Private Agents and Kubernetes

Introduction

A group of Jitterbit Harmony Private Agents can be configured to be auto-registering and autoscaling by using Kubernetes. This page describes how to configure a group of Private Agents using Kubernetes and Helm.

Reference: The Values YAML File

The Agent Group is configured using a specific  values.yaml file. These are the attributes and definitions in the file, based on the register.json defined for the Agent Group (original in the Jitterbit charts GitHub repository):

Installing and Running the Agent Group

Follow these steps to install and run an Agent Group in Kubernetes using Helm:

1. Ensure that your Kubernetes cluster is up and running and that your Helm client is configured properly to use the cluster along with Jitterbit's public Helm repository.

2. Generate the Agent Group's register.json file, if you haven't already. Instructions on this are at Agent Automatic Registration and Autoscaling.

3. Create your Agent Group's values.yaml file. Note that as YAML is a superset of JSON, you can assign agent.registerjson to your register.json's JSON object.  Here's an example values.yaml file created using the register.json file from the previous step. It was converted into YAML by placing it under agent.registerjson:

   nameOverride: private-agent-group
   image:
     repository: jitterbit/agent
     tag: latest
   agent:
     registerjson:
       cloudUrl: https://na-east.jitterbit.com
       username: $00+LongEncryptedAndHashedUsername==
       password: $00LongEncryptedAndHashedPassword=
       agentGroupId: 346591
       agentNamePrefix: "%host%"
       deregisterAgentOnDrainstop: true
   hpa:
     minReplicas: 2
     maxReplicas: 3
   replicas: 3

   WARNING: If you are using a Windows operating system, make sure that the values.yaml file has the UNIX line endings \n instead of DOS's \r\n.

4. Check that your Docker runtime has sufficient resources allocated to it to accommodate the maximum number of replicas you have specified. Each Docker Private Agent requires 0.25 cores and 2 GB memory:

   Resource	Value Calculation	Value Not to Exceed
   Cores	0.25 cores x maxReplicas	(number of CPUs) x (number of cores/CPU)
   Memory	2 GB x maxReplicas	Available memory allocated to the Kubernetes cluster

5. In the example Docker Kubernetes system shown below (a macOS computer with 6 cores—labelled CPUs in the Docker UI—and 16 GB of memory allocated to Docker) you could run at most 7 Docker Private Agents. Eight agents would require 2 cores and 16 GB of memory, using all the available memory. However, as Kubernetes itself runs some containers on the same Docker runtime (known as Kubernetes in Docker), allow at minimum headroom of one core and a portion of the memory rather than dedicating all of the resources available to the Agent Group:

   

6. Add Jitterbit's public Helm chart repository to your client. No credentials are required:

   helm repo add jitterbit https://jitterbit.github.io/charts
   # "jitterbit" has been added to your repositories

7. Update your Helm client's local cache of repositories (recommended). This will ensure that you are using the latest version of the Helm chart:

   helm repo update

8. Install your Agent Group's Helm release:

   helm upgrade --install --force --atomic --timeout 1h -f ./values.yaml private-agent-group jitterbit/agent-group

9. Wait for each agent pod to spin up and monitor using the Jitterbit Harmony Management Console:

   watch kubectl get pods

Accessing an Individual Agent

If you need to access one of the agents, you can execute a Bash shell in its container using the following code, replacing private-agent-group-0 with the Docker name of the particular agent you need to access:

kubectl exec -it private-agent-group-0 bash

Agent Group Teardown

When you are ready to tear down the Agent Group, uninstall the Helm release:

helm uninstall private-agent-group