collectd Integration
System metric collection from collectd agents with write_http plugin support. Forward CPU, memory, disk, and application metrics from your collectd fleet to TigerOps with AI anomaly detection.
How It Works
Install collectd Agent
Install collectd on your Linux hosts using your package manager. Enable the plugins relevant to your workload — cpu, memory, disk, interface, load, processes, and more.
Configure write_http Plugin
Add the write_http plugin to your collectd.conf, pointing to the TigerOps collectd ingestion endpoint. Set your API key in the Authorization header for authentication.
Enable Application Plugins
Activate collectd plugins for your stack: mysql, postgresql, nginx, apache, java (JMX), python, and more. All plugin metrics flow to TigerOps automatically.
Build Fleet Dashboards
TigerOps groups collectd hosts by hostname and plugin. Create fleet-wide dashboards, per-host drill-downs, and AI anomaly alerts across all your collectd agents simultaneously.
What You Get Out of the Box
write_http Plugin Support
Native endpoint compatible with collectd write_http plugin in JSON and PUTVAL format. No format conversion needed — collectd sends metrics directly to TigerOps at your configured interval.
100+ Plugin Metrics Out of the Box
All collectd plugins are supported: cpu, memory, disk, interface, load, processes, vmem, irq, entropy, thermal, uptime, users, and third-party plugins like nginx and mysql.
TypeDB-Aware Metric Parsing
TigerOps understands the collectd type database (types.db). Gauge, counter, derive, and absolute data source types are handled correctly, including counter wrapping and rate conversion.
Plugin-Instance Label Preservation
collectd plugin instances (e.g., disk-sda, interface-eth0) are preserved as dimensional labels. Query metrics for specific disks or interfaces across your entire server fleet.
Network Plugin Aggregation
Use collectd network plugin to aggregate metrics from multiple hosts to a single collectd relay, then forward to TigerOps. Reduces direct connections while maintaining per-host granularity.
AI-Powered Capacity Forecasting
TigerOps applies ML forecasting to collectd disk, memory, and CPU metrics. Get capacity warnings days in advance before hosts become saturated.
collectd.conf write_http Plugin
Add the write_http plugin block to your collectd.conf to forward metrics to TigerOps.
# collectd.conf — TigerOps write_http output
# collectd 5.x compatible
Hostname "web-prod-01"
FQDNLookup false
Interval 10
Timeout 20
ReadThreads 4
WriteThreads 1
# System plugins
LoadPlugin cpu
LoadPlugin memory
LoadPlugin disk
LoadPlugin interface
LoadPlugin load
LoadPlugin processes
LoadPlugin uptime
# Application plugins (enable as needed)
# LoadPlugin mysql
# LoadPlugin postgresql
# LoadPlugin nginx
# TigerOps output via write_http
LoadPlugin write_http
<Plugin write_http>
<Node "tigerops">
URL "https://ingest.atatus.net/collectd"
Format "JSON"
Header "Authorization: Bearer ${TIGEROPS_API_KEY}"
Timeout 10
BufferSize 65536
LowSpeedLimit 0
Metrics true
Notifications false
StoreRates true
# Verify TLS certificate
VerifyPeer true
VerifyHost true
</Node>
</Plugin>Common Questions
Which collectd write plugin should I use with TigerOps?
Use write_http with format="JSON" or format="Command" (PUTVAL). The TigerOps collectd endpoint at https://ingest.atatus.net/collectd accepts both formats. JSON is recommended for richer metadata support.
Does TigerOps handle collectd counter and derive types correctly?
Yes. TigerOps reads collectd data source types from the standard types.db schema. Counter and derive types are automatically converted to per-second rates. Gauge and absolute types are stored as-is.
Can I use collectd network plugin to relay metrics to TigerOps?
Yes. Deploy a collectd relay host with the network input plugin (receiving from agents) and write_http output plugin (forwarding to TigerOps). This is the recommended topology for large deployments with hundreds of hosts.
How do I monitor collectd agent health?
Enable the collectd plugin itself (meta-plugin) and the logfile plugin with TigerOps log ingestion. TigerOps tracks collection intervals, dispatch failures, and missed intervals as agent health metrics.
Is collectd suitable for containerized workloads?
collectd is primarily host-centric. For Kubernetes, we recommend the TigerOps agent or Prometheus with cAdvisor. You can run collectd in a DaemonSet for host-level metrics while using the TigerOps agent for pod-level metrics in parallel.
Upgrade Your collectd Fleet to Modern Observability
Keep your collectd agents. Add AI anomaly detection, capacity forecasting, and cross-signal correlation. Three lines of config.