CVE-2026-71198: 
Linux Debian vulnerability analysis and mitigation

CVE-2026-71198 is a Server-Side Request Forgery (SSRF) vulnerability in OpenStack Glance, the image service for OpenStack. When the HTTP store backend is enabled, an authenticated user can add an image location URL pointing to arbitrary internal network services via the image location APIs (POST /v2/images/{id}/locations or PATCH /v2/images/{id}). Glance validates only the URL scheme but performs no host, IP, or port filtering, allowing the server to make requests to internal endpoints and expose the responses as image data. The CVE was reported on 2026-08-26 and published on 2026-09-04, with a CVSS v3.1 base score of 7.7 (High) (Red Hat CVE, Red Hat Bugzilla).

The root cause is CWE-918 (Server-Side Request Forgery), stemming from insufficient input validation in the image location API code paths. Specifically, glance/common/store_utils.py's validate_external_location() function only blocks certain URL schemes (e.g., file://, filesystem://, swift+config://) but does not validate the host, IP address, or port of the supplied URL. In contrast, the web-download import path (patched separately via OSSA-2026-004 / CVE-2026-34881) correctly calls validate_import_uri() in glance/common/utils.py, which enforces allowed/disallowed host and port filtering — the location API paths bypass this entirely. When a malicious location is added, glance_store issues HTTP HEAD and GET requests to the attacker-supplied URL, and the response body is stored as image data retrievable via GET /v2/images/{id}/file. Exploitation requires only the add_image_location policy (SERVICE_OR_PROJECT_MEMBER) for the new POST API, or show_multiple_locations=True and the set_image_location policy for the legacy PATCH API (Red Hat Bugzilla).

Successful exploitation enables a full-read SSRF, allowing an authenticated attacker to exfiltrate responses from arbitrary internal network endpoints reachable by the Glance server. The most critical risk is exposure of cloud instance metadata services (e.g., http://169.254.169.254/), which can yield temporary cloud credentials, IAM role tokens, and other sensitive configuration data, potentially enabling lateral movement or privilege escalation within the cloud environment. Only confidentiality is impacted; integrity and availability are not directly affected (Red Hat CVE, Red Hat Bugzilla).

There is no public proof-of-concept exploit and no evidence of in-the-wild exploitation at this time (Red Hat CVE). The vulnerability requires authenticated access with at minimum project member privileges, reducing the attack surface compared to unauthenticated vulnerabilities. The CVE status remains "Reserved" and a patch was noted as not yet available at time of initial disclosure, though Red Hat has since tracked it as patched. Detection is available via Nessus plugin 342782 (Tenable). No CISA KEV listing or threat actor attribution has been identified.

1. Reconnaissance: Identify an OpenStack deployment with Glance's HTTP store backend enabled. Confirm the attacker has at minimum project member (authenticated) access to the OpenStack tenant.
2. Identify target internal endpoint: Determine a valuable internal endpoint to probe, such as the cloud metadata service at http://169.254.169.254/latest/meta-data/iam/security-credentials/ or other internal services accessible from the Glance server.
3. Add malicious image location: Using the OpenStack API or CLI, create or select an existing image and add a location URL pointing to the internal target:

   POST /v2/images/{image_id}/locations
   { "url": "http://169.254.169.254/latest/meta-data/iam/security-credentials/", "metadata": {} }

4. Trigger server-side request: Glance-store automatically issues HTTP HEAD and GET requests to the supplied URL to calculate size and hash, fetching the internal resource's response body as image data.
5. Exfiltrate response: Download the image file via GET /v2/images/{image_id}/file to retrieve the response body from the internal endpoint, which may contain cloud credentials or other sensitive data (Red Hat Bugzilla).

• Network: Outbound HTTP requests from the Glance server to internal IP ranges (e.g., 169.254.169.254, 10.0.0.0/8, 172.16.0.0/12, 192.168.0.0/16) or cloud metadata endpoints, particularly HEAD and GET requests originating from the glance-api process.
• Logs: Glance API access logs showing POST /v2/images/{id}/locations or PATCH /v2/images/{id} requests with url values containing private/internal IP addresses or hostnames; subsequent GET /v2/images/{id}/file requests from the same user shortly after.
• File System: Unexpected image files in Glance storage containing structured data (JSON, XML, plaintext credentials) rather than valid image formats.
• Process: Unusual outbound connections initiated by the glance-api or glance-store process to non-standard internal endpoints visible in network flow logs or host-based monitoring.

A patch is available and tracked via Red Hat Bugzilla (Bug 2524517); operators should apply vendor-provided updates for OpenStack Glance as soon as they are released for their distribution (Red Hat Bugzilla). As immediate workarounds: disable the HTTP store backend if it is not required; restrict the add_image_location and set_image_location policies to trusted administrative users only; and implement network-level controls (firewall rules, egress filtering) to prevent the Glance server from reaching internal metadata services and private IP ranges. Additionally, monitoring Glance server outbound connections for requests to internal endpoints can help detect exploitation attempts (Red Hat CVE).

The vulnerability was reported by Cyril Roelandt, OpenStack Glance PTL and Glance Core Security Team member, and disclosed via the oss-security mailing list. Red Hat has assigned it high severity and is tracking remediation through its security response process. No significant broader media coverage or notable public researcher commentary beyond the initial disclosure has been identified at this time (oss-security, Red Hat Bugzilla).