Executive brief
Apache Spark, a popular data processing engine, uses a weak default encryption method for communication between its internal components when encryption is enabled. This flaw could allow an attacker on the same network to intercept and modify data or system messages without being detected. Such an attack could lead to the corruption of data processing tasks or the disruption of Spark operations.
Technical details
Apache Spark versions before 3.4.4, 3.5.2, and 4.0.0 default to AES in CTR mode (AES/CTR/NoPadding) for RPC communication when 'spark.network.crypto.enabled' is set to true but no specific cipher is configured. AES-CTR provides confidentiality but lacks integrity protection (authentication). A man-in-the-middle attacker can perform bit-flipping attacks on the ciphertext to modify RPC traffic, such as heartbeat messages or application data, without detection. The vulnerability is mitigated by explicitly configuring 'spark.network.crypto.cipher' to 'AES/GCM/NoPadding' or by enabling SSL. Patches are available in versions 3.4.4, 3.5.2, and 4.0.0.
Affected products
- Apache Spark < 3.4.4, >= 3.5.0, < 3.5.2
- Apache pyspark < 3.4.4, >= 3.5.0, < 3.5.2
Timeline
- 2025-10-14: disclosed: Initial disclosure on oss-security mailing list
- 2025-10-15: advisory: GitHub Advisory and NVD entry published
References
- https://lists.apache.org/thread/zrgyy9l85nm2c7vk36vr7bkyorg3w4qq
- http://www.openwall.com/lists/oss-security/2025/10/14/11
- https://github.com/apache/spark/commit/05ba74e5f8043601b7af530429ceaf828d7bb329
- https://github.com/apache/spark/commit/405b23b1aef889fad03bf4e3ced6d427dc9a43f4
- https://github.com/apache/spark/commit/f0563ef64c7f42df21b16ccaeb4cf1324ea720f9