Advanced Topics
Monitoring CloudQuery


Monitoring CloudQuery can be done in a number of main ways:

  • Logging
  • OpenTelemetry


CloudQuery utilizes structured logging (in plain and JSON formats) which can be analyzed by local tools such as jq, grep and remote aggregations tools like loki, datadog or any other popular log aggregation that supports structured logging.

OpenTelemetry (Preview)

ELT workloads can be long running and sometimes it is necessary to better understand what calls are taking the most time; to potentially optimize those on the plugin side, ignore them or split them to a different workload. Plugins come with an OpenTelemetry library built in, but it is up to the plugin author to instrument the most important parts--usually the API calls--this way it is possible to see what calls take the longest time, or where throttling and errors are happening.

CloudQuery supports OpenTelemetry (opens in a new tab) tracing out of the box and can be enabled easily via configuration.

To collect traces you need a backend (opens in a new tab) that supports OpenTelemetry protocol. For example you can use Jaeger (opens in a new tab) to visualize and analyze traces.

To start Jaeger locally you can use Docker:

docker run -d \
  -p 16686:16686 \
  -p 4318:4318 \

and then specify in the source spec the following:

kind: source
  name: "aws"
  path: "cloudquery/aws"
  registry: "cloudquery"
  version: "v27.0.0"
  tables: ["aws_s3_buckets"]
  destinations: ["postgresql"]
  otel_endpoint: "localhost:4318"
  otel_endpoint_insecure: true # this is only in development when running local jaeger

After that you can open http://localhost:16686 (opens in a new tab) and see the traces:


In production, it is common to use an OpenTelemetry collector (opens in a new tab) that runs locally or as a gateway to batch the traces and forward it to the final backend. This helps with performance, fault-tolerance and decoupling of the backend in case the tracing backend changes.

OpenTelemetry and Datadog

In this quick example we will show how to connect an open telemetry collector to Datadog via OpenTelemetry exporter.

Firstly, you will need to have an OpenTelemetry Collector (opens in a new tab) running either locally or as a gateway. Here is an example of running it locally with docker:

docker run  -p 4319:4319 -v $(pwd)/config.yml:/etc/otelcol-contrib/config.yaml otel/opentelemetry-collector-contrib:0.91.0

following is an example for OTEL collector config.yml to receive traces locally on 4318 and export them to Datadog:

      site: "" # or your tenant site
      key: "<DATADOG_API_KEY>"

Once ingestion starts you should be able to start seeing the traces in Datadog under ServiceCatalog and Traces with ability to view average p95 latency, error rate, total duration and other useful information you can query to either split the workload better or improve the plugin scheduling if you are the plugin author: