Enable and use the Fluid JindoRuntime FUSE client for monitoring - Container Service for Kubernetes

Fluid offers an out-of-the-box JindoRuntime dashboard to demonstrate the collection of monitoring metrics for each JindoCache distributed cache engine in a Container Service for Kubernetes (ACK) cluster. The collected metrics include server-side metrics of the cache engine and Filesystem in Userspace (FUSE) client metrics. For performance reasons, by default, JindoRuntime FUSE client metrics are not exposed. Therefore, real-time metric data of the FUSE client cannot be viewed on the JindoRuntime dashboard. This topic describes how to enable the collection of Fluid JindoRuntime FUSE client metrics and view the monitoring data on the Fluid JindoRuntime dashboard.

Prerequisites

Managed Service for Prometheus is enabled in the ACK cluster or ACK Serverless cluster. For more information, see Use Managed Service for Prometheus.
The cloud-native AI suite and ack-fluid version 1.0.11 or later are installed in the cluster. For more information, see Deploy the cloud-native AI suite.
Note
If you have installed the open-source Fluid, uninstall it before installing ack-fluid to prevent potential conflicts and ensure a smooth installation process.

Step 1: Integrate Fluid with ARMS

Log on to the ARMS console.
In the left-side navigation pane, click Integration Center. In the AI section, click the Fluid card.
In the Select a Kubernetes cluster section of the Fluid page, select the desired cluster. If the page shows that Fluid is already installed, skip the steps in this section.

In the Configuration Information section, configure the parameters and click OK.

Parameter	Description
Name (Not required)	The unique name of the Fluid exporter. You can leave it empty.
metrics collection interval (seconds)	The interval at which you want the service to collect monitoring data.

You can view integrated components on the Integration Management page of the ARMS console.
1. Log on to the ARMS console.
2. In the left-side navigation pane, click Integration Management. On the Integrated Addons tab, click the Fluid card.
3. On the Environments tab, click View Details in the Actions column to view the Fluid component in the cluster and alert rules.

Step 2: Configure exposure metrics and scraping policy

When deploying JindoRuntime, you can enable exposure and configure scraping policies for metrics for the FUSE client using the spec.fuse.metrics.enabled and spec.fuse.metrics.scrapeTarget parameters respectively. The following YAML template shows an example:

apiVersion: data.fluid.io/v1alpha1
kind: JindoRuntime
metadata:
  name: hadoop
spec:
  replicas: 2
  fuse:
    metrics:
      enabled: true # Set this parameter to true to enable the metrics exposure feature for the JindoRuntime FUSE client.
      scrapeTarget: All # The scraping policy for the JindoRuntime FUSE client metrics.
  tieredstore:
    levels:
      - mediumtype: MEM
        path: /dev/shm
        volumeType: emptyDir
        quota: 2Gi
        high: "0.99"
        low: "0.95"

The following table describes the spec.fuse.metrics.enabled and spec.fuse.metrics.scrapeTarget parameters:

Note

For a comprehensive explanation of JindoRuntime parameter metrics, see Use JindoFS to accelerate access to OSS.

Parameter

Description

Default value

spec.fuse.metrics.enabled

This parameter specifies whether to enable the metrics exposure feature for the JindoRuntime FUSE client.

If spec.fuse.metrics.enabled: true is set without specifying a port for handling metrics requests through the spec.fuse.args parameter, for example, by adding the -ometrics_port=8080 argument, Fluid will automatically assign an available port to the JindoRuntime FUSE client and configure it for external metrics services.

false

spec.fuse.metrics.scrapeTarget

This parameter specifies the scraping policy for the JindoRuntime FUSE client metrics. The following four scraping policies are supported:

None: No JindoRuntime FUSE client metrics are scraped.
MountPod: Only metrics exposed by JindoRuntime FUSE pods in the MountPod mount mode in non-Serverless environments are scraped.
Sidecar: Only metrics exposed by JindoRuntime FUSE Sidecar containers in the Sidecar mount mode in Serverless environments are scraped.
All: Metrics from FUSE clients in both MountPod and Sidecar mount modes are scraped.

Important

Exercise caution when setting spec.fuse.metrics.scrapeTarget to Sidecar or All. In a Serverless environment using Fluid in Sidecar mount mode, each pod starts a FUSE Sidecar container to provide data access services to the main container. If the number of concurrent pods is too high, scraping metrics from all FUSE Sidecar containers may lead to significant overhead and place considerable pressure on the Prometheus instance.

None

Step 3: View the Fluid JindoRuntime dashboard

Log on to the ACK console. In the left-side navigation pane, click Clusters.
On the Clusters page, find the cluster that you want to manage and click its name. In the left-side pane, choose Operations > Prometheus Monitoring.
On the Prometheus Monitoring page, click the Others tab, and click the Fluid JindoRuntime Dashboard tab to view the detailed Fluid control panel dashboard.
- If FUSE client metrics scraping is enabled in the MountPod mount mode by setting scrapeTarget to MountPod or All, FUSE client metrics can be viewed in the FUSE Metrics (via CSI) section, which displays metrics from FUSE pod clients.
- If FUSE client metrics scraping is enabled in the Sidecar mount mode by setting scrapeTarget to Sidecar or All, FUSE client metrics can be viewed in the FUSE Metrics (via Sidecar) section, which displays metrics from FUSE Sidecar container clients.
For a detailed explanation of the dashboard data and monitoring metrics on the Fluid JindoRuntime Dashboard tab, see Fluid dashboard parameters and Introduction to metrics.