Health check

Run a health check

python -m src --healthcheck --all --workers 20

Read-only: only show commands are sent, no config changes. Each device’s raw output is captured to PRE_CHECK/{host}.cfg, parsed into metrics, evaluated against thresholds, and written to SUMMARY/YYYY-MM/HEALTH_timestamp.csv and .html.

Add --notify to send the report through enabled notification channels — HTML table via email, text summary via Telegram.

What is collected

Vendor	CPU / Memory	Virtual chassis
Cisco IOS-XE / Catalyst	show processes cpu, show processes memory	StackWise member state, StackWise Virtual / VSS VSL links
Cisco Nexus (NX-OS)	show system resources	vPC peer + keep-alive status
Alcatel AOS	show health all cpu, show health all memory	Virtual chassis topology
Alcatel WLC	show cpuload, show memory	—

Command files live at config/{vendor}/healthcheck_pre.txt — edit them to add or remove commands per vendor.

Status evaluation

Status	Meaning
OK	All metrics under thresholds
WARN	CPU or memory over the warn threshold
CRIT	CPU/memory over the crit threshold, or virtual chassis degraded (lost stack member, VSL down, vPC peer down, VC split)
ERROR	Device unreachable or no metrics parsed

Thresholds are configurable in config.yaml:

healthcheck:
  cpu_warn: 80
  cpu_crit: 90
  mem_warn: 85
  mem_crit: 95

Tip

A degraded virtual chassis is always CRIT regardless of CPU/memory — a lost stack member or vPC peer means the redundancy you rely on is already gone.