Prometheus is a component in StreamNative Platform and is used to collect metrics from the Pulsar cluster. By default, Prometheus is enabled with StreamNative Platform. To disable it, you can set
monitoring.prometheus: false in the
values.yaml YAML file and then update the resource.
Configure CPU and memory resources
You can configure the CPU, memory, and disk for Prometheus in the
values.yaml YAML file. Then, you can use the
helm upgrade command to restart the StreamNative Platform to make updates effective. For details about how to configure CPU and memory resources, see CPU and memory resources.
prometheus: # keep request = limit to keep this container in guaranteed class resources: requests: cpu: memory: volumes: data: name: size:
Configure WAL compression
When a Pulsar cluster has lots of topics, a quantity of monitoring data is generated, which might cause the Prometheus server to restart due to Out of Memory (OOM) on Write-Ahead Log (WAL) replay. To prevent Prometheus OOM, you can configure
prometheus.extraArgs.--storage.tsdb.wal-compression property in the
values.yaml YAML file as follows and then update the resource. For a full list of options about Prometheus local storage, see Prometheus documentation.
prometheus: extraArgs: # https://prometheus.io/docs/prometheus/latest/storage/#operational-aspects - --storage.tsdb.wal-compression