CVE-2018-17190

In all versions of Apache Spark, its standalone resource manager accepts code to execute on a 'master' host, that then runs that code on 'worker' hosts

published November 19, 2018

CVSS
—
EPSS
9%
Percentile
95.2
In the wild
Unconfirmed
What it is

⚠️ A crafty request can turn the tables in Apache Spark! While the master host is designed to sit back and relax, it might just be tempted to execute unexpected user code. 🤔 Think of the master host like a restaurant manager who usually just oversees the kitchen, but accidentally takes a customer order and cooks up a dish themselves. That's not how it’s supposed to work! 🍽️ If an attacker sends a specially-crafted request, they could trick the master host into executing malicious code. This could lead to unauthorized access or manipulation of the entire Spark cluster, raising concerns over data integrity and system security.

Put simply

Think of the master host like a restaurant manager who usually just oversees the kitchen, but accidentally takes a customer order and cooks up a dish themselves. That's not how it’s supposed to work! 🍽️ This vulnerability allows a specially-crafted request to execute code on the master host of Apache Spark's standalone resource manager, breaking its intended behavior of only executing code on worker hosts.

What to do

If an attacker sends a specially-crafted request, they could trick the master host into executing malicious code. This could lead to unauthorized access or manipulation of the entire Spark cluster, raising concerns over data integrity and system security. To mitigate this risk, ensure authentication is enabled for standalone clusters. Regularly update to the latest version of Apache Spark to benefit from security patches. Monitor for any unusual requests to your master host. You've got this! Follow these steps and your Spark setup will be safer in no time! 🛡️

The record
Technical detail
CVSS
Not scored
CVSS v4.0
Not supplied
EPSS
0.08792 · 95.2th percentile
Published
2018-11-19T14:00Z
EPSS history
Timeline
  • 19 NOV 14:00Z
    In all versions of Apache Spark, its standalone resource manager accepts code to execute on a 'master' host, that then runs that code on 'worker' hosts
    cvelistv5