Files
LokiStack/kubernetes/install.sh
T
opsx 9a46a955d4 feat(k8s): 一键部署脚本 + 跨云适配 + 云相关清单拆分到 examples
新增 kubernetes/install.sh(一条命令部署):
- 云平台识别:读节点 providerID,识别 TKE / EKS / ACK / CCE / GKE / AKS / K3s / 自建
- StorageClass 三级探测:显式 STORAGE_CLASS → is-default-class 注解 → 集群唯一 SC
- 自建集群兜底:无任何 SC 时自动创建静态 local PV
  (hostPath + DirectoryOrCreate + nodeAffinity + WaitForFirstConsumer + Retain)
- Secret 幂等生成:已存在则跳过,绝不覆盖(避免密钥轮换导致历史凭证无法解密)
- 失败分类诊断:ImagePullBackOff / CreateContainerConfigError / PVC Pending
  按云给出对应仓库格式与推荐 SC 名
- 环境变量:WITH_NODEPORT / STORAGE_CLASS / LOCAL_STORAGE / NODE_NAME / PV_PATH / PV_SIZE

云相关清单拆到 kubernetes/examples/:
- ingress.yaml(阿里云 alb)、ingress-tke.yaml(腾讯云 qcloud)、secret.example.yaml
- 原因:留在根目录会导致 kubectl apply -f kubernetes/ 时 ingress 同名互相覆盖,
  且 secret.example.yaml 的占位符会把真实密钥顶掉

README 重写:
- 全部用法速查 + 环境变量一览
- 自建 K8s 存储方案选型(local PV / local-path-provisioner / OpenEBS / Longhorn / NFS)
- hostPath 数据运维:定位、热备份、跨节点迁移、卸载保留、常见坑
2026-10-26 14:03:39 +00:00

334 lines
14 KiB
Bash
Executable File
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
#!/usr/bin/env bash
# ============================================================================
# opsx-loki 一键部署
#
# 自动适配:阿里云 ACK / 腾讯云 TKE / AWS EKS / 华为云 CCE / GCP GKE / 自建 K8s
# 命名空间固定:log
# 幂等:重复执行不覆盖密钥、不重建数据卷
#
# 用法:
# bash kubernetes/install.sh # 自动探测
# WITH_NODEPORT=1 bash kubernetes/install.sh # 同时开 NodePort(配 ingress)
# STORAGE_CLASS=local-path bash kubernetes/install.sh # 手动指定 StorageClass
# ============================================================================
set -euo pipefail
NS="log"
WITH_NODEPORT="${WITH_NODEPORT:-0}"
STORAGE_CLASS="${STORAGE_CLASS:-}"
DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
info() { printf '\033[1;36m==>\033[0m %s\n' "$*"; }
ok() { printf '\033[1;32m✅\033[0m %s\n' "$*"; }
warn() { printf '\033[1;33m⚠️ \033[0m %s\n' "$*"; }
die() { printf '\033[1;31m❌\033[0m %s\n' "$*" >&2; exit 1; }
# ============================================================================
# 0) 前置检查
# ============================================================================
command -v kubectl >/dev/null 2>&1 || die "未找到 kubectl,请先安装"
kubectl version --request-timeout=5s >/dev/null 2>&1 || die "kubectl 无法连接集群,请检查 kubeconfig"
# ============================================================================
# 0.1) 云平台识别(信息展示 + 后续诊断用)
# ============================================================================
detect_cloud() {
local pid=""
pid=$(kubectl get nodes -o jsonpath='{.items[0].spec.providerID}' 2>/dev/null || true)
case "${pid}" in
qcloud://*) echo "腾讯云 TKE" ;;
aws://*) echo "AWS EKS" ;;
aliyun://*) echo "阿里云 ACK" ;;
huaweicloud*|huawei*) echo "华为云 CCE" ;;
gce://*) echo "GCP GKE" ;;
azure://*) echo "Azure AKS" ;;
k3s://*) echo "K3s(自建)" ;;
*) echo "自建 / 其他 K8s" ;;
esac
}
# ============================================================================
# 0.2) StorageClass 探测
# 优先级:显式 STORAGE_CLASS > default 注解 > 唯一 SC > 失败
# 注意:结果写入全局变量(RESOLVED_SC / SC_ERROR),不要用 $(...) 调用,
# 否则子 shell 里的赋值无法回传
# ============================================================================
RESOLVED_SC=""
SC_ERROR=""
resolve_storage_class() {
local sc="" all="" count=0
RESOLVED_SC=""
SC_ERROR=""
# 1) 显式指定
if [ -n "${STORAGE_CLASS}" ]; then
if ! kubectl get sc "${STORAGE_CLASS}" >/dev/null 2>&1; then
SC_ERROR="指定的 StorageClass '${STORAGE_CLASS}' 不存在。现有:$(kubectl get sc -o jsonpath='{range .items[*]}{.metadata.name}{" "}{end}' 2>/dev/null)"
return 1
fi
RESOLVED_SC="${STORAGE_CLASS}"; return 0
fi
# 2) 带 default 注解的 SC
sc=$(kubectl get sc -o jsonpath='{range .items[?(@.metadata.annotations.storageclass\.kubernetes\.io/is-default-class=="true")]}{.metadata.name}{"\n"}{end}' 2>/dev/null | head -1 || true)
if [ -n "${sc}" ]; then RESOLVED_SC="${sc}"; return 0; fi
# 3) 集群里只有一个 SC → 直接用它
all=$(kubectl get sc -o jsonpath='{range .items[*]}{.metadata.name}{"\n"}{end}' 2>/dev/null || true)
count=$(printf '%s' "${all}" | grep -c . || true)
if [ "${count}" = "1" ]; then RESOLVED_SC="${all}"; return 0; fi
# 4) 无法确定
return 1
}
# ============================================================================
# 0.3) 自建集群兜底:无任何 StorageClass 时,创建「静态 local PV」
#
# 设计取舍:
# - 零外部依赖:不拉 local-path-provisioner 镜像(国内网络常常拉不到)
# - hostPath + DirectoryOrCreate:kubelet 自动建目录,无需预先 mkdir
# - nodeAffinity 锁节点:Pod 永远调度到数据所在节点
# - WaitForFirstConsumer:先定 Pod 再绑 PV,避免跨节点错绑
# - reclaimPolicy: Retain:误删 PVC 不丢数据
#
# 代价:数据固定在一台节点,节点损毁需从备份恢复(README 有 CSI 备份说明)
# ============================================================================
LOCAL_SC_NAME="opsxloki-local"
LOCAL_PV_NAME="opsxloki-data-pv"
LOCAL_PV_SIZE="${PV_SIZE:-30Gi}"
LOCAL_PV_PATH="${PV_PATH:-/var/lib/opsxloki-data}"
pick_local_node() {
if [ -n "${NODE_NAME:-}" ]; then echo "${NODE_NAME}"; return 0; fi
kubectl get nodes --no-headers \
-o custom-columns=NAME:.metadata.name,READY:.status.conditions[-1].status 2>/dev/null \
| awk '$2=="True"{print $1; exit}'
}
create_local_storage() {
local node
node="$(pick_local_node)"
[ -n "${node}" ] || { warn "找不到 Ready 节点,无法创建本地存储"; return 1; }
info "创建节点本地存储:节点=${node} 路径=${LOCAL_PV_PATH} 容量=${LOCAL_PV_SIZE}"
cat <<EOF | kubectl apply -f - >/dev/null
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: ${LOCAL_SC_NAME}
provisioner: kubernetes.io/no-provisioner
volumeBindingMode: WaitForFirstConsumer
reclaimPolicy: Retain
EOF
cat <<EOF | kubectl apply -f - >/dev/null
apiVersion: v1
kind: PersistentVolume
metadata:
name: ${LOCAL_PV_NAME}
labels:
app: opsx-loki
spec:
capacity:
storage: ${LOCAL_PV_SIZE}
accessModes:
- ReadWriteOnce
persistentVolumeReclaimPolicy: Retain
storageClassName: ${LOCAL_SC_NAME}
hostPath:
path: ${LOCAL_PV_PATH}
type: DirectoryOrCreate
nodeAffinity:
required:
nodeSelectorTerms:
- matchExpressions:
- key: kubernetes.io/hostname
operator: In
values:
- ${node}
EOF
ok "已创建 StorageClass ${LOCAL_SC_NAME} + PV ${LOCAL_PV_NAME}(节点 ${node})"
warn "注意:数据固定在该节点 ${LOCAL_PV_PATH},Pod 不会漂移到其他节点"
RESOLVED_SC="${LOCAL_SC_NAME}"
return 0
}
CLOUD="$(detect_cloud)"
info "opsx-loki 部署开始(namespace: ${NS})"
info "集群: $(kubectl config current-context 2>/dev/null || echo unknown)"
info "平台: ${CLOUD}"
# LOCAL_STORAGE 三态:auto(默认,无 SC 时自动建)/ off(只提示)/ force(强制建,测试用)
LOCAL_STORAGE="${LOCAL_STORAGE:-auto}"
if [ "${LOCAL_STORAGE}" = "force" ]; then
warn "LOCAL_STORAGE=force:强制使用节点本地存储"
create_local_storage || die "创建节点本地存储失败"
elif ! resolve_storage_class; then
if [ -n "${SC_ERROR}" ]; then
die "${SC_ERROR}"
fi
echo ""
if [ "${LOCAL_STORAGE}" = "auto" ]; then
warn "集群没有可用 StorageClass(自建集群常见)→ 自动启用节点本地存储"
create_local_storage || die "创建节点本地存储失败"
else
warn "未找到可用的 StorageClass,且 LOCAL_STORAGE=off"
echo ""
echo " 现有 StorageClass:"
kubectl get sc 2>/dev/null | sed 's/^/ /' || echo " (无)"
echo ""
echo " 解决办法(选一个):"
echo " 1) 手动指定(推荐,若已有可用 SC):"
echo " STORAGE_CLASS=<sc名称> bash kubernetes/install.sh"
echo ""
echo " 2) 标记某个 SC 为默认:"
echo " kubectl patch sc <sc名称> -p '{\"metadata\":{\"annotations\":{\"storageclass.kubernetes.io/is-default-class\":\"true\"}}}'"
echo ""
echo " 3) 自建集群安装 local-path-provisioner(动态供给,支持多 PVC):"
echo " kubectl apply -f https://raw.githubusercontent.com/rancher/local-path-provisioner/master/deploy/local-path-storage.yaml"
echo " STORAGE_CLASS=local-path bash kubernetes/install.sh"
echo ""
echo " 4) 让脚本用节点本地存储(零依赖):"
echo " 去掉 LOCAL_STORAGE=off 重跑即可"
echo ""
die "StorageClass 未就绪,部署中止(避免 PVC 永久 Pending)"
fi
else
ok "StorageClass: ${RESOLVED_SC}"
fi
# ============================================================================
# 1) 命名空间
# ============================================================================
kubectl apply -f "${DIR}/namespace.yaml"
# ============================================================================
# 2) Secret(幂等:已存在则跳过,绝不覆盖)
# ============================================================================
rand_hex() {
if command -v openssl >/dev/null 2>&1; then
openssl rand -hex 32
elif [ -c /dev/urandom ]; then
head -c 32 /dev/urandom | od -An -tx1 | tr -d ' \n'
else
die "无法生成随机密钥(缺 openssl 且无 /dev/urandom)"
fi
}
if kubectl -n "${NS}" get secret opsxloki-secrets >/dev/null 2>&1; then
info "Secret opsxloki-secrets 已存在,跳过(不覆盖现有密钥)"
else
info "生成随机密钥 JWT_SECRET / ENCRYPTION_KEY"
kubectl -n "${NS}" create secret generic opsxloki-secrets \
--from-literal=JWT_SECRET="$(rand_hex)" \
--from-literal=ENCRYPTION_KEY="$(rand_hex)" >/dev/null
ok "Secret 已创建"
fi
# ============================================================================
# 3) 数据卷(显式写入探测到的 StorageClass)
# ============================================================================
RENDER_DIR="$(mktemp -d)"
trap 'rm -rf "${RENDER_DIR}"' EXIT
awk -v sc="${RESOLVED_SC}" '
/^spec:/ && !done { print; print " storageClassName: " sc; done=1; next }
{ print }
' "${DIR}/pvc.yaml" > "${RENDER_DIR}/pvc.yaml"
kubectl apply -f "${RENDER_DIR}/pvc.yaml"
# ============================================================================
# 4) 工作负载 + 服务
# ============================================================================
kubectl apply -f "${DIR}/deployment.yaml"
kubectl apply -f "${DIR}/service.yaml"
if [ "${WITH_NODEPORT}" = "1" ]; then
kubectl apply -f "${DIR}/service-nodeport.yaml"
ok "NodePort Service 已创建(30080)"
fi
# ============================================================================
# 5) 等待就绪(失败时给出按云定制的诊断)
# ============================================================================
info "等待 Deployment 就绪(最多 180 秒)..."
if ! kubectl -n "${NS}" rollout status deployment/opsx-loki --timeout=180s; then
echo ""
warn "部署未在 180 秒内就绪,开始诊断:"
echo ""
kubectl -n "${NS}" get po -l app=opsx-loki -o wide 2>/dev/null | sed 's/^/ /'
REASON=$(kubectl -n "${NS}" get po -l app=opsx-loki -o jsonpath='{.items[0].status.containerStatuses[0].state.waiting.reason}' 2>/dev/null || true)
echo ""
case "${REASON}" in
ImagePullBackOff|ErrImagePull)
warn "镜像拉取失败(${REASON})"
echo " → 当前镜像: $(grep -E '^\s+image:' "${DIR}/deployment.yaml" | head -1 | awk '{print $2}')"
echo " → 该节点拉不到 Docker Hub。请改为目标云可拉的仓库后重跑:"
echo " 腾讯云 TCR : ccr.ccs.tencentyun.com/<ns>/opsxloki:latest"
echo " 阿里云 ACR : registry.cn-hangzhou.aliyuncs.com/<ns>/opsxloki:latest"
echo " 华为云 SWR : swr.cn-north-4.myhuaweicloud.com/<ns>/opsxloki:latest"
echo " 自建 Harbor: harbor.your.domain/library/opsxloki:latest"
;;
CreateContainerConfigError)
warn "容器配置缺失(通常是 Secret 不存在)"
echo " → 检查: kubectl -n ${NS} get secret opsxloki-secrets"
;;
"")
PVC_PHASE=$(kubectl -n "${NS}" get pvc opsxloki-data -o jsonpath='{.status.phase}' 2>/dev/null || echo unknown)
if [ "${PVC_PHASE}" != "Bound" ]; then
warn "PVC 未绑定(phase=${PVC_PHASE})"
echo " → 检查 StorageClass: kubectl -n ${NS} describe pvc opsxloki-data"
echo " → ${CLOUD} 建议的 SC:"
case "${CLOUD}" in
腾讯云*) echo " cbs" ;;
AWS*) echo " gp3 / gp2" ;;
阿里云*) echo " alicloud-disk-essd" ;;
华为云*) echo " csi-disk" ;;
*) echo " local-path(需装 local-path-provisioner)" ;;
esac
fi
;;
esac
echo ""
kubectl -n "${NS}" describe po -l app=opsx-loki 2>/dev/null | tail -15 | sed 's/^/ /'
die "部署未完成,请按上面提示处理"
fi
ok "Deployment 已就绪"
# ============================================================================
# 6) 输出访问信息
# ============================================================================
POD=$(kubectl -n "${NS}" get po -l app=opsx-loki -o jsonpath='{.items[0].metadata.name}' 2>/dev/null || echo "-")
POD_IP=$(kubectl -n "${NS}" get po -l app=opsx-loki -o jsonpath='{.items[0].status.podIP}' 2>/dev/null || echo "-")
NODE=$(kubectl -n "${NS}" get po -l app=opsx-loki -o jsonpath='{.items[0].spec.nodeName}' 2>/dev/null || echo "-")
echo ""
echo "════════════════════════════════════════════════════════"
ok "opsx-loki 部署完成"
echo "════════════════════════════════════════════════════════"
echo " 平台 : ${CLOUD}"
echo " Namespace : ${NS}"
echo " 存储类 : ${RESOLVED_SC}"
echo " Pod : ${POD} (node=${NODE} ip=${POD_IP})"
echo " Service : opsx-loki.${NS}.svc.cluster.local:8080"
[ "${WITH_NODEPORT}" = "1" ] && echo " NodePort : <任意节点IP>:30080"
echo ""
echo " 初始账号 : admin / admin123 ← 首次登录请立即改密"
echo ""
echo " 本地访问 : kubectl -n ${NS} port-forward svc/opsx-loki 8080:8080"
echo " 浏览器打开 http://localhost:8080"
echo ""
echo " 配 Ingress(可选,先跑 WITH_NODEPORT=1 bash install.sh):"
case "${CLOUD}" in
腾讯云*) echo " kubectl apply -f ${DIR}/examples/ingress-tke.yaml # qcloud class" ;;
阿里云*) echo " kubectl apply -f ${DIR}/examples/ingress.yaml # alb class" ;;
*) echo " 参照 ${DIR}/examples/ 改 ingressClassName 与注解" ;;
esac
echo "════════════════════════════════════════════════════════"