| name | fukuii-node-health-check |
| description | Produce a comprehensive health verdict for a running Fukuii node — client version, sync state, peer count, latest block height, mining state, and a recent-log scan — rolled into HEALTHY / DEGRADED / ACTION REQUIRED. Use when asked to "check the node", "is my node OK", do a routine/scheduled health check, or verify a node before or after maintenance. Read-only; safe to run anytime. |
Fukuii node health check
Read ../CONVENTIONS.md first (locating the node, calling RPC, output contract).
This skill is read-only — no guarded writes.
When to use
- Routine "is everything OK?" checks, dashboards-by-hand, on-call triage.
- Before starting any maintenance, and again after, to confirm a clean state.
- As the first step when a vaguer problem is reported, to localize it.
Inputs to gather first
Network (etc/mordor/sepolia/…), RPC host/port, datadir. See CONVENTIONS §1.
Procedure (all 🟢 read-only)
- Liveness / identity —
web3_clientVersion, net_version,
net_listening. Node up? Right network?
- Sync —
eth_syncing. false = synced; otherwise inspect
currentBlock vs highestBlock (and SNAP pivot if present). Cross-check
with MCP mcp_sync_status.
- Chain tip —
eth_blockNumber; compare against a known-good external tip
for the network if available. A tip far behind wall-clock = stale.
- Peers —
net_peerCount and admin_peers (or mcp_peer_list). Healthy
ETC node: ~20–40 peers. < 5 peers = degraded, 0 = isolated.
- Mining (only if this node mines) —
eth_mining, miner_getStatus,
eth_hashrate, eth_coinbase. See fukuii-mining-operations for depth.
- Resources —
admin_datadir → then df -h <datadir> for free space; flag
< 10% free. (Hand off to fukuii-disk-management if tight.)
- Log scan — tail
~/.fukuii/<network>/logs/fukuii.log for ERROR/WARN
bursts, stack traces, OutOfMemory, repeated peer drops, RocksDB errors.
Decision guide
| Symptom | Verdict | Hand off to |
|---|
| Synced, peers ≥ 10, no error burst | HEALTHY | — |
| Syncing but progressing, peers OK | DEGRADED (expected) | fukuii-sync-troubleshooting if stalled |
| Peers < 5 / 0 | ACTION REQUIRED | fukuii-peer-management |
| Disk < 10% free | ACTION REQUIRED | fukuii-disk-management |
| Repeated ERROR / OOM / RocksDB corruption | ACTION REQUIRED | fukuii-log-triage |
Deep reference
docs/operations/metrics-and-monitoring.md (Prometheus/JMX/Grafana/Kamon)
docs/operations/LOGGING.md
- The in-process MCP prompt
mcp_node_health_check covers the same intent.
Output
Emit the CONVENTIONS §4 block: Verdict, Evidence (actual numbers/lines),
Actions taken (none — read-only), Recommended next steps (with the exact
hand-off skill and commands).