Security Features of HiveMQ

HiveMQ is designed from the ground up with maximum security in mind. For mission-critical IoT and M2M scenarios, secure end-to-end encrypted communication and advanced authentication and authorization features are essential. HiveMQ gives you the flexibility to enable the specific security features that your deployment requires.

If you are unfamiliar with MQTT security concepts, see the MQTT Security Fundamentals blog series.

Authentication & Authorization in HiveMQ

HiveMQ handles authentication and authorization through security extensions.
For example, the HiveMQ Enterprise Security Extension ships as part of the HiveMQ Enterprise Platform bundle. For configuration details, see the HiveMQ Security Extension documentation.
You can also download the HiveMQ File RBAC community extension from the HiveMQ website or use the open HiveMQ Extension SDK to develop your own security extension.

HiveMQ includes a hivemq-allow-all-extension for testing purposes. This extension authorizes all MQTT clients to connect. Before you use HiveMQ in production, you must add an appropriate security extension and remove the hivemq-allow-all-extension.

TLS

Transport Layer Security (TLS) is a cryptographic protocol that allows secure and encrypted communication at the transport layer between a client application and a server. If you enable a TLS listener in HiveMQ, each client connection for that listener is encrypted and secured by TLS.

Multiple listeners
You can configure HiveMQ with multiple listeners so HiveMQ can handle secure and insecure connections simultaneously. For more information, see HiveMQ MQTT Listeners.

For deployments where MQTT messages contain sensitive information, we strongly recommend that you enable TLS. When configured correctly, TLS makes it extremely difficult for attackers to break the encryption and read packets on the wire. TLS encrypts the complete transport layer, which makes TLS a better choice than custom payload encryption when security is the priority.

TLS Overhead Considerations

TLS adds CPU and communication overhead. The TLS handshake adds bandwidth and computation overhead when a connection is established. The additional CPU usage is typically negligible on the broker, but can affect constrained devices that are not designed for computation-intensive tasks. If your deployment uses unreliable connections that frequently drop, consider the increased overhead. For more information, see TLS/SSL - MQTT Security Fundamentals.

Encryption at Transport Layer vs Encryption at Application Layer

Transport layer encryption encrypts the complete connection, including all MQTT messages sent between the client and the server. This ensures that only the connected client can read any message in the communication. Since the payload of the MQTT message remains as raw bytes, full interoperability with other MQTT clients is ensured (including clients that do not use TLS.) All MQTT messages are secured with this method, not only PUBLISH messages.

Application layer encryption encrypts the payload of an MQTT PUBLISH message with application-specific encryption. Only clients that know how to decrypt the payload can read the original message payload. Without TLS, the transport is unencrypted and attackers can read raw messages on the wire. However, if the attacker cannot decrypt the payload, the PUBLISH payload remains secure. NOTE: Application layer encryption only protects PUBLISH payloads. Other information such as topic names remains unencrypted, and other MQTT messages such as CONNECT messages cannot be secured with this method.

You can combine both encryption methods. If only a few trusted clients need to decrypt specific PUBLISH payloads and you also need to secure all communication, use both methods together.

Java Key Stores and Trust Stores

Java key stores and Java trust stores are containers for SSL information such as X.509 certificates and keys. Each store is typically persisted in a single file and protected with a master password.

Key stores and trust stores are conceptually similar but serve different purposes:

  • Key stores provide credentials. A key store contains a public key certificate and the corresponding private key. HiveMQ uses key stores to protect the private key for SSL connections.

  • Trust stores verify credentials. A trust store contains trusted certificates or certificates signed by a Certificate Authority (CA). Clients that connect to HiveMQ store the server certificate (or the CA certificate if the server certificate is signed by a CA) to identify the server as trusted.

If you are unfamiliar with private and public key cryptography, review this topic before you configure SSL.

It is possible to use the same file for the key store and trust store. However, to ensure the security of the private key, we strongly recommend that you use separate files.

For information on how to create a key store, see HiveMQ Platform How-Tos.

Autoreload
HiveMQ reloads key and trust stores during runtime. You can add or remove client certificates from the trust store or change the server certificate in the key store without any downtime. If the same master password is used, you can replace the key store and trust store files without downtime.

Communication Protocol

When no explicit SSL/TLS version is set, HiveMQ automatically uses one of the two default-enabled protocols based on client support. TLSv1.2 or TLSv1.3 are recommended because these protocols tend to be more secure.

When no explicit TLS version is set, HiveMQ uses TLSv1.2 or TLSv1.3 by default, based on the version the client supports.

The default tls-tcp-listener configuration of HiveMQ enables the following TLS protocols by default:

HiveMQ default TLS protocols
TLSv1.3
TLSv1.2
Due to security concerns, the OpenJDK Java platform no longer enables TLSv1 and TLSv1.1 by default. Java applications that use TLS, including HiveMQ, now require TLS 1.2 or later. The change applies to OpenJDK 8u292 onward, OpenJDK 11.0.11 later, and all versions of OpenJDK 16. TLSv1 and TLSv1.1 are not removed from OpenJDK, only the default availability changes.

If you need to support TLSv1 or TLSv1.1, you must explicitly enable them in the TLS version configuration of your HiveMQ listeners (see example explicit HiveMQ TLS configuration).

To enable only specific protocols, you can use an explicit TLS configuration that is similar to the following example. If necessary, you can also use such an explicit configuration to enable legacy protocols such as TLSv1 and TLSv1.1:

Example explicit TLS version configuration
<?xml version="1.0"?>
<hivemq xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">

    ...
    <listeners>
        ...
        <tls-tcp-listener>
            <tls>
                ...
                <!-- Enable specific TLS versions manually -->
                <protocols>
                    <protocol>TLSv1.2</protocol>
                </protocols>
                ...
            </tls>
        </tls-tcp-listener>
    </listeners>
    ...
</hivemq>

Cipher Suites

The security of TLS depends on the cipher suites in use. Usually, JVM vendors enable only secure cipher suites by default. If you need to restrict HiveMQ to specific cipher suites, you can configure them explicitly.

HiveMQ enables the following cipher suites by default:

Default cipher suites
TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384
TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256
TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256
TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA
TLS_ECDHE_RSA_WITH_AES_256_CBC_SHA
AES256 requires JCE unlimited strength jurisdiction policy files.
TLS_RSA cipher suites are disabled by default in Java 21.0.10 and later versions due to lack of forward secrecy.
Use ECDHE cipher suites for secure connections.

If none of the default cipher suites are supported, the cipher suites that your JVM enables are used.

The list of cipher suites that are enabled by default can change with any HiveMQ release. If you depend on specific cipher suites, specify the cipher suites explicitly.
Example configuration to set cipher suites for listeners explicitly
<?xml version="1.0"?>
<hivemq xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">

    ...
    <tls>
        ...
        <!-- Only allow specific cipher suites -->
        <cipher-suites>
            <cipher-suite>TLS_RSA_WITH_AES_128_CBC_SHA</cipher-suite>
            <cipher-suite>TLS_RSA_WITH_AES_256_CBC_SHA256</cipher-suite>
            <cipher-suite>SSL_RSA_WITH_3DES_EDE_CBC_SHA</cipher-suite>
        </cipher-suites>
        ...
    </tls>
    ...
</hivemq>

Each TLS listener can be configured to have its own list of enabled cipher suites.

Native SSL

HiveMQ comes prepackaged with an OpenSSL implementation called BoringSSL that is maintained by Google and can be activated on Linux or macOS.

The main advantage of native SSL is increased performance compared to standard JVM SSL. Native SSL also provides access to additional cipher suites, including:

  • Stronger AES with GCM

  • The ChaCha20 stream cipher

  • Additional cipher suites with elliptic curve algorithms

Limitations:

  • Native SSL is not available on all platforms. If native SSL is not supported on your platform, HiveMQ performs a graceful fallback to the SSL implementation of your JVM.

  • Cluster transport TLS connections cannot use the native SSL implementation.

  • If native SSL is enabled, you cannot disable the SSLv2Hello communication protocol.

  • If you configure cipher suites that are available in OpenSSL but not in JVM SSL, the broker may have no matching cipher suites for any client, and connections cannot be established.

To enable HiveMQ Native SSL, use a configuration similar to the following:

Example Native SSL configuration
<?xml version="1.0"?>
<hivemq xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">

    ...
    <listeners>
        ...
        <tls-tcp-listener>
            <tls>
                ...
                <native-ssl>true</native-ssl>
                ...
            </tls>
        </tls-tcp-listener>
    </listeners>
    ...
</hivemq>

Due to security concerns and to align with the OpenJDK Java Platform, from HiveMQ 4.7 onwards, HiveMQ only enables the following TLS protocols by default for native SSL:

  • TLSv1.3

  • TLSv1.2

If you need to support legacy TLS versions such as TLSv1 or TLSv1.1 for your Native SSL implementation, explicitly enable the versions in your tls-tcp-listener configuration:

Example native SSL configuration with explicit TLS version
<?xml version="1.0"?>
<hivemq xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">

    ...
    <listeners>
        ...
        <tls-tcp-listener>
            <tls>
                ...
                <!-- Enable legacy TLS versions manually -->
                <protocols>
                    <protocol>TLSv1.1</protocol>
                </protocols>
                <native-ssl>true</native-ssl>
                ...
            </tls>
        </tls-tcp-listener>
    </listeners>
    ...
</hivemq>

Randomness

If it is available, HiveMQ uses /dev/urandom as the default source of cryptographically secure randomness. /dev/urandom is considered secure enough for almost all purposes [1] and has a significantly better performance than /dev/random.

If desired, you can revert to /dev/random for your random number generation:

  • Delete the line that starts with the following information from your $HIVEMQ_HOME/bin/run.sh file if you start HiveMQ manually or the -Djava.security.egd=file:/dev/./urandom option from the configuration file of the init service of your choice.

     	    JAVA_OPTS="$JAVA_OPTS -Djava.security.egd=file:/dev/./urandom"

OCSP stapling

The Online Certificate Status Protocol (OCSP) determines the revocation status of an SSL certificate. OCSP is frequently used as an alternative to Certificate Revocation Lists (CRL) because OCSP contains less information and requires less network traffic. The smaller data payload enables more lightweight clients.

In client-driven OCSP, each client requests certificate status directly from the OCSP responder. When many clients use client-driven OCSP, the volume of requests can cause the OCSP responder to become a performance bottleneck.

OCSP-without-stapling
Figure 1. A schematic overview of a client-driven OCSP request for SSL certificate status to an OCSP Responder

OCSP stapling allows the HiveMQ broker, rather than the client, to make the status request to the OCSP responder. The HiveMQ broker regularly obtains an OCSP response about its own certificate from the OCSP responder, caches the response, and sends it directly to the client in the initial TLS handshake. The client does not need to connect to the OCSP responder directly.

OCSP stapling significantly reduces the load on the OCSP responder because a single request per validity period replaces a request per individual client.

OCSP-stapling
Figure 2. A schematic overview of OCSP stapling with the HiveMQ broker

The caching interval defines how frequently the HiveMQ broker sends requests for new status information. Between requests, the HiveMQ broker caches the last status that was received.
If the OCSP responder is not available, HiveMQ temporarily reduces the cache interval to 15 seconds to get status information as soon as possible.
Once a successful OCSP response is received, the interval automatically reverts to the configured value.
If the HiveMQ broker does not receive a valid response within 30 minutes, the cached response is cleaned up and no OCSP response is sent to the client. However, HiveMQ continues to try to establish a connection with the OCSP responder.

HiveMQ initiates requests for status information to the OCSP responder in the following cases:

  • When a TLS listener starts

  • When the configured cache interval expires

  • When a client requires status information and the response is not yet cached

  • If the cached response expires and a new TLS connection is established

OCSP Stapling Configuration Properties

The <ocsp-stapling> element has the following properties:

Name Default Mandatory Description

enabled

false

no

Enables OCSP stapling.

override-url

none

no

Overrides the URL of the OCSP Responder contained in the server certificate. An override URL must be set if no OCSP URL information is included in the server certificate.

cache-interval

3600

no

Interval in seconds to cache the OCSP response on the server side from the OCSP stapling responder.

OCSP stapling configuration

The following configuration enables OCSP stapling for a TLS TCP listener:

<?xml version="1.0"?>
<hivemq xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">

    <listeners>
        ...
        <tls-tcp-listener>
            <port>8883</port>
            <bind-address>0.0.0.0</bind-address>
            <tls>
                <keystore>
                    <path>/path/to/the/key/store.jks</path>
                    <password>password-keystore</password>
                    <private-key-password>password-key</private-key-password>
                </keystore>
                <native-ssl>true</native-ssl>
                <ocsp-stapling>
                    <enabled>true</enabled>
                </ocsp-stapling>
            </tls>
        </tls-tcp-listener>
        ...
    </listeners>
</hivemq>
Preconditions
OCSP stapling is disabled by default. To use OCSP stapling you must set <native-ssl> and <ocsp-stapling><enabled></ocsp-stapling> to true.

The following configuration enables OCSP stapling for a TLS TCP listener with a custom cache-interval and override-url:

<?xml version="1.0"?>
<hivemq xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">

    <listeners>
        ...
        <tls-tcp-listener>
            <port>8883</port>
            <bind-address>0.0.0.0</bind-address>
            <tls>
                <keystore>
                    <path>/path/to/the/key/store.jks</path>
                    <password>password-keystore</password>
                    <private-key-password>password-key</private-key-password>
                </keystore>
                <native-ssl>true</native-ssl>
                <ocsp-stapling>
                    <enabled>true</enabled>
                    <override-url>http://your.ocsp-responder.com:2560</override-url>
                    <cache-interval>3600</cache-interval>
                </ocsp-stapling>
            </tls>
        </tls-tcp-listener>
        ...
    </listeners>
</hivemq>

HiveMQ Audit Log

The audit log provides a unified record of all auditing-relevant events. The audit log can be used in several ways:

  • Review all actions performed on the HiveMQ cluster.

  • Satisfy legal and compliance requirements.

  • Provide data for intrusion-prevention software.

  • Track which users accessed which information and when.

Audit Log Configuration

The audit log is enabled by default. It can be disabled in the HiveMQ configuration file.

HiveMQ Audit Log File

By default, HiveMQ writes the audit log to <HiveMQ Home>/audit/audit.log.

To customize the audit log folder, set the HIVEMQ_AUDIT_FOLDER environment variable or the hivemq.audit.folder system property. For more information, see Manually Setting HiveMQ Folders.

The audit log contains sensitive information. Be sure to set the filesystem permissions of the audit folder accordingly.

The audit log provides automatic log file rolling. Once per day at midnight, HiveMQ archives the old audit log file with the filename audit.<yyyy-MM-dd>.log.

For example, after two days of operation, the audit folder contains the following files:

├─ audit.2019-07-22.log
├─ audit.2019-07-23.log
└─ audit.log
HiveMQ never deletes archived audit log files. If you need to remove old audit logs regularly, you must take additional action. For example, set up a scheduled cron job to alleviate data protection concerns or storage constraints.

Audit Log Statement Format

Audit log statements use the following structure:

<time><time zone> | user:"<user name>" | IP:"<host address>" | node:"<node name>" | source:"<source>" | <event>

Table 1. Log statement arguments
Log argument Description

time

Shows the time when the event happened. The time format is: yyyy-MM-ddTHH:mm:ss,SSS

time zone

Shows the offset of the time zone in which the event happened compared to UTC. The format is: ±HH:mm

user name

The User login that triggered the event.

host address

The IP address from which the user connected. The address can be in IPv4 or IPv6 format.

node name

The identifier of the HiveMQ cluster node on which the event occurred. This name is logged at the start of HiveMQ in the hivemq.log file. For example, "hivemqId": "35yIM".

source

The source of the event that generated the audit log entry. The source can be control-center or rest-api. When available, additional information is provided in square brackets. For example, control-center [Default Login] or rest-api [/api/v1/mqtt/clients].

event

The type of event and additional information for the event. For a list of all events, see Available HiveMQ Audit Log Events.

Available HiveMQ Audit Log Events

The following events are listed in the audit log:

control-center

Table 2. HiveMQ control-center audit log events
Event Additional Information

Login attempt failed

Log in successful

Session timed out

Logged out

Refreshed clients snapshot

Created trace recording

Trace recording name, start time, end time, client filters, topic filters, and packet filters

Stopped trace recording

Trace recording name

Downloaded trace recording

Trace recording name

Deleted trace recording

Trace recording name

Refreshed clients snapshot

Added subscription

Topic filter, QoS, and client ID

Removed subscription

Topic filter, and client ID

Forced client disconnect

With/without will massage and client ID

Forced session delete

Client ID

Created new backup

Downloaded backup file

Backup file name

Deleted backup file

Backup file name

Started import of backup file

Backup file name

Backup aborted

Cleared dropped message statistics

Inspected password

Client ID

Inspected will payload

Client ID

Inspected TLS certificate

Client ID

Inspected proxy protocol TLVs

Client ID

rest-api

Table 3. HiveMQ REST API audit log events
Event Additional Information

Obtained paginated list of all clients

Obtained client details

Client ID

Checked whether the client is connected

Client ID

Forced client disconnect

ClientID

Forced session delete

ClientID

Obtained list of client subscriptions

Client ID

Obtained backup details

Backup ID

Requested create new backup

Backup ID

Requested restore backup

Backup ID

Downloaded backup

Backup ID

Downloaded list of all backups

Requested new diagnostic zip

Zip ID

Downloaded trace recording

Trace Recording ID

Obtained list of all trace recordings

Created trace recording

Trace Recording ID

Deleted trace recording

Trace Recording ID

Stopped trace recording

Trace Recording ID

Started DataHub trial mode

Obtained the FSM state

Client ID

Saved new script

Script ID, Version

Obtained script

Script ID, Version

Obtained filtered list of scripts

Deleted all versions of script

Script ID

Created new schema

Schema ID, Version

Obtained schema

Schema ID, Version

Obtained filtered list of schemas

Deleted all versions of schema

Schema ID

Created new data policy

Data Policy ID

Updated data policy

Data Policy ID

Obtained data policy

Data Policy ID

Obtained filtered list of data policies

Obtained paginated list of data policies

Deleted all versions of data policy

Data Policy ID

Created new behavior policy

Behavior Policy ID

Updated behavior policy

Behavior Policy ID

Obtained behavior policy

Behavior Policy ID

Obtained paginated list of behavior policies

Deleted behavior policy

Behavior Policy ID