Here's a quick summary of what we will be doing in this guide

• Create and configure Cloud SQL
• Create and configure Compute Engine VM instance to deploy Telegraf and OpenTelemetry Collector
• Deploy Telegraf to fetch the metrics from Google Cloud monitoring
• Deploy OpenTelemetry Collector to scrape the metrics from Telegraf
• Send and Visualize the metrics in SigNoz Cloud

Prerequisites

• Google Cloud account with administrative privilege or Cloud SQL Admin privilege.
• SigNoz Cloud Account (we are using SigNoz Cloud for this demonstration, we will also need ingestion details. To get your Ingestion Key and Ingestion URL, sign-in to your SigNoz Cloud Account and go to Settings >> Ingestion Settings)
• Access to a project in GCP

Setup

Get started with Cloud SQL Configuration

Follow the steps mentioned in the Creating Cloud SQL document to create Cloud SQL instance.

Deploy Telegraf to fetch the metrics from Google Cloud Monitoring

You will need a Compute Engine instance to install Telegraf and OTel Collector. You can follow the instructions on the Creating Compute Engine document to create the Compute Engine instance.

Step 1: Install telegraf which will collect metrics from Google Cloud Monitoring for Cloud SQL, see the available metrics for Cloud SQL. The installation process for the respective operating system can be found in official documentation.

After successful installation, the telegraf status should be active and running.

The configuration file for telegraf can be found here:

/etc/telegraf/telegraf.conf

Step 2: Configure the Telegraf input and output plugin by adding configurations to the telegraf.conf file.

# Gather timeseries from Google Cloud Platform v3 monitoring API
[[inputs.stackdriver]]
  ## GCP Project
  project = "omni-new"
  ## Include timeseries that start with the given metric type.
  metric_type_prefix_include = [
    "cloudsql.googleapis.com",
  ]
  ## Most metrics are updated no more than once per minute; it is recommended
  ## to override the agent level interval with a value of 1m or greater.
  interval = "1m"
# Send OpenTelemetry metrics over gRPC
[[outputs.opentelemetry]]
  ## Override the default (localhost:4317) OpenTelemetry gRPC service
  ## address:port
  service_address = "localhost:4317"

We are assuming that the OTel Collector is running on the same machine as Telegraf. If that is not the case, you will have to change the service_address in the outputs.opentelemetry section accordingly.

Deploy OpenTelemetry to scrape the metrics from Telegraf

Step 1: Install and configure OpenTelemetry for scraping the metrics from Telegraf. Follow OpenTelemetry Binary Usage in Virtual Machine guide for detailed instructions.

Step 2: After successful configuration start the OTel Collector using following command:

./otelcol-contrib --config ./config.yaml &> otelcol-output.log & echo "$!" > otel-pid

Step 3: Restart the Telegraf service

Step 4: If the configurations are configured correctly with Telegraf and we can see the output logs from OpenTelemetry as follows:

Send and Visualize the metrics obtained by OpenTelemetry in SigNoz

Step 1: Go to the SigNoz Cloud URL and head over to the dashboard.

Step 2: If not already created, create a new dashboard and select Time Series.

Step 3: Select metric for Cloud SQL

All metrics starting with cloudsql_googleapis_com_ have been collected from Cloud SQL database.

Troubleshooting

If you run into any problems while setting up monitoring for your Cloud SQL's metrics with SigNoz, consider these troubleshooting steps:

• Verify Configuration: Double-check your config.yaml file to ensure all settings, including the ingestion key and endpoint, are correct.
• Review Logs: Look at the logs of both Telegraf and the OpenTelemetry Collector to identify any error messages or warnings that might provide insights into what’s going wrong.
• Update Dependencies: Ensure all relevant packages and dependencies are up-to-date to avoid compatibility issues.
• Consult Documentation: Review the SigNoz, OpenTelemetry, and Telegraf documentation for any additional troubleshooting of the common issues.