Summary
ArcadeDB's IMPORT DATABASE command contains a server-side request forgery vulnerability in the way it validates and connects to remote resources. An authenticated attacker may bypass hostname checks through DNS rebinding or redirects to reach cloud metadata services, internal systems, or local files when the command is available and the server can access those resources.
Affected context
The issue occurs in ArcadeDB deployments where authenticated users can invoke the IMPORT DATABASE command and the server can make outbound requests. Risk is increased when validation resolves one destination while the later connection follows redirects or resolves the original URL again, allowing the effective destination to differ from the one that was checked.
Exploit preconditions
- Attack surface: The server-side URL-fetching functionality used by
IMPORT DATABASE - Required access: An authenticated account with permission to invoke the affected command
- Trigger point: Processing and connecting to a database import URL
- Why exploitation is possible: Validation and subsequent connection handling do not consistently bind the request to the same resolved destination, permitting DNS rebinding or redirect-based validation bypasses
Why this happens
The vulnerability is caused by inconsistent destination validation in a server-side HTTP request flow. The security check resolves and evaluates a hostname, but the later network operation re-resolves the raw URL and permits redirects without ensuring that the final destination remains within the approved policy. This behavior maps to CWE-918: Server-Side Request Forgery, because attacker-influenced URLs can cause the server to make requests to otherwise restricted network locations.
Potential impact
- Cloud instance metadata or temporary credentials could be exposed
- Internal services may be accessed from the server's network position
- Sensitive local files could be read where file access is supported by the import mechanism
- Internal application data or administrative interfaces could be disclosed
- Network segmentation controls may be weakened
General mitigation guidance
Mitigation generally focuses on making outbound request validation consistent, restrictive, and bound to the actual connection destination.
Common defensive patterns include:
- Use an explicit allowlist of permitted schemes, hosts, ports, and destination networks
- Resolve and validate destinations immediately before connection, then bind the connection to the validated result
- Revalidate every redirect and disallow redirects by default when they are unnecessary
- Block loopback, link-local, private, reserved, and cloud metadata address ranges
- Avoid allowing user-controlled URLs when a controlled repository or object store abstraction is sufficient
- Apply least-privilege permissions to authenticated users who can initiate imports
- Restrict outbound network access from database server processes through network policy
How teams detect issues like this
Detection approaches can include:
- Static analysis and code review for URL validation that is separated from connection establishment
- Searches for raw URL reuse after hostname resolution, unrestricted redirect handling, and missing private-address checks
- Tests covering DNS changes between validation and connection, redirect destinations, alternate IP representations, and local-file access
- Runtime monitoring of outbound requests from database servers to metadata endpoints, private ranges, unexpected ports, or unusual redirect chains
- Authorization reviews ensuring that only trusted roles can invoke import functionality
- Policy enforcement requiring centralized, approved outbound-request libraries

