#!/usr/bin/env bash
# k8s-status.sh — 通过 kubectl 读取 K3s 集群状态，解析输出核心字段
# 用法:
#   bash k8s-status.sh                          # 默认 kubectl + 当前 kubeconfig
#   KUBECONFIG=/etc/rancher/k3s/k3s.yaml bash k8s-status.sh
# 环境变量: KUBECTL（kubectl 路径，默认 kubectl）、KUBECONFIG
set -euo pipefail

KUBECTL_CMD=(kubectl)
if [[ -n "${KUBECTL:-}" ]]; then
  KUBECTL_CMD=("$KUBECTL")
fi
if [[ -n "${KUBECONFIG:-}" ]]; then
  KUBECTL_CMD+=(--kubeconfig "$KUBECONFIG")
fi

echo "===================== K3s 集群状态巡检 ====================="
echo "时间: $(date '+%Y-%m-%d %H:%M:%S %Z')"
echo

echo "────── ① 节点状态（全量 -o wide）──────"
"${KUBECTL_CMD[@]}" get nodes -o wide
echo

echo "────── ② 节点健康解析（核心字段，jsonpath 精确提取）──────"
"${KUBECTL_CMD[@]}" get nodes -o jsonpath='
{range .items[*]}
{ .metadata.name}{"\t"}{ .status.conditions[?(@.type=="Ready")].status}{"\t"}{ .metadata.labels.node-role\.kubernetes\.io/control-plane}{"\t"}{ .status.nodeInfo.kubeletVersion}{"\t"}{ .status.addresses[?(@.type=="InternalIP")].address}{"\t"}{ .status.addresses[?(@.type=="ExternalIP")].address}{"\t"}{ .status.nodeInfo.osImage}{"\t"}{ .status.nodeInfo.kernelVersion}{"\t"}{ .status.nodeInfo.containerRuntimeVersion}{"\n"}
{end}' | awk -F'\t' 'NF {
  role = ($3 == "true") ? "control-plane" : "worker"
  printf "  %-26s %-7s %-14s %-14s 内网IP=%-16s 公网IP=%-16s %s | %s | %s\n", $1, $2, role, $4, $5, $6, $7, $8, $9
}'
echo

echo "────── ③ 节点资源占用（CPU/内存）──────"
if "${KUBECTL_CMD[@]}" top nodes 2>/dev/null; then
  :
else
  echo "  (metrics-server 暂不可用，跳过)"
fi
echo

echo "────── ④ Pod 总数 & 按节点分布 ──────"
TOTAL=$("${KUBECTL_CMD[@]}" get pods -A --no-headers 2>/dev/null | wc -l | tr -d ' ')
echo "  集群 Pod 总数: ${TOTAL:-0}"
"${KUBECTL_CMD[@]}" get pods -A -o wide --no-headers 2>/dev/null | awk '{print $8}' | sort | uniq -c | awk '{printf "  %s: %s 个 pod\n", $2, $1}'
echo

echo "────── ⑤ 异常 Pod（非 Running/Completed）──────"
"${KUBECTL_CMD[@]}" get pods -A --no-headers 2>/dev/null | awk '$4 != "Running" && $4 != "Completed" { printf "  [%s] %s/%s: %s (%s)\n", $1, $2, $3, $4, $5 }'
echo "  (无输出即全部健康)"
echo

echo "────── ⑥ 最近 Warning 事件（最多 5 条）──────"
"${KUBECTL_CMD[@]}" get events -A --field-selector type=Warning --sort-by=.lastTimestamp 2>/dev/null | awk 'NR==1 {next} {print "  " $0}' | tail -5 || true
echo
echo "===================== 巡检完成 ====================="