Cross-Site Request Forgery & weak access control in QRadar ConfigServices webservice

Abstract

The QRadar web application is deployed with Apache Axis to expose a number of SOAP services. No measures have been implemented in Axis and/or QRadar to prevent Cross-Site Request Forgery attacks against these webservices. Due to this it is possible for an attacker to call any exposed service via Cross-Site Request Forgery. A successful attack requires that the attacker tricks/forces a logged in victim to visit the attacker's specially crafted URL.

Besides the lack of Cross-Site Request Forgery protection, most methods also lack proper access control checks. A handful of these methods perform some form of access control, but most methods can be called by any authenticated user. This could for example be used by a logged in attacker to gain access to sensitive information (eg, login credentials).

Tested versions

This issue was successfully verified on QRadar Community Edition version 7.3.1.6 (7.3.1 Build 20180723171558).

Fix

IBM reports that Apache Axis is no longer used and therefore this issues has been resolved in upstream builds. In addtion, it is stated that thist issue is resolved in QRadar Community Edition version 7.3.3.

Introduction

QRadar is IBM's enterprise SIEM solution. A free version of QRadar is available that is known as QRadar Community Edition. This version is limited to 50 events per second and 5,000 network flows a minute, supports apps, but is based on a smaller footprint for non-enterprise use.

The QRadar web application is deployed with Apache Axis to expose a number of SOAP services. By default, Axis allows users to call the SOAP services via a GET request. The GET request is internally converted to a SOAP envelope, before it is processed by Axis. No measures have been implemented in Axis and/or QRadar to prevent Cross-Site Request Forgery attacks against the webservices exposed by Axis. Due to this it is possible for an attacker to call any exposed service via Cross-Site Request Forgery. A successful attack requires that the attacker tricks/forces a logged in victim to visit the attacker's specially crafted URL.

Besides the lack of Cross-Site Request Forgery protection, most methods also lack proper access control checks. A handful of these methods perform some form of access control, but most methods can be called by any authenticated user. This could for example be used by a logged in attacker to gain access to sensitive information.

By calling the getNvaProperty() method, it is possible to retrieve any 'NVA' configuration setting. Sensitive settings, like passwords, are stored encrypted, however there is also a getDecrypted() method that allows these values to be decrypted. Some passwords are reused for different services, which also allows users to elevate their own privileges. For example, the property jpa.connection.password is used for connecting to PostgreSQL, but is also used as the password for the ConfigServices account.

Details

Apache Axis provides a SOAP implementation, services can be configured in various ways. In case of QRadar the services are configured in the server-config.wsdd file, located under WEB-INF. Three service classes are currently configured:

  • AdminService
  • Version
  • configservices

The first two are distributed with Axis, the latter one is custom for QRadar. The AdminService allows for deploying and undeploying of webservers, however it is configured to only be accessible from localhost.

The implementation of the configservices webservice can be found in the class com.q1labs.configservices.core.ConfigurationServices. Any public method in this class can be called through Axis. The webservice is mapped to the path /console/services/configservices. There are two ways to call these methods:

  • POST request containing a SOAP envelope. The first tag in the SOAP body should have the same name as the method that needs to be invoked. Method parameters are provided as child elements within this tag.
  • GET request; the URL parameters are converted into a SOAP message by Axis. The method is provided via the method URL parameter, its arguments are provided as URL parameter with the same name as the method argument.

No measures have been implemented in Axis and/or QRadar to prevent Cross-Site Request Forgery attacks against the webservices exposed by Axis. Due to this it is possible for an attacker to call any exposed service via Cross-Site Request Forgery. Methods that can be called include:

  • saveScannerConfigFile(data)
  • addManagedHost(ipAddress, password, isTunneled, isCompressed, ipPublicAddress, natId, consoleIpToUse, runPrecheck, remappingip)
  • startComponent/stopComponent/restartComponent(host, type, componentName)
  • startComponents/stopComponents/restartComponents(host)
  • startSystem/stopSystem/restartSystem()
  • injectDeploymentModel/saveDeploymentModelToStaging(deploymentModel)
  • deployStagingConfiguration(fullDeploy)
  • restoreFromBackupDeployment()
  • saveFile(fileName, data)

An attacker can create a URL that when visited by a logged in target executes one or more of the exposed methods, for example:

https:///console/services/configservices?method=stopSystem

Besides the lack of Cross-Site Request Forgery protection, most methods also lack proper access control checks. A handful of these methods perform some form of access control, but most methods can be called by any authenticated users. For example the saveFile() and retrieveFile() methods check if the logged on user has permission to write or access the requested file.

*com.q1labs.configservices.core.ConfigurationServices:*

public byte[] retrieveFile(String fileName, boolean staging) throws ConfigServicesFault {
	try {
		if (!fileName.contains("/") && !fileName.contains("\\")) {
		String qradarUsername = this.requestSession.getQradarUsername();
		**boolean canAccess = RequestSession.hasPermission(qradarUsername);**
		if (!canAccess) {
			throw new UnauthorizedException("Provided username from security token does not have permission to use this method");
[...]

A logged on attacker can call all methods without proper access control. One notable attack is to call the getNvaProperty() method. By calling this method it is possible to retrieve any 'NVA' configuration setting - including sensitive information like (encrypted) credentials. Credentials are stored encrypted on disk, in some case they are stored decrypted in memory. If the are already decrypted, getNvaProperty() will return the plaintext value. If they are encrypted they can easily be decrypted by calling the getDecrypted() method of the webservice.

Some passwords are reused for different services, which allows users to elevate their own privileges. For example, the property jpa.connection.password is used for connecting to PostgreSQL, but is also used as the password for the ConfigServices account. A low privileged user can request the password for the PostgreSQL user. Since PostgreSQL is normally not exposed over the network it would still not be possible to log in. However due to the password reuse it is possible to use the same password to login as the ConfigService user.

https:///console/services/configservices?method=getNvaProperty&key=jpa.connection.password

Questions or feedback?