This page provides an overview of how to secure a Camunda installation. For Camunda’s security policy, a list of security notices and a guide how to report vulnerabilities, please visit the general security documentation.
In order to secure a Camunda installation, Camunda itself must be configured correctly and it must be integrated correctly into its environment. This section also identifies areas where we consider security issues to be relevant for the specific Camunda Platform product and listed those in the subsequent sections. Compliance for those areas is ensured based on common industry best practices and influenced by security requirements of standards like OWASP Top 10 and others
Deployment Options and Components
There are different ways of using Camunda Platform and different components are provided: the process engine itself, the REST API, the web applications. Depending on how Camunda is deployed and which components are used, different security considerations apply. The following list gives a general overview over deployment options and components outlining the main differences from a security point of view. The remainder of this chapter elaborates on the different configuration options.
- Embedded Java library inside an application: in this case, the Camunda engine is embedded inside a custom Java Application. Usually the application takes care of securing access to Camunda’s APIs and the APIs are not directly exposed to an end user. In this case, the application typically takes care of ensuring authentication and preventing access by unauthorized users.
- Shared Process Engine: in this scenario, the Process Engine is deployed as a container service into an application server such that it can be used by the applications deployed into the same container / server. This case is similar to the embedded Java library case.
- REST API: the REST API provides access to Camunda’s core APIs through HTTP. In this case users can directly access Camunda’s APIs. Usually, it is necessary to configure authentication, authorization and also secure the connection to the REST API using SSL (HTTPS).
- Web applications (Cockpit, Tasklist, …): similar considerations to the REST API apply.
Keep in mind that it is not recommended to use the pre-packaged distribution in production environment rather install the full distribution manually (for example Tomcat manual installation).
The pre-packaged distribution is intended for users who want a getting started experience. In case you still want to use it in production, consider un-deploying the invoice application and removing the demo user.
Security Configuration inside Camunda
Camunda provides a number of configuration options which are relevant from a security perspective. Most prominently: authentication, authorization and the control of custom code (scripts) which can be executed on the server.
Authentication controls who can access Camunda’s APIs and Applications.
Do I need Authentication?
Authentication is only needed in the following cases:
- Camunda’s REST API is used
- Camunda’s web applications are used
In these cases, direct access to Camunda’s core APIs is provided over HTTP and authentication must be enabled.
By contrast, authentication is generally not done by Camunda when embedded as a library into an application. In this case the application takes care of authentication itself.
Enabling Authentication for the REST API
For ease of use by developers, the REST API’s authentication is disabled by default. When deploying the REST API in production, it is therefore required to enable authentication. Check the corresponding section in the REST API documentation.
Authentication in the Web Applications
For the web applications, authentication is enabled by default and it is not possible to disable it.
Internal (database backed) User Management
To perform authentication, Camunda can use two sources: a database or LDAP.
When using the the database, usernames and passwords are stored inside the
ACT_ID_USER table (see documentation on database schema). To protect the passwords stored in the database, Camunda uses two concepts:
- hashing: instead of storing the password in plain text, a hash is stored. When authenticating, the same hash is generated from the user’s input and compared against the hash in the database. If both hashes are equal the authentication attempt is successful. Camunda allows users to configure and customize the hash function used. Please refer the documentation section on password hashing for details.
- salted hashes to protect the database against rainbow table attacks, Camunda uses salted hashes. Similar to hashing itself, this function can be configured and extended to a user’s needs. Please refer the documentation section on password hashing for details.
As an alternative to the database, Camunda can use LDAP for verifying user credentials on authentication. Camunda has read-only access to LDAP.
Authorization controls what data a user can access and change in Camunda once authenticated. Authentication is a pre-requisite to authorization.
Do I need to enable Authorizations?
Similar considerations as for authentication apply. For an in-depth discussion, see the documentation section on authorizations
Restricting Data Access with Authorizations
Authorizations can be used to restrict a user from accessing a data object (such as a process or a task) and can be used to restrict how the user can interact with such data objects (read-only vs. modifications). Authorizations in Camunda are very powerful and it is recommended to read the corresponding documentation entry on authorizations.
Prevent enumerating user accounts by brute-force creating new users
Under certain circumstances, an attacker can enumerate user accounts by brute-force creating new users:
You don’t centrally manage user accounts (e.g., with the help of LDAP or a custom implementation of
WritableIdentityProvider) but instead …
- … use an account “self-service” approach: Unauthenticated users are allowed to create accounts; i.e., you have implemented a custom REST endpoint which can create users without authentication
- … an authenticated user has
USERresource to create new user accounts and an untrusted person has access to this account (please see the Authorization Service docs to learn how permissions are granted to resources)
As soon as the attacker has obtained information about existing user ids, they can put all their efforts on cracking passwords.
We strongly recommend you to use the product with centrally managed user accounts. It is certainly not advisable to manage accounts via the ways mentioned above.
We think that the before mentioned scenarios are uncommon for organizations using the Camunda Platform Runtime. However, we want to inform you about the options to prevent unrecommended usage, which makes the product vulnerable to attacks.
Deployments to the process engine can contain resources that are interpreted like code:
- BPMN, DMN, CMMN models that the process engine executes on the Camunda server
- Java EL expressions that BPMN, DMN, CMMN models include and that are executed on the Camunda server
- Forms that a client application like Camunda Tasklist renders in the browser of the end user
Camunda does not provide a safe sandbox environment for the execution and rendering of these resources. Attackers that are able to make deployments can effectively perform remote code execution in the Camunda system. It is therefore critical that only trusted users and systems can make deployments.
For example, you can restrict deployment access in the following ways:
- Using authorizations, administrators grant the
CREATEpermission on the
Deploymentresource only to trusted users
- An application that embeds the Camunda Java API can choose to not expose the deployment API on untrusted channels (such as to HTTP requests)
- System administrators ensure that only trusted users have network access to the Camunda installation
See also the user guide section Custom Code & Security for further information.
Throttle login attempts
The engine gives option to throttle login attempts. The mechanism behind this is enabled by default. You can read more about it under Identity Service in User Guide.
Custom Whitelist for User, Group and Tenant IDs
To determine if the provided ID is acceptable or not, IDs can be matched against a Whitelist Pattern. You can read more about it under Identity Service in User Guide.
For users that are managed within the engine (i.e. not LDAP-managed users) it is possible to specify a password policy to ensure that all user passwords meet a certain security standard. While choosing a strong password policy will make the users choose better passwords it can annoy them if the policy is too strict. Since version 7.11 a default policy is available that requires passwords to follow only a few rules. However, a much higher level of security can be achieved by using a more sophisticated password policy. (e.g. by password topology blacklisting, also see OWASP guide on password complexity)
If you consider adding your own password policy you can find more information about how password policies in the engine work in our Password Policy user guide.
Camunda offers different types of forms which are primarily used in Tasklist. In the input inside of this forms you can call and execute scripts which allows you to achieve easily your business logic. Please validate this input each time to prevent malicious behaviour.
Expressions in Queries
Consider disabling execution of expressions in queries. See also: Custom Code & Security
One of the options to query data from the engine is using native queries. Which means to provide own SQL queries to retrieve engine entities if the Query API lacks the possibilities you need. However, use the native queries with care. Please bear in mind of the SQL Injection while using this approach.
Maximum Results Limit in Queries
Using the REST API or the Webapps querying for results without restricting the maximum number of results or querying for a vast number of results can lead to a high memory consumption or even to out of memory exceptions.
You can mitigate the risk of an attack by defining a limit for the maximum number of results
queryMaxResultsLimit) in the process engine configuration.
To gain the full feature set of the Webapps, and not suffer any UX degradation due to unavailable data, the
queryMaxResultsLimit must be set to
Please see the User Guide to learn more about the Query Maximum Results Limit.
CSRF Prevention in the Webapps
A CSRF filter is enabled by default, validating each modifying request performed through the webapps. Please also see the detailed overview on how to configure CSRF Prevention.
The CSRF Prevention makes use of a cookie. By default, some security-related configurations are present for this cookie. To ensure full security, please consult the documentation about Cookie Security to learn more about it.
Camunda handles many XML files containing configurations of process engines, definitions of process models and more. In order to mitigate possible vulnerabilities that can be introduced by XML files, the following measures are activated by default:
- Prevention against XML eXternal Entity (XXE) injections according to OWASP
- Feature Secure Processing (FSP) of XML files according to Oracle which introduces limits for several XML properties
If the limitations on XML files introduced by XXE prevention need to be removed, XXE processing can be enabled via
enableXxeProcessing in the process engine configuration.
FSP itself can not be disabled in the engine. All properties that are influenced by this can however be configured in the environment via system properties and the
jaxp.properties file. See the Oracle documentation on how to determine the right limits and how to set them.
Since BPMN schema validation requires external XSD documents, the property
http://javax.xml.XMLConstants/property/accessExternalSchema is by default configured to value
all, which enables referencing XML schemas by any supported protocol. This can be overridden via the system property
javax.xml.accessExternalSchema, however a value set via
jaxp.properties does not take effect.
HTTP Header Security in Webapps
Out-of-the-box the web applications provide the following security-related HTTP headers:
- XSS Protection
- Content Security Policy
- Content-Type Options
These headers enable browser-side security mechanisms which help to improve the protection against several attacking scenarios.
According to your project requirements, some of these headers can be configured more strict or lax. Please see the documentation about the HTTP Header Security to learn more about the several headers, the defaults and how to configure the HTTP headers according to your needs.
Variable Values from Untrusted Sources
Process variables can be submitted as Java objects using the JDK built-in
application/x-java-serialized-object data format, JSON or XML along with a Java class name via the Camunda REST API and web applications.
On server side, they can then be deserialized into Java objects, so that Java code can work with them in a native way. See Camunda Spin for details and this REST API endpoint for an example.
If an attacker can access these endpoints, they can exploit so-called serialization gadgets, i.e. classes that run vulnerable code during deserialization resulting in remote code execution in the general case. For example, consider a class constructor that makes a REST request based on a field value. An attacker could submit a forged variable value so that during deserialization, when the constructor is called, the application server would make an arbitrary REST request to a destination of the attacker’s choice. For details, see OWASP’s description of Deserialization of untrusted data.
Java objects using the JDK built-in
application/x-java-serialized-object data format
Starting with version 7.9, by default, it is not possible to set variables of type
Object AND the data format
The behavior can be restored with the process engine configuration flag
However, please bear in mind that enabling the java serialization format might make the process engine vulnerable against the aforementioned attacking scenario.
JSON/XML serialized objects using Spin
Therefore, we recommend enabling the whitelisting of allowed Java classes by enabling the property deserializationTypeValidationEnabled in the process engine configuration. With this, the process engine validates the class names of submitted variables against a whitelist of allowed Java class and package names. Any non-whitelisted content is rejected. The default values are safe, but may be too restrictive for your use case. You can use the engine properties
deserializationAllowedClasses to extend the default whitelist with package and class names of Java types that you consider save to deserialize in your environment.
In case this default behavior needs further adjustment, a custom validator can be implemented and registered in the engine with the engine property
The provided object needs to be a subtype of
org.camunda.bpm.engine.runtime.DeserializationTypeValidator and offer an implementation of the
In case you want to rely on allowed package and class names from the engine configuration as well, you can provide a subtype of
An implementation of this interface registered as validator will be provided with the defined packages and classes from the engine configuration upon initialization of the engine via
Jackson Type Whitelisting
Spin’s JSON implementation is based on Jackson. If you configure Camunda Spin to deserialize polymorphic classes based on type information included in the JSON itself (i.e. where the JSON contains explicit class names), we strongly recommend to additionally enable Jackson’s Whitelisting feature starting with version 2.10. Camunda’s whitelisting feature does not cover this case.
Security Configuration in the external Environment
Camunda integrates into an environment, most prominently the database and, when using the web applications or the REST API, also a webserver. In order to secure your Camunda deployment as a whole, the integration is relevant.
Camunda stores its data into a relational database. In order to protect access to this data, it must be configured correctly. The documentation section on supported environments provides a list of supported databases.
To prevent unauthorized access to the data stored in the Camunda database you must follow best practices around operating the database, including data encryption. Please make sure to consult the manual provided by your database vendor.
Securing the database connection
To access the database, Camunda needs to establish a connection. Usually the connection is configured either directly through the JDBC configuration options or through a datasource configured inside the application server. Most database drivers support encrypted connections and transport layer security when connecting to the database. When operating Camunda and the database in an untrusted network it is recommended to enable these features. Please consider the manuals of your database, database driver and your application server to do so.
Securing Database Credentials
To establish the connection to the database, the database credentials need to be provided. As opposed to providing the credentials as plain text in a configuration file, some application servers support storing the credentials securely in an encrypted form. In that case, consult the manual of your application server to learn how to use these features.
Web Server (applicable when using REST API or Web Applications)
When deploying the REST API or the Camunda web applications, Camunda is integrated with a third party web server. The documentation section on supported environments provides a list of supported web servers / application servers. It is strongly recommended to consider applying the following configurations.
Enabling SSL / HTTPS
Configure SSL / HTTPS when deploying the Camunda REST APIs or web applications. This can be achieved by configuring HTTPS either on the web server itself or through a reverse proxy. Consider disable HTTP and configure HTTPS only for your web applications. Please consult the manual of your web server or reverse proxy for details.
Setting up the session timeout is usually done via
web.xml deployment descriptor. Please consult the Java Servlet specification or manual of your application server.
The session cookies domain is configured in web server specific configuration. If you want to set such kind of cookies please consult the manual of your web server for details, e.g. for Tomcat check this docs.
Maximum POST size in server (REST API)
Restriction of the maximum size in bytes of the POST requests is specific to your web server. Please consult the manual of your web server for details, e.g. for Tomcat server, check this documentation page.
Securing Cookies (Web Applications)
The container provides the session cookie. Please consult the documentation about Cookie Security to learn what configurations are necessary to ensure full security for the session cookie.
When it comes to error handling, from a security perspective, the top goal is to prevent attackers from obtaining technical details about the system, which for example, a stack trace could reveal (see OWASP’s Improper Error Handling article for more information).
In this section, we describe what we do to prevent disclosing technical details about the system.
In the Webapps, a generic default error page is provided when an error occurs. This error page does not include any technical details like stack traces.
The REST API only displays the type and a short error message when an error is thrown.
Prevent Disclosure of Application Server Internals
In the Error handling paragraph, we explain our technical measures not to disclose any technical details about the Camunda Platform Runtime.
However, technical details cannot only be disclosed on the application level, but also by the application server itself. Therefore, it is recommended to configure the application server in a way that no technical details are disclosed when an error occurs on the application server level.
The exact configuration and the defaults differ among application servers.
Please find below external documentation on how to configure your application server correctly:
- Tomcat 7.0+
- Wildfly 10.1+: Official Documentation
- JBoss EAP 7.0+: Official Documentation
- JBoss EAP 6.4
- Official Documentation
- Alternative Resources
- Web Subsystem: Disable stacktrace in response body (JBoss AS 7.1)
- Camunda Run/Spring Boot 2.2+