managing-etcd

Use when working with Etcd — etcd cluster management, member health monitoring, key-space analysis, and Raft consensus diagnostics. Covers endpoint health, leader election status, alarm states, compac

SKILL.md

GitHub

etcd Management Skill

Monitor, analyze, and optimize etcd clusters safely.

MANDATORY: Discovery-First Pattern

Always check cluster health and member list before any key operations. Never assume key prefixes or cluster topology.

Phase 1: Discovery

#!/bin/bash

ETCD_ENDPOINTS="${ETCD_ENDPOINTS:-localhost:2379}"
ETCD_OPTS="${ETCD_CACERT:+--cacert=$ETCD_CACERT} ${ETCD_CERT:+--cert=$ETCD_CERT} ${ETCD_KEY:+--key=$ETCD_KEY}"

echo "=== Endpoint Health ==="
etcdctl endpoint health --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS 2>&1

echo ""
echo "=== Endpoint Status ==="
etcdctl endpoint status --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS -w table 2>&1

echo ""
echo "=== Member List ==="
etcdctl member list --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS -w table 2>&1

echo ""
echo "=== Alarms ==="
etcdctl alarm list --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS 2>&1

echo ""
echo "=== Version ==="
etcdctl version 2>&1

Phase 1 outputs: Endpoint health, leader ID, Raft term/index, member list, active alarms.

Phase 2: Analysis

#!/bin/bash

ETCD_ENDPOINTS="${ETCD_ENDPOINTS:-localhost:2379}"
ETCD_OPTS="${ETCD_CACERT:+--cacert=$ETCD_CACERT} ${ETCD_CERT:+--cert=$ETCD_CERT} ${ETCD_KEY:+--key=$ETCD_KEY}"

echo "=== DB Size per Member ==="
etcdctl endpoint status --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS -w json 2>/dev/null | python3 -c "
import sys, json
data = json.load(sys.stdin)
for ep in (data if isinstance(data, list) else [data]):
    status = ep.get('Status', ep)
    print(f\"  Endpoint: {ep.get('Endpoint','?')} | DB size: {status.get('dbSize',0)//1048576}MB | Leader: {status.get('leader','?')} | RaftIndex: {status.get('raftIndex','?')}\")
" 2>/dev/null

echo ""
echo "=== Key Count by Prefix (top-level) ==="
etcdctl get / --prefix --keys-only --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS 2>/dev/null | \
    awk -F'/' '{if(NF>1) print "/"$2}' | sort | uniq -c | sort -rn | head -15

echo ""
echo "=== Active Leases ==="
etcdctl lease list --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS 2>/dev/null | head -10

echo ""
echo "=== Auth Status ==="
etcdctl auth status --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS 2>&1

echo ""
echo "=== Compaction Info ==="
etcdctl endpoint status --endpoints="$ETCD_ENDPOINTS" $ETCD_OPTS -w json 2>/dev/null | python3 -c "
import sys, json
data = json.load(sys.stdin)
for ep in (data if isinstance(data, list) else [data]):
    status = ep.get('Status', ep)
    print(f\"  Raft term: {status.get('raftTerm','?')} | Applied: {status.get('raftApplied', status.get('raftIndex','?'))}\")
" 2>/dev/null

Output Format

ETCD ANALYSIS
=============
Members: [count] | Leader: [member_id]
DB Size: [size] | Raft Term: [term] | Alarms: [count]

ISSUES FOUND:
- [issue with affected member/alarm]

RECOMMENDATIONS:
- [actionable recommendation]

Anti-Hallucination Rules

  1. NEVER assume resource names — always discover via CLI/API in Phase 1 before referencing in Phase 2.
  2. NEVER fabricate metric names or dimensions — verify against the service documentation or --help output.
  3. NEVER mix CLI commands between service versions — confirm which version/API you are targeting.
  4. ALWAYS use the discovery → verify → analyze chain — every resource referenced must have been discovered first.
  5. ALWAYS handle empty results gracefully — an empty response is valid data, not an error to retry.

Counter-Rationalizations

Shortcut Counter Why
"I'll skip discovery and check known resources" Always run Phase 1 discovery first Resource names change, new resources appear — assumed names cause errors
"The user only asked for a quick check" Follow the full discovery → analysis flow Quick checks miss critical issues; structured analysis catches silent failures
"Default configuration is probably fine" Audit configuration explicitly Defaults often leave logging, security, and optimization features disabled
"Metrics aren't needed for this" Always check relevant metrics when available API/CLI responses show current state; metrics reveal trends and intermittent issues
"I don't have access to that" Try the command and report the actual error Assumed permission failures prevent useful investigation; actual errors are informative