 5.4),
Same weakness CWE-400 – Uncontrolled Resource Consumption
View allSame technique Denial Of Service
View allVendor StatusVendor
SUSE
Severity: High| Product | Status |
|---|---|
| openSUSE Tumbleweed | Fixed |
| SUSE Linux Enterprise Module for Package Hub 15 SP7 | Fixed |
| SUSE Linux Enterprise Server 16.0 | Fixed |
| SUSE Linux Enterprise Server for SAP applications 16.0 | Fixed |
| SUSE Manager Client Tools 15 | Fixed |
| SUSE Manager Client Tools for SLE 15 | Fixed |
| SUSE Multi-Linux Manager Client Tools for SLE 15 | Fixed |
| SUSE Manager Proxy LTS 4.3 | Fixed |
| SUSE Manager Proxy Module 4.3 | Fixed |
| SUSE Enterprise Storage 6 | Fixed |
| SUSE Linux Enterprise Module for Package Hub 15 SP5 | Fixed |
| SUSE Linux Enterprise Module for Package Hub 15 SP6 | Fixed |
| SUSE Manager Proxy Module 4.1 | Fixed |
| SUSE Manager Proxy Module 4.2 | Fixed |
| openSUSE Leap 15.3 | Fixed |
| openSUSE Leap 15.4 | Fixed |
| openSUSE Leap 15.5 | Fixed |
| openSUSE Leap 15.6 | Fixed |
| SUSE Multi Linux Manager Tools SLE-15 | Fixed |
Exploit-DB