Vulnerabilities / Threats

6/18/2018
05:28 PM
Connect Directly
Twitter
LinkedIn
RSS
E-Mail
100%
0%

Exposed Container Orchestration Systems Putting Many Orgs at Risk

More than 22,600 open container orchestration and API management systems discovered on the Internet.

New research confirms that many organizations are deploying workloads to the public cloud without adequate security controls and processes in place first.

Lacework recently used the Shodan search engine, SSL data mining techniques, and some internally developed tools to uncover as many as 22,672 open container orchestration dashboards and API management systems on the Internet. The applications that Lacework discovered during its research were Kubernetes, Mesos Marathon, Swagger API, Red Hat Openshift and Docker Swarm's Portainer and Swarmpit.

Some 95% of the exposed dashboards and management systems were hosted inside of Amazon Web Services. More than half of those exposed services were hosted on AWS regions in the US. Though a vast majority of the open container orchestration interfaces had credentials for controlling access, the fact that they were exposed to the Internet at all is troublesome, says Dan Hubbard, chief security at Lacework.

The dashboard provides top-level access to all the functions required to manage and administer container clusters. Anyone with access to the dashboard can carry out tasks like starting and stopping workloads, adding or modifying applications and setting key security controls.

Container management interfaces that are not properly configured and are exposed on the Internet pose a major risk for enterprises. "Even with credentials set up, information related to the organization might be disclosed," Hubbard points out. This includes information contained in host headers or SSL certificates. Lacework researchers, for instance, were relatively easily able to derive company names from certificates and hostnames without even having to access the exposed container interfaces.

Weak authentication schemes - like inadequate credential requirements - can allow an attacker to brute-force the password of an exposed container management interface. Similarly, not using SSL can allow an attacker to gain access via clear text authentication. Any vulnerabilities in the container application itself are also likely to be easier to exploit via exposed management dashboards.

Lacework's research also uncovered some 300 container management dashboards with no authentication at all, giving attackers a way to easily spin up and stop containers, delete or destroy information, and harvest data about the infrastructure, Hubbard says. Such completely open dashboards also give attackers a way to potentially move laterally to other cloud resources, log into cloud instances and launch new instances for abuse like Bitcoin mining and DDoS. In addition, attackers can exploit the dashboards to set up a backdoor or to drop malicious code for performing command-and-control and other activities.

Because Kubernetes is currently one of the most popular and fastest-growing container orchestration and management systems, Lacework researchers decided to drill down into some of the issues related to the use of these dashboards in public clouds. Although Kubernetes comes with security features like default SSL use and default authentication, Lacework researchers discovered many open Kubernetes dashboards that were still in the process of being set up; dashboards with no authentication; dashboards that could be brute-forced and those that were leaking information about the organization.

Researchers also discovered 38 servers running a container health check service called "healthz" for Kubernetes environments, live on the Internet with no authentication at all. Besides potentially giving attackers a way to potentially perform full remote code execution capabilities, the app also provides a way to monitor workloads and stop them from running via the dashboard.

The data highlights the need for organizations to follow fundamental security best practices when deploying containers in the public cloud. Examples include protecting them behind a proxy, using multifactor authentication and role-based models to control access, and enforcing the use of SSL, Hubbard says. Limiting the scope of what you can do via a management interface is also a good idea. For example, it is not a great idea to have one management hub for all workloads because if one pod is compromised, all workloads get put at risk, he says.

"The general theme of our discoveries is that there needs to be a bridge in communication, process, and technology between DevOps and security," Hubbard notes. "The operating model here is very different and that needs to be acknowledged."

Related Contents:

  

 

Top industry experts will offer a range of information and insight on who the bad guys are – and why they might be targeting your enterprise. Click for more information

Jai Vijayan is a seasoned technology reporter with over 20 years of experience in IT trade journalism. He was most recently a Senior Editor at Computerworld, where he covered information security and data privacy issues for the publication. Over the course of his 20-year ... View Full Bio

Comment  | 
Print  | 
More Insights
Comments
Newest First  |  Oldest First  |  Threaded View
Crowdsourced vs. Traditional Pen Testing
Alex Haynes, Chief Information Security Officer, CDL,  3/19/2019
BEC Scammer Pleads Guilty
Dark Reading Staff 3/20/2019
Register for Dark Reading Newsletters
White Papers
Video
Cartoon Contest
Write a Caption, Win a Starbucks Card! Click Here
Latest Comment: Well, at least it isn't Mobby Dick!
Current Issue
5 Emerging Cyber Threats to Watch for in 2019
Online attackers are constantly developing new, innovative ways to break into the enterprise. This Dark Reading Tech Digest gives an in-depth look at five emerging attack trends and exploits your security team should look out for, along with helpful recommendations on how you can prevent your organization from falling victim.
Flash Poll
The State of Cyber Security Incident Response
The State of Cyber Security Incident Response
Organizations are responding to new threats with new processes for detecting and mitigating them. Here's a look at how the discipline of incident response is evolving.
Twitter Feed
Dark Reading - Bug Report
Bug Report
Enterprise Vulnerabilities
From DHS/US-CERT's National Vulnerability Database
CVE-2015-3965
PUBLISHED: 2019-03-23
Hospira Symbiq Infusion System 3.13 and earlier allows remote authenticated users to trigger "unanticipated operations" by leveraging "elevated privileges" for an unspecified call to an incorrectly exposed function.
CVE-2016-10743
PUBLISHED: 2019-03-23
hostapd before 2.6 does not prevent use of the low-quality PRNG that is reached by an os_random() function call.
CVE-2019-9947
PUBLISHED: 2019-03-23
An issue was discovered in urllib2 in Python 2.x through 2.7.16 and urllib in Python 3.x through 3.7.2. CRLF injection is possible if the attacker controls a url parameter, as demonstrated by the first argument to urllib.request.urlopen with \r\n (specifically in the query string or PATH_INFO) follo...
CVE-2019-9948
PUBLISHED: 2019-03-23
urllib in Python 2.x through 2.7.16 supports the local_file: scheme, which makes it easier for remote attackers to bypass protection mechanisms that blacklist file: URIs, as demonstrated by triggering a urllib.urlopen('local_file:///etc/passwd') call.
CVE-2019-9945
PUBLISHED: 2019-03-23
SoftNAS Cloud 4.2.0 and 4.2.1 allows remote command execution. The NGINX default configuration file has a check to verify the status of a user cookie. If not set, a user is redirected to the login page. An arbitrary value can be provided for this cookie to access the web interface without valid user...