Configuring Compute Resources for the connectware-agent Helm Chart
Compute resource requests are used by Kubernetes to schedule workloads on a node that has the required resources available. Compute resource limits are used to limit a workloads consumption.
In order to be scheduled in Kubernetes highest Quality of Service class, agents installed through the connectware-agent
Helm chart come with a default of using the same value for requests and limits. These values are:
CPU
2000m
Memory
2000Mi
To adjust the compute resources for the agent, specify a Kubernetes Quantity for cpu
and memory
for resources.requests
and resources.limits
inside the agents entry in protocolMapperAgents
context of your values.yaml file.
Example
Hint: cpu
is usually specified in m
, representing “Milli-CPU”, with 1000m
roughly meaning one CPU core. memory
is usually specified in Mi
, representing “Mebibytes”.
Removing Default Values
In order to be scheduled in Kubernetes highest Quality of Service class, agents installed through the connectware-agent
Helm chart come with a default of using the same value for requests and limits. These values are:
CPU
2000m
Memory
2000Mi
To remove either you can set the value resources.request
and resources.limits
inside the agents entry in protocolMapperAgents
context of your values.yaml file to an empty YAML object {}
.
Example
You can also completely remove resource definitions by setting resources
inside the agents entry in protocolMapperAgents
context of your values.yaml file to an empty YAML object {}
.
Example
Full Example: Individual Resources for a Single Agent
In this example, we will set lower default resources, but for our agent welder-robots
we will request more resources and remove limits, as we do want it to not be limited for its important work.
Example
Last updated
Was this helpful?