Information disclosure in image export API in Canonical LXD before 6.5 and 5.21.4 on Linux allows network attackers to determine project existence without authentication via crafted requests using wildcard fingerprints.
Information disclosure in images API in Canonical LXD before 6.5 and 5.21.4 on all platforms allows unauthenticated remote attackers to determine project existence via differing HTTP status code responses.
Path traversal in Canonical LXD LXD-UI versions before 6.5 and 5.21.4 on all platforms allows remote authenticated attackers to access or modify unintended resources via crafted resource names embedded in URL paths.
Cross-Site Request Forgery (CSRF) in LXD-UI in Canonical LXD versions >= 5.0 on Linux allows an attacker to create and start container instances without user consent via crafted HTML form submissions exploiting client certificate authentication.
Template Injection in instance snapshot creation component in Canonical LXD (>= 4.0) allows an attacker with instance configuration
permissions to read arbitrary files on the host system via specially crafted snapshot pattern templates using the Pongo2 template engine.
Information Spoofing in devLXD Server in Canonical LXD versions 4.0 and above on Linux container platforms allows attackers with root privileges within any container to impersonate other containers and obtain their metadata, configuration, and device information via spoofed process names in the command line.
Due to insufficient verification, an attacker could use a malicious client to bypass authentication checks and run RPC commands in a region. This has been addressed in MAAS and updated in the corresponding snaps.
In Canonical Multipass up to and including version 1.15.1 on macOS, incorrect default permissions allow a local attacker to escalate privileges by modifying files executed with administrative privileges by a Launch Daemon during system startup.
In Juju versions prior to 3.6.8 and 2.9.52, any authenticated controller user was allowed to upload arbitrary agent binaries to any model or to the controller itself, without verifying model membership or requiring explicit permissions. This enabled the distribution of poisoned binaries to new or upgraded machines, potentially resulting in remote code execution.
The /log endpoint on a Juju controller lacked sufficient authorization checks, allowing unauthorized users to access debug messages that could contain sensitive information.