Apache Aurora and Thanos Integration

Input and output integration overview

This plugin gathers metrics from Apache Aurora schedulers, providing insights necessary for effective monitoring of Aurora clusters.

This plugin sends metrics from Telegraf to Thanos using the Prometheus remote write protocol over HTTP, allowing efficient and scalable ingestion into Thanos Receive components.

Integration details

Apache Aurora

The Aurora plugin is designed to gather metrics from Apache Aurora schedulers. This plugin connects to specified schedulers using their respective URLs and retrieves operational metrics that help in monitoring the health and performance of Aurora clusters. It primarily captures numeric data from the /vars endpoint, ensuring key metrics related to task execution and resource utilization are monitored. The plugin enhances operational insights by utilizing HTTP Basic Authentication for secure access. With optional TLS configuration, it further bolsters security when transmitting data. The plugin provides a robust way to interface with Apache Aurora, reflecting a focus on operational reliability and ongoing performance assessment across distributed systems.

Thanos

Telegraf’s HTTP plugin can send metrics directly to Thanos via its Remote Write-compatible Receive component. By setting the data format to prometheusremotewrite, Telegraf can serialize metrics into the same protobuf-based format used by native Prometheus clients. This setup enables high-throughput, low-latency metric ingestion into Thanos, facilitating centralized observability at scale. It is particularly useful in hybrid environments where Telegraf is collecting metrics from systems outside Prometheus’ native reach, such as SNMP devices, Windows hosts, or custom apps, and streams them directly to Thanos for long-term storage and global querying.

Configuration

Apache Aurora

[[inputs.aurora]]
  ## Schedulers are the base addresses of your Aurora Schedulers
  schedulers = ["http://127.0.0.1:8081"]

  ## Set of role types to collect metrics from.
  ##
  ## The scheduler roles are checked each interval by contacting the
  ## scheduler nodes; zookeeper is not contacted.
  # roles = ["leader", "follower"]

  ## Timeout is the max time for total network operations.
  # timeout = "5s"

  ## Username and password are sent using HTTP Basic Auth.
  # username = "username"
  # password = "pa$$word"

  ## Optional TLS Config
  # tls_ca = "/etc/telegraf/ca.pem"
  # tls_cert = "/etc/telegraf/cert.pem"
  # tls_key = "/etc/telegraf/key.pem"
  ## Use TLS but skip chain & host verification
  # insecure_skip_verify = false

Thanos

[[outputs.http]]
  ## Thanos Receive endpoint for remote write
  url = "http://thanos-receive.example.com/api/v1/receive"

  ## HTTP method
  method = "POST"

  ## Data format set to Prometheus remote write
  data_format = "prometheusremotewrite"

  ## Optional headers (authorization, etc.)
  # [outputs.http.headers]
  #   Authorization = "Bearer YOUR_TOKEN"

  ## Optional TLS configuration
  # tls_ca = "/path/to/ca.pem"
  # tls_cert = "/path/to/cert.pem"
  # tls_key = "/path/to/key.pem"
  # insecure_skip_verify = false

  ## Request timeout
  timeout = "10s"

Input and output integration examples

Apache Aurora

1. Dynamic Resource Allocation Monitoring: Utilize the Aurora plugin to build a real-time dashboard displaying metrics related to resource allocation in your Aurora clusters. By aggregating data from multiple schedulers, you can visualize how resources are distributed among various roles (leader and follower), enabling proactive management of resource utilization and helping prevent bottlenecks in production workloads.

2. Alerting on Scheduler Health: Implement alerting mechanisms where the Aurora plugin checks the health of schedulers periodically. If a scheduler role responds with a status that indicates a failure to communicate (non-200 status), alerts can be automatically generated and sent to the operations team via email or messaging apps, ensuring immediate attention to critical issues and maintaining availability in service management.

3. Performance Benchmarking Over Time: By continuously collecting metrics such as job update events and execution times, this plugin can assist teams in benchmarking the performance of their Apache Aurora deployment over time. Relevant metrics can be logged into a time-series database, enabling historical analysis, trend identification, and understanding how changes in the system, such as configuration adjustments or workload changes, impact performance.

4. Integration with CI/CD Pipelines: Integrate the metrics collected via the Aurora plugin with CI/CD pipeline tools to monitor how deployments affect runtime metrics in Aurora. Teams can thereby ensure that new releases do not adversely impact scheduler performance and can roll back changes seamlessly if any metric exceeds predefined thresholds after deployment.

Thanos

1. Agentless Cloud Monitoring: Deploy Telegraf agents across cloud VMs to collect system and application metrics, then stream them directly into Thanos using Remote Write. This provides centralized observability without requiring Prometheus nodes at each location.

2. Scalable Windows Host Monitoring: Use Telegraf on Windows machines to collect OS-level metrics and send them via Remote Write to Thanos Receive. This enables observability across heterogeneous environments with native Prometheus support only on Linux.

3. Cross-Region Metrics Federation: Telegraf agents in multiple geographic regions can push data to region-local Thanos Receivers using this plugin. From there, Thanos can deduplicate and query metrics globally, reducing latency and network egress costs.

4. Integrating Third-Party Data into Thanos: Collect metrics from custom telemetry sources such as REST APIs or proprietary logs using Telegraf inputs and forward them to Thanos via Remote Write. This brings non-native data into a Prometheus-compatible, long-term analytics pipeline.

Feedback

Thank you for being part of our community! If you have any general feedback or found any bugs on these pages, we welcome and encourage your input. Please submit your feedback in the InfluxDB community Slack.