/

CVE-2026-44168

Severity:

HIGH RISK

(8.0)

Command Injection (CWE-78) in MariaDB SST Donor Handling

CVE-2026-44168

Modified

Vulnerability type

Category:

Command Injection

CWE:

CWE-78

Impact:

Remote Code Execution on donor node during SST

Severity:

HIGH RISK

(8.0)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

A command injection vulnerability in MariaDB's State Snapshot Transfer (SST) flow allows a malicious joiner node to influence how the donor node constructs shell commands. By sending crafted parameters that are interpolated into the donor's command line, an attacker can execute arbitrary shell commands on the donor host. This becomes risky in clustered or replicated deployments where untrusted or compromised nodes can participate in SST using the mariabackup method.

Affected context

The vulnerability occurs in MariaDB server deployments that use replication or clustering mechanisms relying on State Snapshot Transfer, specifically with the mariabackup SST method.

It manifests when a donor node builds shell command lines using parameters provided by a joiner node and fails to robustly validate or sanitize all of those parameters.

Environments that allow semi-trusted or potentially compromised nodes to join a cluster, or where control over joiner configuration is weak, are more exposed to this risk.

Exploit preconditions

  • Attack surface: MariaDB donor node participating in SST with the mariabackup method.
  • Required access: Ability to act as, control, or impersonate a joiner node that initiates an SST process.
  • Trigger point: The moment the donor processes SST parameters from the joiner and interpolates them into a shell command line.
  • Why exploitation is possible: Donor-side code constructs shell commands by interpolating joiner-supplied parameters without strict validation, quoting, or argument separation, enabling injection of additional shell tokens and commands.

Why this happens

The root cause aligns with CWE-78: Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection').

The donor node assembles shell commands for the mariabackup SST by interpolating values supplied by the joiner directly into a command string.

Because special shell characters and separators are not consistently neutralized or the parameters are not passed as strongly-typed arguments, an attacker-controlled joiner can smuggle arbitrary shell syntax into these commands.

At an abstract level, this reflects the insecure pattern of dynamically building OS command strings using untrusted input instead of using parameterized or API-based execution primitives with strict validation.

Potential impact

  • Arbitrary command execution on the donor node under the privileges of the MariaDB/SST process.
  • Access to or modification of database data hosted on the donor instance.
  • Installation of backdoors or further malware on the donor host.
  • Lateral movement to adjacent systems reachable from the compromised donor server.
  • Disruption of cluster availability by corrupting data or interfering with replication.

General mitigation guidance

Mitigation is typically achieved by eliminating direct interpolation of untrusted values into shell command strings and tightening trust boundaries between cluster nodes.

Recommended defensive patterns include:

  • Avoid constructing shell commands via string concatenation or interpolation with values received from peer nodes; instead, use APIs or argument arrays that separate executable paths from data.
  • Apply strict, allow-list-based validation on all SST parameters accepted from joiner nodes, rejecting or normalizing any content that contains shell metacharacters or unexpected values.
  • Constrain the execution environment of SST processes using least-privilege accounts, restricted file system permissions, and OS-level sandboxing to limit impact if command execution occurs.
  • Treat cluster join operations as a privileged action, ensuring that only authenticated and explicitly trusted nodes can initiate SST.
  • Reduce the amount of joiner-controlled configuration that influences donor-side command execution, preferring fixed, server-side configuration where feasible.

How teams detect issues like this

Detection efforts can focus on both code and runtime behavior.

For code-level detection:

  • Use static analysis rules to flag patterns where external input influences shell command strings, particularly in SST or replication-related code paths.
  • During code review, look for joiner-supplied parameters being interpolated into system(), popen(), exec*(), or shell wrappers without robust validation and safe argument passing.
  • Enforce repository policies that disallow ad-hoc shell construction with peer-supplied values in cluster orchestration logic.

For runtime detection:

  • Monitor SST-related processes for unusual command-line arguments or unexpected subprocess invocations originating from donor nodes.
  • Correlate SST events with system audit logs or intrusion detection alerts that indicate suspicious shell commands or privilege escalation attempts.
  • Apply host-based intrusion detection or EDR solutions to capture anomalous process trees spawned from MariaDB or SST components.

Sources

SourceLink
github.comgithub.com/MariaDB/server/security/advisories/GHSA-vwf7-w26c-9w5h
jira.mariadb.orgjira.mariadb.org/browse/MDEV-39413
access.redhat.comaccess.redhat.com/errata/RHSA-2026:25143
access.redhat.comaccess.redhat.com/errata/RHSA-2026:25145
access.redhat.comaccess.redhat.com/errata/RHSA-2026:33093
access.redhat.comaccess.redhat.com/errata/RHSA-2026:33412
access.redhat.comaccess.redhat.com/errata/RHSA-2026:33464
access.redhat.comaccess.redhat.com/errata/RHSA-2026:33481
access.redhat.comaccess.redhat.com/errata/RHSA-2026:33482
access.redhat.comaccess.redhat.com/errata/RHSA-2026:49522
access.redhat.comaccess.redhat.com/security/cve/CVE-2026-44168
bugzilla.redhat.combugzilla.redhat.com/show_bug.cgi?id=2488450
security.access.redhat.comsecurity.access.redhat.com/data/csaf/v2/vex/2026/cve-2026-44168.json

Frequently asked

questions

It can be, if an attacker can control or impersonate a joiner node that participates in SST, potentially over networked cluster links. The exploitability depends on how cluster membership and node authentication are managed.

The attacker must be able to influence the parameters sent by a joiner during SST, which typically requires access to a MariaDB node configuration or the ability to join or simulate a node in the replication or cluster topology.

The safest pattern is to stop interpolating joiner-supplied values directly into shell command strings on the donor side and instead rely on strict allow-listed parameters, argument-array execution APIs, and hardening of which configuration elements can be influenced by peers.

Future prevention relies on secure coding practices around command execution: never build shell commands from untrusted input, validate and normalize all inter-node parameters, restrict trust between cluster components, and incorporate security-focused code review and static analysis into the development lifecycle.

Find out what's already
exploitable in your codebase.