calico 配置 BGP Route Reflectors

Calico作为k8s的一个流行网络插件,它依赖BGP路由协议实现集群节点上的POD路由互通;而路由互通的前提是节点间建立 BGP Peer 连接。BGP 路由反射器(Route Reflectors,简称 RR)可以简化集群BGP Peer的连接方式,它是解决BGP扩展性问题的有效方式;具体来说:

  • 没有 RR 时,所有节点之间需要两两建立连接(IBGP全互联),节点数量增加将导致连接数剧增、资源占用剧增
  • 引入 RR 后,其他 BGP 路由器只需要与它建立连接并交换路由信息,节点数量增加连接数只是线性增加,节省系统资源

calico-node 版本 v3.3 开始支持内建路由反射器,非常方便,因此使用 calico 作为网络插件可以支持大规模节点数的K8S集群。

本文档主要讲解配置 BGP Route Reflectors,建议首先阅读 基础calico文档

前提条件

实验环境为按照kubeasz安装的2主2从集群,calico 版本 v3.3.2

$ kubectl get node
NAME           STATUS                     ROLES    AGE    VERSION
192.168.1.1   Ready,SchedulingDisabled   master   178m   v1.13.1
192.168.1.2   Ready,SchedulingDisabled   master   178m   v1.13.1
192.168.1.3   Ready                      node     178m   v1.13.1
192.168.1.4   Ready                      node     178m   v1.13.1
$ kubectl get pod -n kube-system -o wide | grep calico
calico-kube-controllers-77487546bd-jqrlc   1/1     Running   0          179m   192.168.1.3   192.168.1.3   <none>           <none>
calico-node-67t5m                          2/2     Running   0          179m   192.168.1.1   192.168.1.1   <none>           <none>
calico-node-drmhq                          2/2     Running   0          179m   192.168.1.2   192.168.1.2   <none>           <none>
calico-node-rjtkv                          2/2     Running   0          179m   192.168.1.4   192.168.1.4   <none>           <none>
calico-node-xtspl                          2/2     Running   0          179m   192.168.1.3   192.168.1.3   <none>           <none>

查看当前集群中BGP连接情况:可以看到集群中4个节点两两建立了 BGP 连接

$ ansible all -m shell -a '/opt/kube/bin/calicoctl node status'
192.168.1.3 | SUCCESS | rc=0 >>
Calico process is running.

IPv4 BGP status
+--------------+-------------------+-------+----------+-------------+
| PEER ADDRESS |     PEER TYPE     | STATE |  SINCE   |    INFO     |
+--------------+-------------------+-------+----------+-------------+
| 192.168.1.1 | node-to-node mesh | up    | 03:08:20 | Established |
| 192.168.1.2 | node-to-node mesh | up    | 03:08:18 | Established |
| 192.168.1.4 | node-to-node mesh | up    | 03:08:19 | Established |
+--------------+-------------------+-------+----------+-------------+

IPv6 BGP status
No IPv6 peers found.

192.168.1.2 | SUCCESS | rc=0 >>
Calico process is running.

IPv4 BGP status
+--------------+-------------------+-------+----------+-------------+
| PEER ADDRESS |     PEER TYPE     | STATE |  SINCE   |    INFO     |
+--------------+-------------------+-------+----------+-------------+
| 192.168.1.4 | node-to-node mesh | up    | 03:08:17 | Established |
| 192.168.1.3 | node-to-node mesh | up    | 03:08:18 | Established |
| 192.168.1.1 | node-to-node mesh | up    | 03:08:20 | Established |
+--------------+-------------------+-------+----------+-------------+

IPv6 BGP status
No IPv6 peers found.

192.168.1.1 | SUCCESS | rc=0 >>
Calico process is running.

IPv4 BGP status
+--------------+-------------------+-------+----------+-------------+
| PEER ADDRESS |     PEER TYPE     | STATE |  SINCE   |    INFO     |
+--------------+-------------------+-------+----------+-------------+
| 192.168.1.2 | node-to-node mesh | up    | 03:08:21 | Established |
| 192.168.1.3 | node-to-node mesh | up    | 03:08:21 | Established |
| 192.168.1.4 | node-to-node mesh | up    | 03:08:21 | Established |
+--------------+-------------------+-------+----------+-------------+

IPv6 BGP status
No IPv6 peers found.

192.168.1.4 | SUCCESS | rc=0 >>
Calico process is running.

IPv4 BGP status
+--------------+-------------------+-------+----------+-------------+
| PEER ADDRESS |     PEER TYPE     | STATE |  SINCE   |    INFO     |
+--------------+-------------------+-------+----------+-------------+
| 192.168.1.2 | node-to-node mesh | up    | 03:08:17 | Established |
| 192.168.1.3 | node-to-node mesh | up    | 03:08:19 | Established |
| 192.168.1.1 | node-to-node mesh | up    | 03:08:20 | Established |
+--------------+-------------------+-------+----------+-------------+

IPv6 BGP status
No IPv6 peers found.

配置全局禁用全连接(BGP full mesh)

$ cat << EOF | calicoctl create -f -
apiVersion: projectcalico.org/v3
kind: BGPConfiguration
metadata:
  name: default
spec:
  logSeverityScreen: Info
  nodeToNodeMeshEnabled: false
  asNumber: 64512
EOF

上述命令配置完成后,再次使用命令ansible all -m shell -a '/opt/kube/bin/calicoctl node status'查看,可以看到之前所有的bgp连接都消失了。

配置 BGP node 与 Route Reflector 的连接建立规则

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
$ cat << EOF | calicoctl create -f -
kind: BGPPeer
apiVersion: projectcalico.org/v3
metadata:
  name: peer-to-rrs
spec:
  # 规则1:普通 bgp node 与 rr 建立连接
  nodeSelector: "!has(i-am-a-route-reflector)"
  peerSelector: has(i-am-a-route-reflector)

---
kind: BGPPeer
apiVersion: projectcalico.org/v3
metadata:
  name: rr-mesh
spec:
  # 规则2:route reflectors 之间也建立连接
  nodeSelector: has(i-am-a-route-reflector)
  peerSelector: has(i-am-a-route-reflector)
EOF

上述命令配置完成后,使用命令:calicoctl get bgppeer calicoctl get bgppeer rr-mesh -o yaml 检查配置是否正确。

选择并配置 Route Reflector 节点

首先查看当前集群中的节点:

$ calicoctl get node -o wide
NAME     ASN       IPV4              IPV6   
k8s401   (64512)   192.168.1.1/24          
k8s402   (64512)   192.168.1.2/24          
k8s403   (64512)   192.168.1.3/24          
k8s404   (64512)   192.168.1.4/24

可以在集群中选择1个或多个节点作为 rr 节点,这里先选择节点:k8s401

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
# 1.先导出 node k8s401 的配置,准备修改
$ calicoctl get node k8s401 --export -o yaml |tee rr01.yml
apiVersion: projectcalico.org/v3
kind: Node
metadata:
  creationTimestamp: null
  name: k8s401
spec:
  bgp:
    ipv4Address: 192.168.1.1/24
    ipv4IPIPTunnelAddr: 172.20.7.128
  orchRefs:
  - nodeName: 192.168.1.1
    orchestrator: k8s

# 2.修改上述 rr01.yml 的配置如下
apiVersion: projectcalico.org/v3
kind: Node
metadata:
  creationTimestamp: null
  name: k8s401
  labels:
    # 设置标签
    i-am-a-route-reflector: true
spec:
  bgp:
    ipv4Address: 192.168.1.1/24
    ipv4IPIPTunnelAddr: 172.20.7.128
    # 设置集群ID
    routeReflectorClusterID: 224.0.0.1
  orchRefs:
  - nodeName: 192.168.1.1
    orchestrator: k8s

# 3.应用修改后的 rr node 配置
$ calicoctl apply -f rr01.yml

查看增加 rr 之后的bgp 连接情况

$ ansible all -m shell -a '/opt/kube/bin/calicoctl node status'
192.168.1.4 | SUCCESS | rc=0 >>
Calico process is running.

IPv4 BGP status
+--------------+-----------+-------+----------+-------------+
| PEER ADDRESS |   PEER TYPE   | STATE |  SINCE   |    INFO     |
+--------------+-----------+-------+----------+-------------+
| 192.168.1.1 | node specific | up    | 11:02:55 | Established |
+--------------+-----------+-------+----------+-------------+

IPv6 BGP status
No IPv6 peers found.

192.168.1.3 | SUCCESS | rc=0 >>
Calico process is running.

IPv4 BGP status
+--------------+-----------+-------+----------+-------------+
| PEER ADDRESS | PEER TYPE | STATE |  SINCE   |    INFO     |
+--------------+-----------+-------+----------+-------------+
| 192.168.1.1 | node specific | up    | 11:02:55 | Established |
+--------------+-----------+-------+----------+-------------+

IPv6 BGP status
No IPv6 peers found.

192.168.1.1 | SUCCESS | rc=0 >>
Calico process is running.

IPv4 BGP status
+--------------+---------------+-------+----------+-------------+
| PEER ADDRESS |   PEER TYPE   | STATE |  SINCE   |    INFO     |
+--------------+---------------+-------+----------+-------------+
| 192.168.1.2 | node specific | up    | 11:02:55 | Established |
| 192.168.1.3 | node specific | up    | 11:02:55 | Established |
| 192.168.1.4 | node specific | up    | 11:02:55 | Established |
+--------------+---------------+-------+----------+-------------+

IPv6 BGP status
No IPv6 peers found.

192.168.1.2 | SUCCESS | rc=0 >>
Calico process is running.

IPv4 BGP status
+--------------+-----------+-------+----------+-------------+
| PEER ADDRESS | PEER TYPE | STATE |  SINCE   |    INFO     |
+--------------+-----------+-------+----------+-------------+
| 192.168.1.1 | node specific | up    | 11:02:55 | Established |
+--------------+-----------+-------+----------+-------------+

IPv6 BGP status
No IPv6 peers found.

可以看到所有其他节点都与所选rr节点建立bgp连接。

再增加一个 rr 节点

步骤同上述选择第1个 rr 节点,这里省略;添加成功后可以看到所有其他节点都与两个rr节点建立bgp连接,两个rr节点之间也建立bgp连接。

  • 对于节点数较多的K8S集群建议配置3-4个 RR 节点

参考文档

更多 BGP 路由协议相关知识请查阅思科/华为相关网络文档。