etcd单节点扩容成3节点

背景：

我的集群是通过kubeadm部署的etcd单节点，现需扩容成3节点，达到高可用的目的。

步骤：

1.下载并配置etcdctl

wget https://github.com/etcd-io/etcd/releases/download/v3.5.2/etcd-v3.5.2-linux-amd64.tar.gz
tar xvf etcd-v3.5.2-linux-amd64.tar.gz
cd etcd-v3.5.2-linux-amd64
cp etcdctl /usr/sbin

我的etcd版本是3.5，如果是3.4以下，需要设置
export ETCDCTL_API=3

设置别名
echo "alias etcdctl='etcdctl --endpoints=https://[127.0.0.1]:2379 --cacert=/etc/kubernetes/pki/etcd/ca.crt --cert=/etc/kubernetes/pki/etcd/healthcheck-client.crt --key=/etc/kubernetes/pki/etcd/healthcheck-client.key'" >> /root/.bashrc
source /root/.bashrc
查看集群状态
etcdctl member list
查看endpoint状态
etcdctl endpoint status

2.备份etcd

扩容过程中，需要将原来的etcd库删除，会导致kubernetes集群的master节点信息丢失。
因此在扩容之前，建议使用etcdctl snapshot命令进行备份。或者另建etcd节点，将原来的数据传送过去。这里使用snapshot备份。

etcdctl snapshot save /root/etcd$(date +%Y%m%d_%H%M%S)_snapshot.db
ll etcd20220215_001358_snapshot.db

如需恢复原有集群，使用如下命令：

etcdctl --data-dir=/var/lib/etcd \
--initial-advertise-peer-urls=https://192.168.0.6:2380 \
--initial-cluster=master=https://192.168.0.6:2380 \
--name=master \
snapshot restore etcd20220215_001358_snapshot.db

3.生成并复制证书

安装cfssl

curl -s -L -o /usr/bin/cfssl https://pkg.cfssl.org/R1.2/cfssl_linux-amd64
curl -s -L -o /usr/bin/cfssljson https://pkg.cfssl.org/R1.2/cfssljson_linux-amd64

用以下json文件来生成证书。

cd /etc/kubernets/pki/etcd
cat ca-config.json
{
    "signing": {
        "default": {
            "expiry": "43800h"
        },
        "profiles": {
            "server": {
                "expiry": "43800h",
                "usages": [
                    "signing",
                    "key encipherment",
                    "server auth"
                ]
            },
            "client": {
                "expiry": "43800h",
                "usages": [
                    "signing",
                    "key encipherment",
                    "client auth"
                ]
            },
            "peer": {
                "expiry": "43800h",
                "usages": [
                    "signing",
                    "key encipherment",
                    "server auth",
                    "client auth"
                ]
            }
        }
    }
}
cat ca-csr.json
{
    "CN": "My own CA",
    "key": {
        "algo": "rsa",
        "size": 2048
    },
    "names": [
        {
            "C": "US",
            "L": "CA",
            "O": "My Company Name",
            "ST": "San Francisco",
            "OU": "Org Unit 1",
            "OU": "Org Unit 2"
        }
    ]
}
cat server.json
{
    "CN": "etcd0",
    "hosts": [
        "127.0.0.1",
        "192.168.0.6",
        "192.168.1.151",
        "192.168.1.25"
    ],
    "key": {
        "algo": "ecdsa",
        "size": 256
    },
    "names": [
        {
            "C": "US",
            "L": "CA",
            "ST": "San Francisco"
        }
    ]
}

cat member1.json  # 填本机IP
{
    "CN": "etcd0",
    "hosts": [
        "192.168.0.6"
    ],
    "key": {
        "algo": "ecdsa",
        "size": 256
    },
    "names": [
        {
            "C": "US",
            "L": "CA",
            "ST": "San Francisco"
        }
    ]
}

cat client.json
{
    "CN": "client",
    "hosts": [
       ""
    ],
    "key": {
        "algo": "ecdsa",
        "size": 256
    },
    "names": [
        {
            "C": "US",
            "L": "CA",
            "ST": "San Francisco"
        }
    ]
}

生成证书

生成前请看坑1（-.-!）

cfssl gencert -initca ca-csr.json | cfssljson -bare ca -
cfssl gencert -ca=ca.pem -ca-key=ca-key.pem -config=ca-config.json -profile=server server.json | cfssljson -bare server
cfssl gencert -ca=ca.pem -ca-key=ca-key.pem -config=ca-config.json -profile=peer member1.json | cfssljson -bare member1
cfssl gencert -ca=ca.pem -ca-key=ca-key.pem -config=ca-config.json -profile=client client.json | cfssljson -bare client

将master节点的/etc/kubernetes/pki目录复制到子节点。

scp -r /etc/kubernetes/pki 192.168.1.25:/etc/kubernetes/
scp -r /etc/kubernetes/pki 192.168.1.151:/etc/kubernetes/

在子节点上修改member1.json中的ip，重新生成证书

cfssl gencert -ca=ca.pem -ca-key=ca-key.pem -config=ca-config.json -profile=peer member1.json | cfssljson -bare member1

4.停止apiserver和etcd

mv /etc/kubernetes/manifests /etc/kubernetes/manifests_bak

5.添加节点到etcd集群

添加node1到etcd集群
etcdctl member add node1 --peer-urls=https://192.168.1.151:2380
添加node2到etcd集群
etcdctl member add node2 --peer-urls=https://192.168.1.25:2380

6.复制并编辑etcd.yaml

将etcd.yaml文件放入各个子节点的/etc/kubernetes/manifests目录下，在kubelet启动时将会自动启动，/etc/kubernetes/manifests下的所有*.yaml实例为静态pod。

scp /etc/kubernetes/manifests/etcd.yaml 192.168.1.151:/etc/kubernetes/manifests/
scp /etc/kubernetes/manifests/etcd.yaml 192.168.1.25:/etc/kubernetes/manifests/
在node1和node2上
vim /etc/kubernetes/manifest/etcd.yaml

7.编辑kube-apiserver.yaml

8.开启apiserver和etcd

mv /etc/kubernetes/manifests_bak /etc/kubernetes/manifests

9.查看etcd集群状态

etcdctl member list
etcdctl endpoing health

坑

1.etcd日志报错certificate specifies an incompatible key usage

WARNING: 2022/02/15 13:20:30 grpc: addrConn.createTransport failed to connect to {0.0.0.0:2379  <nil> 0 <nil>}. Err: connection error: desc = "transport: authentication handshake failed: remote error: tls: bad certificate". Reconnecting...
{"level":"warn","ts":"2022/02/15T13:20:30.516+0800","caller":"embed/config_logging.go:198","msg":"rejected connection","remote-addr":"127.0.0.1:47329","server-name":"","error":"tls: failed to verify client certificate: x509: certificate specifies an incompatible key usage"}

原因：

ca-config.json中的server-usages参数中缺少client auth。
ca.crt不止用于server认证也用于client认证。

解决：

修改ca-config.json，添加client auth

2.cluster ID mismatch

原因：

etcd这个报错是因为data-dir目录没有清空，有缓存导致。

解决：

停止删除etcd容器，清空或删除目录，重新生成etcd pod。

3.member 2ce221743acad866 has already been bootstrapped

原因：

当前节点已经添加过集群，etcd配置文件中如果没有配置的话，默认 --initial-cluster-state是new。

解决：

修改etcd.yaml，添加如下字段，删除数据目录，重新生成pod。

--initial-cluster-state=existing

4.x509:certificate signed by unknown authority

原因：

具体原因未知，报错是证书不信任，但是我把证书文件都添加到/etc/ssl/certs/ca-bundle.trust.crt中了。

解决：

参考这里：
https://github.com/etcd-io/etcd/blob/e205d09895e6e9d810a88923a64104474002c0c4/Documentation/op-guide/security.md#example-1-client-to-server-transport-security-with-https

在etcd启动文件中添加 --peer-auto-tls 字段重新生成pod就好了，离谱。

回滚

我这里扩容成功后，restore snapshot没有报错，但是集群没有恢复，查看snapshot status只有2.4m，而备份有41m，不知道是哪里的问题，重新恢复了几次也没有成功。。。估计是扩容有问题生成了一个新的集群，exiting参数也加了。。。只好回滚，再重新扩容了。

分别在各个节点上创建备份目录，停止apiserver和etcd，回滚etcd.yaml和apiserver.yaml，清空etcd数据目录。修改etcdctl环境变量。
etcdctl snapshot restore恢复如果不指定数据目录的话会在当前目录生成一个default.etcd目录，移动 default.etcd/member 到数据目录。注意数据目录的权限必须是700，不然etcd会报错。
重新启动apiserver和etcd。
恢复成功。

回滚后如果有多个网卡，calico会未达终态。
参考这篇文章：
https://wghdr.top/archives/97