Deployments

A deployment is a rapyuta.io resource that represents a unique instantiation of a rapyuta.io package. It holds information about the package deployed, the configuration used and interfaces exposed. It possesses a unique identifier and provides a mechanism to introspect its phase and state that are needed to ascertain the state of a system.

Tooling such as logs, debug terminals and other automation leverage this uniquely identifiable resource to allow the operator to manage, debug and observe a particular running instance of their application.

Deployments may support linking and binding to allow the user to combine multiple different applications to help realize a potentially complex robotics solution.

The deployment object exposes information to help you introspect the state of the system and set policies on mechanisms that attempt recovery of the desired state on degradation.

The following sections let the developer learn more about

Phases

The lifecycle of a deployment consists of multiple phases. The DEPLOYMENT PHASE indicates the current phase of the deployment in its lifecycle.

The below table lists the phases of deployment as they appear in the lifecycle:

Deployment phase Description
In progress accepts request to deploy a package and starts deployment process
Provisioning pulls a docker image and creates a running instance of the image (docker container) for each executable of the component
Succeeded each executable of every component is successfully started
Failed to start error occurred during In progress phase
Partially deprovisioned you deprovisioned a deployment, but there is at least one component that could not be deprovisioned
Deployment stopped you deprovisioned a deployment, and all of its components are stopped

Deployment

Status

rapyuta.io enables you to monitor the current status of each executable of a component that is deployed. The status of deployment depends on the combined status of all components participating in the deployment.

The following table lists the statuses you may see during the Provisioning deployment phase:

Status Description
Pending docker image is being pulled, or docker container is being created
Error error occurs while pulling a docker image or creating a docker container

The following table lists the statuses you may see during the Succeeded deployment phase:

Status Description
Running executables of components are running
Pending restarting executable due to runtime error in the application or rapyuta.io software
Error runtime error occurred
Unknown rapyuta.io is unaware of the current status

If the status of an executable reads Pending or Error, you are provided the cause of the status as Reason.

Reason field

Error Codes

If the overall deployment STATUS is Error, rapyuta.io displays an error code along with a brief description of the error. The following table lists available error codes, short descriptions and the recommendations you should take:

Error code Description Recommended action
DEP_E151 device is either offline or not reachable check the internet connection of the device
DEP_E152 executables of the component deployed on the device either exited too early or failed troubleshoot the failed component by analyzing deployment logs
DEP_E153 unable to either pull the docker image or build the source code for the component deployed on cloud verify that the docker image provided while adding the package still exists at the specified registry endpoint
DEP_E154 executables of the component deployed on cloud exited too early troubleshoot the failed component by analyzing deployment logs
DEP_E155 executables of the component deployed on cloud failed troubleshoot the failed component by analyzing deployment logs
DEP_E156 dependent deployment is in error state troubleshoot the dependent deployment that is in error state
DEP_E161 docker image not found for executables of components deployed on device verify that the path of the docker image is valid
DEP_E162 Validation error. Cases include:
  • Inconsistent values of ROS distro and CPU architecture variables for the device and package being provisioned.
  • rapyuta.io docker images not present on docker device.
  • Create package with appropriate values for ROS distro and CPU architecture variables.
  • Onboard the device again.
DEP_E163 application has stopped and exited unexpectedly, and crashes continuously debug the application using the corresponding deployment logs
DEP_E2xx internal rapyuta.io error in the components deployed on cloud report the issue together with the relevant details to the support team
DEP_E3xx internal rapyuta.io error in the components deployed on a device report the issue together with the relevant details to the support team
DEP_E4xx internal rapyuta.io error report the issue together with the relevant details to the support team

Restart Policy

Unlike deployments running on the cloud, which automatically restart if stopped due to an error, deployments that are running on devices do not automatically restart if exited due to an error or when devices are rebooted.

You can configure the behavior of deployments running on devices by setting their restart policies. There are three kinds of restart policies available for a device deployment:

  • Always
    Always restart deployments if the deployment executables are in an error state or if the device is rebooted.
  • On-failure
    Restart deployments only if the deployment executables exit due to an error, and the exit code is non-zero.
  • Never
    Do not restart deployments under any circumstance.

There are a couple of exceptions while applying the restart policies:

  • Restarting a deployment running on a device may fail if its executables are missing in $PATH. It is due to the components of the deployment failing to start.
  • rapyuta.io components that are shared between deployments running on the same device (like ROS Master) will have the same restart policy as that of the first deployment on a device irrespective of the restart policies of any subsequent deployments on the device.
  • If a deployment running on a device is stopped manually by stopping its docker container, both on-failure and always policies (if selected) are not applied unless the device is rebooted.

You can modify or override the initial setting of restart policy while deploying a package. Read deploying a package topic to learn how to do so.

For a deployment running on a device, the variable Restart Count (on the deployment details page) represents the number of times the deployment has restarted due to restarting of deployment components.

Deploying A Package

To deploy a package in rapyuta.io, follow the steps:

  1. On the left navigation bar, click CATALOG.
  2. Select the package you want to deploy.
  3. Click Deploy package.
  4. In the Name of deployment box, enter a name for the specific deployment you are creating for the package.
  5. A LABEL is a key-value pair. If you want to add a label, click Add label.
  6. If a component of the package has cloud runtime, skip to instruction 9.
  7. If a component of the package has device runtime, you must select the device you want to deploy the component on. Click Refresh the list of online devices to retrieve an updated list of online devices.
  8. Select the device from the Select device for deploying the component drop-down list.

    The list of devices is a set of online devices, which are pre-filtered to match the architecture (amd64, arm32v7, arm64v8) and device runtime (docker or preinstalled) required by the component in question.

  9. If you want to add a dependent deployment, click Add dependency, and select a deployment you want to add as a dependency from the drop-down list of deployment IDs.
  10. If you want to add a volume, click Add volume. Ensure that a running volume deployment is available before you add one.
  11. If you want to modify the initial setting of restart policy of components with device runtime, click Modify. Modify restart policy
  12. Click CREATE DEPLOYMENT > Confirm.

Deploy demo package You will be redirected to the Details page of the newly created deployment. The package is successfully deployed when the green colored bar moves from In progress to Provisioning to Succeeded indicating that the DEPLOYMENT PHASE has succeeded and the deployment STATUS is Running.

Deployment example

Furthermore, if dependent deployments are added then each dependency’s STATUS must read Running.

You may analyze the corresponding deployment logs generated while deploying a package.

If a deployment fails, the DEPLOYMENT PHASE will read Failed to start. You may have to click Deprovision Deployment, delete the package, create the package all over again, and try deploying it.