Noodara
Operate

Metrics and status

What Noodara samples on each server, how often, how long it keeps it, and when a status pill turns amber or red.

Noodara samples every connected server and the service containers on it. Server and service pages show the samples as charts, with status pills for the thresholds below.

Sampling

  • One sample per connected server every 60 seconds, over the same SSH connection the panel already uses. Nothing is installed on the server.
  • Change the interval with NOODARA_METRICS_INTERVAL_MS, from 30000 (30 seconds) to 300000 (5 minutes). A tick that is still running when the next one is due skips it.
  • Samples are kept for 168 hours (7 days) at full resolution, then pruned. Change it with NOODARA_METRICS_RETENTION_HOURS, from 1 to 720 hours (30 days).

Both variables are listed in Configuration variables.

What is measured

MetricServerService
cpu_percentHost CPUContainer CPU
memory_used_bytesHost memory in useContainer memory in use
memory_percentOf host memoryOf the container limit, or of host memory without a limit
disk_used_percentRoot filesystem—
restart_count—Container restarts

A chart shows at most 500 points. Longer ranges are grouped into buckets of at least 60 seconds. A bucket without samples is a gap in the chart: missing data is never drawn as zero.

Status thresholds

PillWhen
High diskThe root filesystem is at 90% or more.
High memoryMemory is at 90% or more for the last 3 samples in a row.
Crash loopThe container restarted 3 times or more within 10 minutes.
Out of memoryDocker reports the container was OOM killed, or it exited with code 137 while its memory was at 90% or more of the limit.

The service page also shows the healthcheck result, the last deployment and the restart count.

Errors

CodeMeaning
METRICS_UNAVAILABLENo samples yet. Check that the server is connected and try again in a minute.

A freshly connected server shows its first sample after one interval.

On this page