Skip to main content
The KloudMate Agent is built on the OpenTelemetry Collector framework, which uses a modular architecture composed of four main component types: receivers, processors, exporters, and extensions.

​
Component Architecture

The collector processes telemetry data through a pipeline architecture:

Receivers

Collect telemetry data from various sources including hosts, databases, cloud platforms, and applications

Processors

Transform, filter, enrich, and manipulate telemetry data as it flows through pipelines

Exporters

Send processed telemetry data to backends like KloudMate or other observability platforms

Extensions

Add capabilities like health checks, storage, and diagnostics without direct data access

​
Pipeline Configuration

Components are organized into pipelines in the collector configuration. Each pipeline processes a specific telemetry type (metrics, logs, or traces).

​
Example Pipeline Structure

service:
  pipelines:
    metrics:
      receivers: [otlp, hostmetrics]
      processors: [resourcedetection, batch]
      exporters: [otlphttp]
    logs:
      receivers: [otlp, filelog]
      processors: [resource, batch]
      exporters: [otlphttp]
    traces:
      receivers: [otlp]
      processors: [batch]
      exporters: [otlphttp]

​
Component Categories

​
Receivers

KloudMate Agent includes receivers for:
  • Infrastructure: Host metrics, Docker, Kubernetes
  • Databases: MySQL, PostgreSQL, MongoDB, Redis, Oracle, SQL Server, SAP HANA, Elasticsearch
  • Web Servers: Apache, Nginx, IIS
  • Message Queues: RabbitMQ, Kafka
  • Cloud Platforms: AWS CloudWatch, Azure Monitor, Google Cloud Monitoring
  • Logs: File logs, journald, syslog, Fluent Forward
  • Network: Netflow, SNMP
  • Custom: Prometheus metrics, HTTP checks, SQL queries

​
Processors

Processors handle data transformation and enrichment:
  • Essential: Batch, Memory Limiter
  • Resource & Attributes: Resource Detection, K8s Attributes, Attributes
  • Transformation: Transform (OTTL), Metrics Transform, Cumulative to Delta, Delta to Rate
  • Filtering: Filter, Probabilistic Sampler, Redaction
  • Organization: Group by Attributes, Group by Trace

​
Exporters

KloudMate Agent supports these exporters:
  • OTLP: Standard protocol for sending to KloudMate and other OTLP-compatible backends
  • OTLP HTTP: HTTP-based OTLP protocol variant
  • Debug: Console output for troubleshooting
  • Nop: No-operation for testing

​
Extensions

Extensions provide supporting functionality:
  • Health Check: Kubernetes liveness and readiness probes
  • File Storage: Persistent state across restarts
  • Z-Pages: In-process diagnostics
  • Memory Limiter: Memory usage control

​
Platform-Specific Components

The KloudMate Agent includes platform-specific components that are only available on certain operating systems or deployment modes.

​
Linux-Specific

  • Docker Stats Receiver: Container metrics from Docker daemon
  • Journald Receiver: Systemd journal logs

​
Windows-Specific

  • Windows Event Log Receiver: Windows event logs
  • Windows Performance Counters Receiver: Performance counter metrics
  • IIS Receiver: Microsoft IIS web server metrics
  • Windows Service Receiver: Windows service status monitoring
  • SQL Server Receiver: Microsoft SQL Server metrics

​
Kubernetes-Specific

  • K8s Cluster Receiver: Cluster-level metrics
  • K8s Objects Receiver: Kubernetes object events as logs
  • K8s Events Receiver: Cluster events
  • Kubelet Stats Receiver: Pod and container metrics
  • K8s Attributes Processor: Kubernetes metadata enrichment

​
Configuration Best Practices

1

Start with essential components

Begin with OTLP receiver, batch processor, and OTLP HTTP exporter for basic functionality
2

Add monitoring receivers

Enable host metrics and application-specific receivers based on your infrastructure
3

Configure processors

Add resource detection and attributes processors to enrich telemetry with context
4

Optimize performance

Tune batch sizes, memory limits, and collection intervals for your workload
5

Enable extensions

Configure health checks for production deployments and Z-pages for debugging

​
Component Lifecycle

Components are initialized when the collector starts and shut down gracefully when the collector stops. Failed components can cause the entire collector to fail, so ensure proper configuration and testing.

​
Initialization Order

  1. Extensions are initialized first
  2. Receivers start collecting data
  3. Processors begin processing data from receivers
  4. Exporters connect to backends and start sending data

​
Shutdown Order

Shutdown occurs in reverse order to prevent data loss:
  1. Receivers stop accepting new data
  2. Processors finish processing queued data
  3. Exporters flush remaining data to backends
  4. Extensions shut down last

​
Remote Configuration

KloudMate Agent supports remote configuration, allowing you to manage component configurations centrally without restarting agents. See the Remote Configuration guide for details.
Changing component configurations requires careful testing. Invalid configurations can cause the collector to fail to start or drop telemetry data.

​
Next Steps

Explore Receivers

Learn about available receivers and their configuration options

Configure Processors

Understand how to transform and enrich your telemetry data

Setup Exporters

Configure exporters to send data to KloudMate or other backends

Add Extensions

Enable supporting capabilities like health checks and diagnostics

Build docs developers (and LLMs) love

Get started for freeTalk to us