Creating and managing agent pools
Similar to resource groups, agent pools help you organize and manage agents that are installed in different environments.
You can use an agent pool to spread the deployment processing work among participating agents and provide relief to otherwise overburdened agents. Users assign agents to pools and pools are assigned to resources, just like lone agents. When an agent pool is assigned to a resource, work items are sent to eligible agents. If you don't use a pool, then the agent that is assigned to the resource is assigned all the processing work regardless of its current workload. Such a workload might lead to bottlenecks.Note: Starting in IBM DevOps Deploy (Deploy) 7.0.3,
there is a new agent pool algorithm which will assign the workload to the most available
agents in the pool. Most available agents are determined by which agents have the
minimum running plugin steps in the agent pool. If there are multiple agents in the pool
with the same plugin step count, the agents the database returns first will be assigned
the workload. Prior to Deploy 7.0.3,
agents were assigned the work at random.
Creating an agent pool
Procedure
To create an agent pool, specify the agents that are the members of the pool.
- From the Resources tab, click Agent Pools.
-
From the Agent Pools pane, click Create Agent
Pool.
The Create Agent Pool window opens.
- Specify the pool name.
- Optional: Enter a description in the Description field.
- Optional: To manage the teams that can access the agent pool, click the plus icon by Teams.
-
To add agents to the pool, click the Add Agents field.
A list of the available agents appears.
-
Select the agent or agents to add to the pool.
You can filter the listed agents by entering search text into the field.
- When you are finished, click Save.
Managing agent pools
Procedure
You can manage agent pools from the web interface.
Agent pools for containers
Procedure
Agent pools can be used to provide efficiency with your use of Deploy licenses. You can take advantage of running your agents in a Kubernetes-based environment where you can spin agent instances up and down quickly.
Follow these steps in order to achieve this efficiency: