Dgraph fixes unauthenticated pprof token leak enabling admin takeover
TL;DR — Dgraph Alpha exposes /debug/pprof/cmdline without auth, leaking the configured admin token and enabling unauthenticated callers to reach admin-only endpoints by replaying it.
What happened
Dgraph is an open-source distributed GraphQL database, where the “Alpha” node exposes HTTP endpoints for queries and administration.
CVE-2026-40173 describes a critical credential disclosure chain: the unauthenticated debug endpoint /debug/pprof/cmdline can expose the full process command line, including an admin token configured via --security "token=...". An attacker can then reuse the leaked credential in the X-Dgraph-AuthToken header to access admin-only endpoints such as /admin/config/cache_mb, resulting in unauthorized privileged administrative access (configuration/operational control actions) when the Alpha HTTP port is reachable by untrusted parties.
Why this matters: leaving debug/diagnostic endpoints reachable on production listeners is a repeatable failure mode that turns “strong auth” into a paper barrier once secrets are exposed via ancillary routes.
Who is impacted
- Deployments running Dgraph where Alpha’s HTTP listener is reachable by untrusted networks and an admin token is configured via
--security "token=...".
| Component | Affected versions (per CVE record) | Fixed version (per CVE record) |
|---|---|---|
dgraph | <= 25.3.1 | 25.3.2 |
What to do now
- Follow vendor remediation guidance and apply the latest patched release available at the time of writing.
"This issue has been fixed in version 25.3.2."
- Inventory where Dgraph Alpha HTTP is exposed (Kubernetes
Servicetype, ingress rules, load balancers, security groups) and confirm/debug/pprof/*is not reachable from untrusted networks. - Treat this as a credential exposure scenario: rotate the Dgraph admin token and review recent access to
/debug/pprof/cmdlineand/admin/*endpoints in HTTP logs. - If you rely on process-argument secrets (e.g.,
--security "token=..."), reassess how those values can surface via diagnostics, crash reports, or metrics endpoints in your runtime environment.
Content is AI-assisted and reviewed by our team, but issues may be missed and best practices evolve rapidly, send corrections to [email protected]. Always consult official documentation and validate key implementation decisions before making design or security choices.
