Skip to content
CVE-2026-100397

Rucio Unauthenticated Admin Impersonation via SSH Challenge Tokens (CVE-2026-100397)

Detects exploitation of CVE-2026-100397, an authentication bypass in Rucio. The function validate_auth_token in lib/rucio/core/authentication.py does not distinguish short-lived SSH challenge tokens from real authentication tokens, because both are stored in the same token table. Challenge tokens are generated from an account name supplied in the X-Rucio-Account header without prior authentication, so an unauthenticated attacker can request a challenge token for any account, including an administrative one, and present it as a valid auth token. The attacker requests a token at /auth/ssh_challenge_token, receives it in the X-Rucio-SSH-Challenge-Token header, then sends it as X-Rucio-Auth-Token to other API endpoints as that account. This is tracked as CWE-287 and CWE-305. Detection focuses on authentication requests to the Rucio SSH auth endpoints, anomalous account values in X-Rucio-Account headers, and subsequent privileged administrative API calls (account creation, identity addition, RSE management) performed with freshly minted tokens from unexpected source IPs.

Vulnerability Intelligence

Public PoC

What is CVE-2026-100397 Rucio Unauthenticated Admin Impersonation via SSH Challenge Tokens (CVE-2026-100397)?

Rucio Unauthenticated Admin Impersonation via SSH Challenge Tokens (CVE-2026-100397) (CVE-2026-100397) maps to the Initial Access and Privilege Escalation and Credential Access tactics — the adversary is trying to get into your network in MITRE ATT&CK.

This page provides production-ready detection logic for Rucio Unauthenticated Admin Impersonation via SSH Challenge Tokens (CVE-2026-100397), covering the data sources and telemetry it touches: IIS/Reverse-proxy access logs, Rucio server HTTP logs. The queries below are rated critical severity at medium confidence, and ship for 7 SIEM platforms — KQL, SPL, Elastic, QRadar, Sumo, YARA-L, LogScale.

MITRE ATT&CK

Tactic
Initial Access Privilege Escalation Credential Access
Microsoft Sentinel / Defender
kusto
// df00tech DF00-5ce6ae0814 · https://df00tech.com/detections/CVE-2026-100397 · licence: https://df00tech.com/detections/licence
let sshAuthEndpoints = dynamic(["/auth/ssh", "/auth/ssh_sign", "/auth/ssh_challenge_token"]);
let suspiciousWindow = 10m;
// Rucio server access logs ingested as W3CIISLog or custom CommonSecurityLog
let sshAuth = W3CIISLog
| where csUriStem has_any (sshAuthEndpoints)
| where scStatus == 200
| project TimeGenerated, cIP, csUriStem, csUserAgent, Account = csUsername, scStatus;
sshAuth
| summarize AuthCount = count(), Endpoints = make_set(csUriStem), Accounts = make_set(Account) by cIP, bin(TimeGenerated, suspiciousWindow)
| where AuthCount >= 3 or array_length(Accounts) > 1
| extend Reason = "Repeated or multi-account Rucio SSH challenge-token auth from single source"
| project TimeGenerated, cIP, AuthCount, Endpoints, Accounts, Reason

Flags bursts of requests to the Rucio SSH challenge-token authentication endpoint from a single source IP, or requests cycling multiple account identities, consistent with challenge-token abuse to mint an admin-scoped token.

critical severity medium confidence

Data Sources

IIS/Reverse-proxy access logs Rucio server HTTP logs

Required Tables

W3CIISLog

False Positives

  • Legitimate automated Rucio clients (DIRAC, FTS, data management daemons) authenticating with SSH keys at high frequency
  • Health-check or monitoring probes hitting the auth endpoints
  • A shared service node authenticating multiple service accounts in quick succession

Query fingerprint DF00-5ce6ae0814 · licence

Sigma logsource & community rules

The detection logic for Rucio Unauthenticated Admin Impersonation via SSH Challenge Tokens (CVE-2026-100397) (CVE-2026-100397) above is provided in a vendor-neutral form so you can deploy it on any SIEM. The same logic is shipped here as native KQL (Microsoft Sentinel / Defender), SPL (Splunk), Elastic (Elastic Security (EQL)), QRadar (IBM QRadar (AQL)), Sumo (Sumo Logic CSE), YARA-L (Google Chronicle / SecOps), LogScale (CrowdStrike LogScale (CQL)) queries. In Sigma terms, this detection targets the following logsource:

logsource:
  category: network_connection
  product: windows

Browse the community-maintained Sigma rules for this technique:


Testing Methodology

Validate this detection against 3 adversary techniques from Atomic Red Team. Each test below lists the behaviour to exercise and the telemetry you should expect to see. Executable commands and cleanup steps are further down this page.

  1. Test 1 Rucio SSH auth endpoint burst

    Expected signal: Five HTTP GET requests to /auth/ssh_challenge_token with X-Rucio-Account: root from one source IP in the web/proxy access logs.

  2. Test 2 Account-cycling against SSH auth endpoint

    Expected signal: Multiple /auth/ssh_challenge_token requests with differing X-Rucio-Account values from one source IP.

  3. Test 3 Privileged API call after SSH auth

    Expected signal: An /auth/ssh_challenge_token request followed within minutes by a POST /accounts/* from the same source IP.


Response Playbook

Triage

  1. Confirm the Rucio server version from the deployment (pip show rucio / container image tag) and compare against affected ranges (>=1.14.0.post1 <35.9.1, >=36.0.0rc1 <38.6.1, >=39.0.0rc1 <40.4.2, >=41.0.0rc1 <41.1.1).
  2. Review the Rucio auth server logs for source IP(s) hitting /auth/ssh_challenge_token and identify which account identity was requested via the X-Rucio-Account header.
  3. Correlate any admin-scoped token issuance with subsequent privileged API calls (account/identity/RSE creation or modification) within the same session window.
  4. Determine whether the source IP belongs to a known grid service, monitoring system, or an unrecognized/external host.

Containment

  1. Expire or revoke all active Rucio auth tokens to force re-authentication and kill any impersonated admin sessions.
  2. Block the offending source IP(s) at the reverse proxy/WAF and temporarily disable the SSH authentication method on the Rucio server until patched.
  3. Upgrade Rucio to a fixed release (35.9.1, 38.6.1, 40.4.2, or 41.1.1 and later) immediately.

Evidence Collection

  1. Preserve the Rucio httpd/gunicorn access and error logs covering the suspected exploitation window.
  2. Export the Rucio tokens table and identities/account_map tables to capture issued tokens and any attacker-added SSH identities.
  3. Snapshot the Rucio database and capture auth server configuration showing enabled authentication methods.

Escalation Criteria

  • ! An admin-scoped token was issued to a source IP not matching any known administrator or service account.
  • ! New Rucio accounts, identities, or RSEs were created or modified following a suspicious SSH authentication, indicating successful privilege abuse.

Investigation Guide

Forensic Artifacts

  • > Rucio tokens table rows with admin-scoped accounts and unexpected issuance IPs
  • > New or modified rows in the identities and account_map tables (attacker-added SSH keys)
  • > Rucio httpd access log entries to /auth/ssh_challenge_token with X-Rucio-Account headers

Tuning Guidance

Build an allowlist of source IPs for known grid service clients (FTS, DIRAC, data management daemons) and monitoring probes, then exclude them from the volume thresholds. Lower the auth_count threshold in low-traffic environments and raise it where legitimate automated clients authenticate frequently. Prioritize any event where an admin-scoped token is followed by write operations to /accounts, /identities, or /rses endpoints.


Hunting Queries

Enumerate all historical Rucio SSH authentication attempts grouped by source IP and requested account to baseline normal vs anomalous callers.

Hunting — KQL
kql
// df00tech DF00-5ce6ae0814 · https://df00tech.com/detections/CVE-2026-100397 · licence: https://df00tech.com/detections/licence
W3CIISLog | where csUriStem has "/auth/ssh" | summarize by cIP, csUsername, bin(TimeGenerated, 1h) | order by TimeGenerated desc
Hunting — SPL
spl
``` df00tech DF00-5ce6ae0814 · https://df00tech.com/detections/CVE-2026-100397 · licence: https://df00tech.com/detections/licence ```
index=web uri_path="/auth/ssh*" | stats count values(rucio_account) by src_ip | sort - count

Atomic Red Team Tests

Test 1 Rucio SSH auth endpoint burst
linux

Simulate repeated requests to the Rucio SSH challenge-token endpoint from a single host to trigger the volume-based detection.

Command

bash
for i in $(seq 1 5); do curl -s -o /dev/null -w '%{http_code}\n' -H 'X-Rucio-Account: root' https://rucio-lab.example.org/auth/ssh_challenge_token; done

Cleanup

bash
echo 'No persistent changes; clear test logs if desired'

Expected Telemetry

Five HTTP GET requests to /auth/ssh_challenge_token with X-Rucio-Account: root from one source IP in the web/proxy access logs.

Expected Detection

KQL/SPL volume rule fires on auth_count >= 3 from a single source IP.

Test 2 Account-cycling against SSH auth endpoint
linux

Issue auth requests cycling multiple account identities to trigger the multi-account branch of the detection.

Command

bash
for acct in root admin ddmadmin svcuser; do curl -s -o /dev/null -H "X-Rucio-Account: $acct" https://rucio-lab.example.org/auth/ssh_challenge_token; done

Cleanup

bash
echo 'No cleanup required'

Expected Telemetry

Multiple /auth/ssh_challenge_token requests with differing X-Rucio-Account values from one source IP.

Expected Detection

Rule fires on distinct_accounts > 1 for a single source IP.

Test 3 Privileged API call after SSH auth
linux

Simulate a successful SSH challenge-token request followed by a privileged account-creation API call to trigger the EQL sequence correlation.

Command

bash
curl -s -o /dev/null -H 'X-Rucio-Account: root' https://rucio-lab.example.org/auth/ssh_challenge_token; curl -s -o /dev/null -X POST -H 'X-Rucio-Auth-Token: TESTTOKEN' https://rucio-lab.example.org/accounts/attacker

Cleanup

bash
curl -s -X DELETE -H 'X-Rucio-Auth-Token: TESTTOKEN' https://rucio-lab.example.org/accounts/attacker || true

Expected Telemetry

An /auth/ssh_challenge_token request followed within minutes by a POST /accounts/* from the same source IP.

Expected Detection

Elastic EQL sequence rule correlates the auth event with the subsequent privileged admin API call.

Related Detections

Blog posts about this detection

Browse by tactic

Full ATT&CK matrix Search all detections