Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB
CVE-2026-111383 Overview
Nezha versions from 2.0.13 before 2.3.15 contain a resource exhaustion vulnerability in the /api/v1/ws/transfer WebSocket endpoint. Authenticated users can open unbounded concurrent WebSocket connections to the dashboard. Each connection creates a subscription through ServerTransferClass.Subscribe without enforcing per-user or global limits. Attackers can hold thousands of simultaneous subscriptions, exhausting file descriptors, memory, and goroutines on the server. The condition blocks agent gRPC reconnections and degrades monitoring availability. The flaw is tracked under [CWE-770: Allocation of Resources Without Limits or Throttling].
Critical Impact
An authenticated attacker can exhaust server file descriptors, memory, and goroutines by opening unbounded concurrent WebSocket connections, blocking agent gRPC reconnections and causing denial of service.
Affected Products
- Nezha monitoring platform versions 2.0.13 through 2.3.14
- Dashboard component exposing /api/v1/ws/transfer
- Deployments reachable by authenticated low-privilege users
Discovery Timeline
- 2026-10-16 - CVE-2026-111383 published to NVD
- 2026-10-16 - Last updated in NVD database
Technical Details for CVE-2026-111383
Vulnerability Analysis
The vulnerability resides in the Nezha dashboard WebSocket handler at /api/v1/ws/transfer. The endpoint accepts WebSocket upgrade requests from any authenticated user. Each upgrade invokes ServerTransferClass.Subscribe, allocating a goroutine, a file descriptor, and in-memory buffers for the subscription. The handler lacks per-user caps, global connection caps, and idle timeouts.
An authenticated attacker can script thousands of concurrent upgrades from a single account. The server retains each subscription until the client disconnects, so the resource footprint grows linearly with attacker intent. Once file descriptors or goroutine budgets are exhausted, the agent gRPC listener cannot accept new reconnections, breaking telemetry collection across the fleet.
Root Cause
The root cause is missing throttling on resource allocation at a WebSocket entry point. The code path trusts the authentication boundary to limit abuse and does not count or cap live subscriptions created by ServerTransferClass.Subscribe. There is no read limit, no idle disconnect, and no quota enforcement for connections per principal.
Attack Vector
The attack vector is network-based and requires low privileges. Any valid dashboard account can establish the WebSocket upgrade. No user interaction is required. The attacker iterates connection creation while holding each session open. The impact targets availability only; confidentiality and integrity are not affected.
// Source: https://github.com/nezhahq/nezha/commit/36b742a7e9a81ed782363dd19efd77b7030fbd0f
// Patch excerpt from cmd/dashboard/controller/fm.go - hardening dashboard WebSocket boundaries
// The official file-manager client uploads in 1 MiB WebSocket messages. Cap
// each complete message at that protocol boundary so gorilla/websocket cannot
// buffer an attacker-sized frame before the IO stream relay sees it.
const fileManagerWebSocketInputLimit int64 = 1024 * 1024
type websocketReadLimiter interface {
SetReadLimit(limit int64)
}
func limitFileManagerWebSocketInput(conn websocketReadLimiter) {
conn.SetReadLimit(fileManagerWebSocketInputLimit)
}
Source: Nezha security commit 36b742a. The patch introduces read limits and hardens dashboard WebSocket boundaries as part of the broader fix shipped in 2.3.15.
Detection Methods for CVE-2026-111383
Indicators of Compromise
- Sudden growth in open file descriptors on the Nezha dashboard process without a corresponding rise in managed agents
- High number of concurrent WebSocket sessions originating from a single authenticated principal to /api/v1/ws/transfer
- Agent-side gRPC reconnection failures and gaps in telemetry while the dashboard process remains running
- Goroutine counts in Nezha metrics growing without plateau
Detection Strategies
- Instrument the reverse proxy or ingress fronting Nezha to count WebSocket upgrades per source IP and per authenticated user, alerting on outliers
- Correlate dashboard process metrics (open FDs, resident memory, goroutine count) with agent reconnect error rates
- Alert on repeated 101 Switching Protocols responses for /api/v1/ws/transfer from the same session within short time windows
Monitoring Recommendations
- Export Go runtime metrics from the dashboard and set thresholds on go_goroutines and process_open_fds
- Review access logs for sustained bursts against the /api/v1/ws/transfer path and tie them to user identity
- Monitor agent-side logs for repeated gRPC reconnect failures that coincide with dashboard resource pressure
How to Mitigate CVE-2026-111383
Immediate Actions Required
- Upgrade Nezha to version 2.3.15 or later, which addresses the unbounded subscription issue
- Restrict dashboard access to trusted operators and remove unused accounts that could be abused for the attack
- Place the dashboard behind an authenticating reverse proxy that enforces per-user connection quotas
Patch Information
The fix is contained in Nezha commit 36b742a7e9a81ed782363dd19efd77b7030fbd0f and released in version 2.3.15. See the GitHub Security Advisory GHSA-ch6h-w872-77mx and the VulnCheck Denial of Service Advisory for the official write-ups.
Workarounds
- Enforce connection concurrency caps at the reverse proxy for the /api/v1/ws/transfer path per source IP and per session cookie
- Apply idle timeouts and maximum session durations at the proxy to force periodic reconnection and prevent indefinite subscription hold
- Reduce the number of dashboard accounts with login privileges until upgrading to 2.3.15
# Example nginx rate limit and connection cap for the vulnerable endpoint
http {
limit_conn_zone $binary_remote_addr zone=ws_conn:10m;
limit_req_zone $binary_remote_addr zone=ws_req:10m rate=5r/s;
server {
location /api/v1/ws/transfer {
limit_conn ws_conn 5;
limit_req zone=ws_req burst=10 nodelay;
proxy_read_timeout 60s;
proxy_send_timeout 60s;
proxy_pass http://nezha_backend;
proxy_http_version 1.1;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection "upgrade";
}
}
}
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.
Sources: National Vulnerability Database, CISA KEV, Exploit-DB

