Skip to content

Getting started

The ZDM library is central to the IoT development with Zerynth and still it is very simple to use. The simplest usage requires just 3 lines of code:

from zdm import zdm

agent = zdm.Agent()

The above snippet shows the steps required to enable a ZDM connection:

  • import the ZDM module
  • create an instance of the Agent class
  • start the agent


Under the hood a MQTT client is started and a secure connection is established using the device certificates contained in the secure element of every Zerynth hardware.

In order for the connection to be established correctly it is necessary that the physical device has been associated with a ZDM device. This procedure links the physical device identity contained in the secure element to its cloud identity represented by the ZDM device unique identifier. The procedure must be performed just once and can be done both from the command line and VSCode extension.

After the connection is established, the Agent will update the real time clock to the current time and will try to keep it synchronized. The Agent will also take care of the reliability of the connection by automatically retrying to connect in case of network failure.

Upon successful connection, the Agent also sends some information on the physical device, like the firmware version and the device type.


Once connected the Agent can handle remote requests for jobs. A job is a task that is performed by a function in the firmware. It is possible to configure the Agent to call a specific function by assigning it to the job name. For example:

def job1(agent, arguments):
    return "OK"

def job2(agent, arguments):
    raise IOError

agent = zdm.Agent(jobs={"myjob":job1, "another_job":job2})
The above snippet configures the Agent to call function job1 when the myjob job is triggered, and job2 for another_job. Notice that the functions have the same signature: the first argument is the instance of the agent, the second is a dictionary containing the arguments of the job.

Jobs are executed by the Agent thread and their result is sent to the cloud for reference. If a job fails with an exception, the error is also sent to the cloud for reference.


The Agent is capable of sending raw data points such as sensor readings to the cloud where they are temporarily stored and fed into the various available integrations (i.e. sent to Azure or to the Zerynth Storage for further elaboration). Data is sent in JSON format as a dictionary called payload thus allowing for the maximum flexibility. The payload is also associated to a tag, a label that can be used to better organize IoT data and to easily retrieve it from the cloud.

Sending data is very easy:

agent = zdm.Agent()

agent.publish(payload={"temperature": 25.3, "humidity": 35.2}, tag="room")

Firmware updates

The Agent is capable of accepting a FOTA update request and changing the current firmware with a new one. The FOTA process can be customized by adding checks at different steps. The customization point are: - version check: the firmware can check the new version and decide if the FOTA should proceed or not. For example, a policy that prevents switching to a version older than the current can be easily implemented with this check. - new firmware acceptance: the new firmware must be accepted as valid to be run at the next reboot. This is a very critical step since accepting a bad firmware could make the device incapable of performing further updated. By default a new firmware is accepted when it is capable of reconnecting to the ZDM, receiving a job request and confirming it. In some use cases these checks can be not enough and some custom logic may be need. For example a diagnostic function can be run to check that the new firmware is still capable of reading all the sensors and controlling the actuators

Configuring the Agent for customized FOTA is easy:

def fota_checks(agent, step, arg):
    if step=="check_version":
        # don't care about version, any version is ok
        return False
    elif step=="accept":
        # do not accept any firmware
        return "sorry, firmware not accepted"

agent = zdm.Agent(on_fota=fota_checks)
In the snippet above, a function is provided to the on_fota parameter. This function will be called multiple times at different steps in the FOTA process. When fota_checks returns None or something False the step is considered valid, whereas when a non truthy value is returned the step is canceled and the return valued is sent to the cloud as the reason for the FOTA failure.

ZDM Agent

class Agent

class Agent(cfg=None, jobs={}, set_clock_every=300, on_fota=None, host="")

Create an Agent instance. The Agent class accepts various parameters:

  • cfg is an instance of the Config class detailing the transport connection parameters. It is set to None by default using standard parameters.
  • jobs is a dictionary that defines the ZDM jobs the agent can handle. The keys of the dictionary are strings representing the name of the jobs and the values are functions that are called each time a job is triggered. When set to empty dictionary {} the only jobs that can be triggered are reset and fota.
  • set_clock_every is the number of seconds after which re-synchronizing the real time clock. If set to zero or negative, clock synchronization is disabled and is up to the firmware to find alternative ways of setting the current time (i.e. by ntp protocol)
  • on_fota is a function accepting one ore more arguments that will be called at different steps of the FOTA process to validate or refuse the step.
  • host is the hostname of the ZDM broker, do not change.

method start


Start the Agent. Since the connection time can be considerable depending on the network status, the method returns either upon successful connection or after a certain amount of time whatever comes first. The amount of time to wait for a working agent can be set in seconds with the wait_working parameter.

method online


Return True if the following conditions are all satisfied: - the Agent is started - the MQTT connection has been established - the MQTT client has subscribed to all required topics - the Agent has received the current time

method connected


Return True if the MQTT connection has been established and the MQTT client has subscribed to all required topics.

method has_time


Return True if the Agent has received the current time.

method firmware


Return the current firmware version. Version strings have a specific format: - an optional prefix terminated with a dash - the firmware id - a colon - the firmware id version

When a firmware is loaded into a device manually via the toolchain, the version string is set to "Factory". The firmware id is determined by the ZDM when a new firmware is created while the firmware version can be decided by the user that loads the new version. By default, the new version is equal to the previous one incremented by one.

When the a new firmware is launched and it has not been validate yet, the version is prefixed with "testing-"

method publish

publish(payload, tag="default", timeout=0)

Send data contained in payload labeling it with tag. This method calls into the underlying MQTT client that is configured by default with a qos of one. The timeout specifies the time in milliseconds to wait for an acknowledgement form the broker when qos > 0; when qos = 0, the timeout is ignored. If the acknowledgement is not received by the timeout, the MQTTPublishTimeout exception is raised.

method reset


Reset the device gracefully by disconnecting the MQTT client, closing network interfaces (i.e. deassociating from WiFi access point) and finally resetting the device. It is used mainly during the FOTA process to start the new firmware but it can also be called externally when there is a need for a clean reset.

Agent configuration

class Config

Creates a Config instance that set parameters for the connection to the MQTT broker.

In particular, the following parameters are available:

  • keepalive is the MQTT keepalive timeout, in seconds
  • reconnect_after is the time after which retrying to connect upon error, in milliseconds.
  • network_timeout is the timeout in milliseconds for network operations like the TCP connect. If no answer is received after the timeout, the operation is aborted.
  • clean_session is the MQTT clean session flag
  • qos_publish is the quality of service when publishing to the MQTT broker
  • qos_subscribe isthe quality of service when subscribing to the MQTT broker