Skip to content

Restart an agent

You might need to restart your agent from time to time for any of the following reasons:

  • Agents can sometimes crash or become unstable.
  • You changed the agent's configuration.
  • You added a new certificate to be used for setting up a proxy server.
  • You added new Python libraries.
  • You added new third-party drivers.
  • To receive agent updates.

You can restart the agent in one of two ways:

  • From the Agents screen in the Data Productivity Cloud.

  • From the cloud provider's control panel:

    Note

    A changed agent deployment configuration may require a restart of the cloud provider service to take effect.

Read Pause an agent for an alternative way of managing agents that need restarting.

Restart from the Agents screen

  1. In the left navigation, click the Agents & Instances icon . Then, select Agents from the menu.
  2. Locate the agent you want to restart and click ....
  3. Click Restart agent.
  4. Click Yes restart to confirm.

This feature takes advantage of how a cloud provider will manage an installed agent service by ensuring that the required amount of agent instances (tasks/pods, depending on your deployment) are always started if any drop.

The Restart agent option sends a graceful shutdown request to all the instances (tasks/pods) contained within the agent service on the cloud provider. Each instance will complete any pipeline tasks currently in progress, or wait for any open transactions to complete, and then shut itself down. The agent service will then start up new, replacement instances.

Running pipelines will continue as normal. Any pipeline tasks that are sent to an agent that's currently restarting will be queued until the new service is back up and running. Your pipelines will be unaffected, though they may take a few minutes longer than normally expected for execution to complete.

Assuming that there are no running tasks, it usually takes a few minutes for an agent instance to shutdown and a new instance to start back up. If you have multiple instances and one is working on a longer pipeline task or currently has an open transaction, then the other restarted agent instances will take on any new pipeline tasks.

Restart an AWS ECS Fargate agent

You will need access to your AWS account with the appropriate permissions.

To restart your AWS agent, follow these steps:

  1. Log in to the AWS console.
  2. In the AWS console, type Elastic Container Service in the search bar, and select that service.
  3. Locate the ECS service running your agent, and click Update service.
  4. Select the latest task definition and click Update.

Restart an Azure agent

You will need access to your Azure account with the appropriate permissions.

There are two ways you can restart an Azure agent:

  • The Azure Portal
  • The Azure CLI

Using the Azure Portal

To restart your agent using the Azure Portal, follow these steps:

  1. Log in to the Azure Portal.
  2. In the search bar at the top of the Azure Portal, enter "Container apps", and click the Container Apps link in the results menu.
  3. Click your chosen container app from the list.
  4. Click Application to display a context menu.
  5. Click Revisions and replicas.
  6. Click on the name of the intended Active Revision.
  7. A panel to the right will appear. At the top of Revision details, click Restart.

Using the Azure CLI

To restart your agent, you can open the Azure CLI either locally on your machine, or through the Azure portal.

Use the following command:

az containerapp revision restart --revision <REVISION_NAME> --resource-group <RESOURCE_GROUP_NAME> --subscription <SUBSCRIPTION_NAME>"

For example, if you have an agent called cloud-agent-dev in a resource group called cloud-agent-dev, and your Azure subscription is called Matillion Dev, you are required to use the following command:

az containerapp revision restart --revision cloud-agent-dev --resource-group cloud-agent-dev --subscription "Matillion Dev"

Restart a Snowflake agent

You can restart the agent by suspending and resuming the compute pool from within Snowflake. Suspending the compute pool will prevent the agent from being used by the Data Productivity Cloud, and will stop all running tasks. Resuming the compute pool will make the agent available for use again.

  1. From your Snowflake Home screen, click Data Products → Apps. You must be using the role that originally installed the application.
  2. Locate Matillion Data Productivity Cloud in the list of apps, and click to select it. If you have multiple installs of the Native App, select the one you wish to restart.
  3. Click the Control Panel tab.

  4. Click Suspend Compute Pool, then click Suspend to confirm. This will put the pool into a Stopping state and change the agent status to Suspending.

    Warning

    This will suspend the compute pool and all agent instances that it supports. This will stay suspended until you actively click Resume Compute Pool.

  5. Wait for the agent to stop. This may take a few minutes.

  6. When the agent is in a Suspended state, click Resume Compute Pool. This will put the pool into a Starting state and change the agent status to Pending.

  7. Wait for the agent to start. This may take a few minutes.

    Note

    After the compute pool changes to an Active state, the agent status may remain as Pending for a few minutes before changing to Running.

© Copyright 2024 Matillion Ltd | Legal | Privacy Policy | Disclaimer | Modern Slavery Statement