tags: master, kube-apiserver
本文档讲解部署一个三实例 kube-apiserver 集群的步骤,它们通过 kube-nginx 进行代理访问,从而保证服务可用性。
注意:如果没有特殊指明,本文档的所有操作均在 zhangjun-k8s01 节点上执行,然后远程分发文件和执行命令。
下载最新版本的二进制文件、安装和配置 flanneld 参考:06-1.部署master节点.md
创建证书签名请求:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > kubernetes-csr.json <<EOF
{
"CN": "kubernetes",
"hosts": [
"127.0.0.1",
"172.27.137.240",
"172.27.137.239",
"172.27.137.238",
"${CLUSTER_KUBERNETES_SVC_IP}",
"kubernetes",
"kubernetes.default",
"kubernetes.default.svc",
"kubernetes.default.svc.cluster",
"kubernetes.default.svc.cluster.local."
],
"key": {
"algo": "rsa",
"size": 2048
},
"names": [
{
"C": "CN",
"ST": "BeiJing",
"L": "BeiJing",
"O": "k8s",
"OU": "4Paradigm"
}
]
}
EOF
-
hosts 字段指定授权使用该证书的 IP 和域名列表,这里列出了 master 节点 IP、kubernetes 服务的 IP 和域名;
-
kubernetes 服务 IP 是 apiserver 自动创建的,一般是
--service-cluster-ip-range
参数指定的网段的第一个IP,后续可以通过下面命令获取:$ kubectl get svc kubernetes NAME CLUSTER-IP EXTERNAL-IP PORT(S) AGE kubernetes 10.254.0.1 <none> 443/TCP 1d
生成证书和私钥:
cfssl gencert -ca=/opt/k8s/work/ca.pem \
-ca-key=/opt/k8s/work/ca-key.pem \
-config=/opt/k8s/work/ca-config.json \
-profile=kubernetes kubernetes-csr.json | cfssljson -bare kubernetes
ls kubernetes*pem
将生成的证书和私钥文件拷贝到所有 master 节点:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
ssh root@${node_ip} "mkdir -p /etc/kubernetes/cert"
scp kubernetes*.pem root@${node_ip}:/etc/kubernetes/cert/
done
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > encryption-config.yaml <<EOF
kind: EncryptionConfig
apiVersion: v1
resources:
- resources:
- secrets
providers:
- aescbc:
keys:
- name: key1
secret: ${ENCRYPTION_KEY}
- identity: {}
EOF
将加密配置文件拷贝到 master 节点的 /etc/kubernetes
目录下:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
scp encryption-config.yaml root@${node_ip}:/etc/kubernetes/
done
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > audit-policy.yaml <<EOF
apiVersion: audit.k8s.io/v1beta1
kind: Policy
rules:
# The following requests were manually identified as high-volume and low-risk, so drop them.
- level: None
resources:
- group: ""
resources:
- endpoints
- services
- services/status
users:
- 'system:kube-proxy'
verbs:
- watch
- level: None
resources:
- group: ""
resources:
- nodes
- nodes/status
userGroups:
- 'system:nodes'
verbs:
- get
- level: None
namespaces:
- kube-system
resources:
- group: ""
resources:
- endpoints
users:
- 'system:kube-controller-manager'
- 'system:kube-scheduler'
- 'system:serviceaccount:kube-system:endpoint-controller'
verbs:
- get
- update
- level: None
resources:
- group: ""
resources:
- namespaces
- namespaces/status
- namespaces/finalize
users:
- 'system:apiserver'
verbs:
- get
# Don't log HPA fetching metrics.
- level: None
resources:
- group: metrics.k8s.io
users:
- 'system:kube-controller-manager'
verbs:
- get
- list
# Don't log these read-only URLs.
- level: None
nonResourceURLs:
- '/healthz*'
- /version
- '/swagger*'
# Don't log events requests.
- level: None
resources:
- group: ""
resources:
- events
# node and pod status calls from nodes are high-volume and can be large, don't log responses for expected updates from nodes
- level: Request
omitStages:
- RequestReceived
resources:
- group: ""
resources:
- nodes/status
- pods/status
users:
- kubelet
- 'system:node-problem-detector'
- 'system:serviceaccount:kube-system:node-problem-detector'
verbs:
- update
- patch
- level: Request
omitStages:
- RequestReceived
resources:
- group: ""
resources:
- nodes/status
- pods/status
userGroups:
- 'system:nodes'
verbs:
- update
- patch
# deletecollection calls can be large, don't log responses for expected namespace deletions
- level: Request
omitStages:
- RequestReceived
users:
- 'system:serviceaccount:kube-system:namespace-controller'
verbs:
- deletecollection
# Secrets, ConfigMaps, and TokenReviews can contain sensitive & binary data,
# so only log at the Metadata level.
- level: Metadata
omitStages:
- RequestReceived
resources:
- group: ""
resources:
- secrets
- configmaps
- group: authentication.k8s.io
resources:
- tokenreviews
# Get repsonses can be large; skip them.
- level: Request
omitStages:
- RequestReceived
resources:
- group: ""
- group: admissionregistration.k8s.io
- group: apiextensions.k8s.io
- group: apiregistration.k8s.io
- group: apps
- group: authentication.k8s.io
- group: authorization.k8s.io
- group: autoscaling
- group: batch
- group: certificates.k8s.io
- group: extensions
- group: metrics.k8s.io
- group: networking.k8s.io
- group: policy
- group: rbac.authorization.k8s.io
- group: scheduling.k8s.io
- group: settings.k8s.io
- group: storage.k8s.io
verbs:
- get
- list
- watch
# Default level for known APIs
- level: RequestResponse
omitStages:
- RequestReceived
resources:
- group: ""
- group: admissionregistration.k8s.io
- group: apiextensions.k8s.io
- group: apiregistration.k8s.io
- group: apps
- group: authentication.k8s.io
- group: authorization.k8s.io
- group: autoscaling
- group: batch
- group: certificates.k8s.io
- group: extensions
- group: metrics.k8s.io
- group: networking.k8s.io
- group: policy
- group: rbac.authorization.k8s.io
- group: scheduling.k8s.io
- group: settings.k8s.io
- group: storage.k8s.io
# Default level for all other requests.
- level: Metadata
omitStages:
- RequestReceived
EOF
分发审计策略文件:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
scp audit-policy.yaml root@${node_ip}:/etc/kubernetes/audit-policy.yaml
done
创建证书签名请求:
cat > proxy-client-csr.json <<EOF
{
"CN": "aggregator",
"hosts": [],
"key": {
"algo": "rsa",
"size": 2048
},
"names": [
{
"C": "CN",
"ST": "BeiJing",
"L": "BeiJing",
"O": "k8s",
"OU": "4Paradigm"
}
]
}
EOF
- CN 名称需要位于 kube-apiserver 的
--requestheader-allowed-names
参数中,否则后续访问 metrics 时会提示权限不足。
生成证书和私钥:
cfssl gencert -ca=/etc/kubernetes/cert/ca.pem \
-ca-key=/etc/kubernetes/cert/ca-key.pem \
-config=/etc/kubernetes/cert/ca-config.json \
-profile=kubernetes proxy-client-csr.json | cfssljson -bare proxy-client
ls proxy-client*.pem
将生成的证书和私钥文件拷贝到所有 master 节点:
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
scp proxy-client*.pem root@${node_ip}:/etc/kubernetes/cert/
done
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
cat > kube-apiserver.service.template <<EOF
[Unit]
Description=Kubernetes API Server
Documentation=https://github.com/GoogleCloudPlatform/kubernetes
After=network.target
[Service]
WorkingDirectory=${K8S_DIR}/kube-apiserver
ExecStart=/opt/k8s/bin/kube-apiserver \\
--advertise-address=##NODE_IP## \\
--default-not-ready-toleration-seconds=360 \\
--default-unreachable-toleration-seconds=360 \\
--feature-gates=DynamicAuditing=true \\
--max-mutating-requests-inflight=2000 \\
--max-requests-inflight=4000 \\
--default-watch-cache-size=200 \\
--delete-collection-workers=2 \\
--encryption-provider-config=/etc/kubernetes/encryption-config.yaml \\
--etcd-cafile=/etc/kubernetes/cert/ca.pem \\
--etcd-certfile=/etc/kubernetes/cert/kubernetes.pem \\
--etcd-keyfile=/etc/kubernetes/cert/kubernetes-key.pem \\
--etcd-servers=${ETCD_ENDPOINTS} \\
--bind-address=##NODE_IP## \\
--secure-port=6443 \\
--tls-cert-file=/etc/kubernetes/cert/kubernetes.pem \\
--tls-private-key-file=/etc/kubernetes/cert/kubernetes-key.pem \\
--insecure-port=0 \\
--audit-dynamic-configuration \\
--audit-log-maxage=15 \\
--audit-log-maxbackup=3 \\
--audit-log-maxsize=100 \\
--audit-log-truncate-enabled \\
--audit-log-path=${K8S_DIR}/kube-apiserver/audit.log \\
--audit-policy-file=/etc/kubernetes/audit-policy.yaml \\
--profiling \\
--anonymous-auth=false \\
--client-ca-file=/etc/kubernetes/cert/ca.pem \\
--enable-bootstrap-token-auth \\
--requestheader-allowed-names="aggregator" \\
--requestheader-client-ca-file=/etc/kubernetes/cert/ca.pem \\
--requestheader-extra-headers-prefix="X-Remote-Extra-" \\
--requestheader-group-headers=X-Remote-Group \\
--requestheader-username-headers=X-Remote-User \\
--service-account-key-file=/etc/kubernetes/cert/ca.pem \\
--authorization-mode=Node,RBAC \\
--runtime-config=api/all=true \\
--enable-admission-plugins=NodeRestriction \\
--allow-privileged=true \\
--apiserver-count=3 \\
--event-ttl=168h \\
--kubelet-certificate-authority=/etc/kubernetes/cert/ca.pem \\
--kubelet-client-certificate=/etc/kubernetes/cert/kubernetes.pem \\
--kubelet-client-key=/etc/kubernetes/cert/kubernetes-key.pem \\
--kubelet-https=true \\
--kubelet-timeout=10s \\
--proxy-client-cert-file=/etc/kubernetes/cert/proxy-client.pem \\
--proxy-client-key-file=/etc/kubernetes/cert/proxy-client-key.pem \\
--service-cluster-ip-range=${SERVICE_CIDR} \\
--service-node-port-range=${NODE_PORT_RANGE} \\
--logtostderr=true \\
--v=2
Restart=on-failure
RestartSec=10
Type=notify
LimitNOFILE=65536
[Install]
WantedBy=multi-user.target
EOF
--advertise-address
:apiserver 对外通告的 IP(kubernetes 服务后端节点 IP);--default-*-toleration-seconds
:设置节点异常相关的阈值;--max-*-requests-inflight
:请求相关的最大阈值;--etcd-*
:访问 etcd 的证书和 etcd 服务器地址;--experimental-encryption-provider-config
:指定用于加密 etcd 中 secret 的配置;--bind-address
: https 监听的 IP,不能为127.0.0.1
,否则外界不能访问它的安全端口 6443;--secret-port
:https 监听端口;--insecure-port=0
:关闭监听 http 非安全端口(8080);--tls-*-file
:指定 apiserver 使用的证书、私钥和 CA 文件;--audit-*
:配置审计策略和审计日志文件相关的参数;--client-ca-file
:验证 client (kue-controller-manager、kube-scheduler、kubelet、kube-proxy 等)请求所带的证书;--enable-bootstrap-token-auth
:启用 kubelet bootstrap 的 token 认证;--requestheader-*
:kube-apiserver 的 aggregator layer 相关的配置参数,proxy-client & HPA 需要使用;--requestheader-client-ca-file
:用于签名--proxy-client-cert-file
和--proxy-client-key-file
指定的证书;在启用了 metric aggregator 时使用;--requestheader-allowed-names
:不能为空,值为逗号分割的--proxy-client-cert-file
证书的 CN 名称,这里设置为 "aggregator";--service-account-key-file
:签名 ServiceAccount Token 的公钥文件,kube-controller-manager 的--service-account-private-key-file
指定私钥文件,两者配对使用;--runtime-config=api/all=true
: 启用所有版本的 APIs,如 autoscaling/v2alpha1;--authorization-mode=Node,RBAC
、--anonymous-auth=false
: 开启 Node 和 RBAC 授权模式,拒绝未授权的请求;--enable-admission-plugins
:启用一些默认关闭的 plugins;--allow-privileged
:运行执行 privileged 权限的容器;--apiserver-count=3
:指定 apiserver 实例的数量;--event-ttl
:指定 events 的保存时间;--kubelet-*
:如果指定,则使用 https 访问 kubelet APIs;需要为证书对应的用户(上面 kubernetes*.pem 证书的用户为 kubernetes) 用户定义 RBAC 规则,否则访问 kubelet API 时提示未授权;--proxy-client-*
:apiserver 访问 metrics-server 使用的证书;--service-cluster-ip-range
: 指定 Service Cluster IP 地址段;--service-node-port-range
: 指定 NodePort 的端口范围;
如果 kube-apiserver 机器没有运行 kube-proxy,则还需要添加 --enable-aggregator-routing=true
参数;
关于 --requestheader-XXX
相关参数,参考:
- https://github.com/kubernetes-incubator/apiserver-builder/blob/master/docs/concepts/auth.md
- https://docs.bitnami.com/kubernetes/how-to/configure-autoscaling-custom-metrics/
注意:
- requestheader-client-ca-file 指定的 CA 证书,必须具有 client auth and server auth;
- 如果
--requestheader-allowed-names
为空,或者--proxy-client-cert-file
证书的 CN 名称不在 allowed-names 中,则后续查看 node 或 pods 的 metrics 失败,提示:
[root@zhangjun-k8s01 1.8+]# kubectl top nodes
Error from server (Forbidden): nodes.metrics.k8s.io is forbidden: User "aggregator" cannot list resource "nodes" in API group "metrics.k8s.io" at the cluster scope
替换模板文件中的变量,为各节点生成 systemd unit 文件:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for (( i=0; i < 3; i++ ))
do
sed -e "s/##NODE_NAME##/${NODE_NAMES[i]}/" -e "s/##NODE_IP##/${NODE_IPS[i]}/" kube-apiserver.service.template > kube-apiserver-${NODE_IPS[i]}.service
done
ls kube-apiserver*.service
- NODE_NAMES 和 NODE_IPS 为相同长度的 bash 数组,分别为节点名称和对应的 IP;
分发生成的 systemd unit 文件:
cd /opt/k8s/work
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
scp kube-apiserver-${node_ip}.service root@${node_ip}:/etc/systemd/system/kube-apiserver.service
done
- 文件重命名为 kube-apiserver.service;
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
ssh root@${node_ip} "mkdir -p ${K8S_DIR}/kube-apiserver"
ssh root@${node_ip} "systemctl daemon-reload && systemctl enable kube-apiserver && systemctl restart kube-apiserver"
done
- 启动服务前必须先创建工作目录;
source /opt/k8s/bin/environment.sh
for node_ip in ${NODE_IPS[@]}
do
echo ">>> ${node_ip}"
ssh root@${node_ip} "systemctl status kube-apiserver |grep 'Active:'"
done
确保状态为 active (running)
,否则查看日志,确认原因:
journalctl -u kube-apiserver
source /opt/k8s/bin/environment.sh
ETCDCTL_API=3 etcdctl \
--endpoints=${ETCD_ENDPOINTS} \
--cacert=/opt/k8s/work/ca.pem \
--cert=/opt/k8s/work/etcd.pem \
--key=/opt/k8s/work/etcd-key.pem \
get /registry/ --prefix --keys-only
$ kubectl cluster-info
Kubernetes master is running at https://127.0.0.1:8443
To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.
$ kubectl get all --all-namespaces
NAMESPACE NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
default service/kubernetes ClusterIP 10.254.0.1 <none> 443/TCP 12m
$ kubectl get componentstatuses
NAME STATUS MESSAGE ERROR
controller-manager Unhealthy Get http://127.0.0.1:10252/healthz: dial tcp 127.0.0.1:10252: connect: connection refused
scheduler Unhealthy Get http://127.0.0.1:10251/healthz: dial tcp 127.0.0.1:10251: connect: connection refused
etcd-0 Healthy {"health":"true"}
etcd-2 Healthy {"health":"true"}
etcd-1 Healthy {"health":"true"}
-
如果执行 kubectl 命令式时输出如下错误信息,则说明使用的
~/.kube/config
文件不对,请切换到正确的账户后再执行该命令:The connection to the server localhost:8080 was refused - did you specify the right host or port?
-
执行
kubectl get componentstatuses
命令时,apiserver 默认向 127.0.0.1 发送请求。当 controller-manager、scheduler 以集群模式运行时,有可能和 kube-apiserver 不在一台机器上,这时 controller-manager 或 scheduler 的状态为 Unhealthy,但实际上它们工作正常。
$ sudo netstat -lnpt|grep kube
tcp 0 0 172.27.137.240:6443 0.0.0.0:* LISTEN 101442/kube-apiserv
- 6443: 接收 https 请求的安全端口,对所有请求做认证和授权;
- 由于关闭了非安全端口,故没有监听 8080;
在执行 kubectl exec、run、logs 等命令时,apiserver 会将请求转发到 kubelet 的 https 端口。这里定义 RBAC 规则,授权 apiserver 使用的证书(kubernetes.pem)用户名(CN:kuberntes)访问 kubelet API 的权限:
kubectl create clusterrolebinding kube-apiserver:kubelet-apis --clusterrole=system:kubelet-api-admin --user kubernetes