链接地址:https://blog.csdn.net/kjh2007abc/article/details/86751730
k8s的网络模型假定了所有Pod都在一个可以直接连通的扁平的网络空间中。这是因为k8s出自Google,而在GCE里面是提供了网络模型作为基础设施的,所以k8s就假定这个网络已经存在。而在大家私有的平台设施里搭建k8s集群,就不能假定这种网络已经存在了。我们需要自己实现这个网络,将不同节点上的Docker容器之间的互相访问先打通,然后运行k8s。
目前已经有多个开源组件支持容器网络模型。本节介绍几个常见的网络组件及其安装配置方法,包括Flannel、Open vSwitch、直接路由和Calico。
1. Flannel
1.1 Flannel通信原理
Flannel之所以可以搭建k8s依赖的底层网络,是因为它能实现以下两点。
(1)它能协助k8s,给每一个Node上的Docker容器分配互相不冲突的IP地址。
(2)它能在这些IP地址之间建立一个覆盖网络(Overlay Network),通过这个覆盖网络,将数据包原封不动地传递到目标容器内。
我们通过下图来看看Flannel是如何实现这两点的。
可以看到,Flannel首先创建了一个名为flannel0的网桥,这个网桥的一端连接docker0网桥,另一端连接一个叫作flanneld的服务进程。
flanneld进程很重要:
flanneld首先要连接etcd,利用etcd来管理可分配的IP地址段资源,同时监控etcd中每个Pod的实际地址,并在内存中建立了一个Pod节点路由表;
然后flanneld进程下连docker0和物理网络,使用内存中的Pod节点路由表,将docker0发给它的数据包包装起来,利用物理网络的连接将数据包投递到目标flanneld上,从而完成Pod到Pod之间的直接地址通信。
Flannel之间的底层通信协议的可选余地很多,有UDP, VxLAN, AWS VPC等多种方式,只要能通到对端的Flannel就可以了。源flanneld加包,目标flanneld解包,最终docker0看到的就是原始的数据,非常透明,根本感觉不到中间Flannel的存在。常用的是UDP。
Flannel是如何做到为不同Node上的Pod分配IP且不产生冲突的?因为Flannel使用集中的etcd服务管理这些地址资源信息,它每次分配的地址段都在同一个公共区域获取,这样自然能随时协调,避免冲突了。在Flannel分配好地址段后,接下来的工作就转交给Docker完成了。Flannel通过修改Docker的启动参数将分配给它的地址段传递进去。
--bip=172.17.18.1/24
通过这些操作,Flannel就控制了每个Node节点上的docker0地址段的地址,也能保障所有Pod的IP地址在同一水平的网络中且不产生冲突了。
Flannel完美地解决了对k8s网络的支持,但是它引入了多个网络组件,在网络通信时需要转到flannel0网络接口,再转到用户态的flanneld程序,到对端后还需要走这个过程的反过程,所以会引入一些网络的延时消耗。
另外,Flannel模型默认使用了UDP作为底层传输协议,UDP协议本身的非可靠性,在大流量、高并发应用场景下还需要反复测试,确保没有问题。
1.2 Flannel的安装和配置方法
1)安装etcd
由于Flannel使用etcd作为数据库,所以需要预先安装好,这里不做描述。
2)安装Flannel
需要在每台Node上都安装Flannel。Flannel软件的下载地址为:https://github.com/coreos/flannel/releases 。将下载好的flannel-<version>-linux-amd64.tar.gz解压,把二进制文件flanneld和mk-docker-opts.sh复制到/usr/bin中,即可完成对Flannel的安装。
3)配置Flannel
此处以使用systemd系统为例对flanneld服务进行配置。
编辑服务配置文件/usr/lib/systemd/system/flanneld.service:
[[email protected] sysconfig]# more /usr/lib/systemd/system/flanneld.service
[Unit]
Description=flanneld overlay address etcd agent
After=network.target
Before=docker.service
[Service]
Type=notify
EnvironmentFile=/etc/sysconfig/flannel
ExecStart=/usr/bin/flanneld -etcd-endpoints=http://10.0.2.15:2379 $FLANNEL_OPTIONS
[Install]
RequiredBy=docker.service
WantedBy=multi-user.target
编辑配置文件/etc/sysconfig/flannel,设置etcd的URL地址:
[[email protected] sysconfig]# more flannel
# flanneld configuration options
# etcd url location. Point this to the server where etcd runs
FLANNEL_ETCD="http://10.0.2.15:2379"
# etcd config key. This is the configuration key that flannel queries
# For address range assignment
FLANNEL_ETCD_KEY="/coreos.com/network"
在启动flanneld服务之前,需要在etcd中添加一条网络配置记录,这个配置将用于flanneld分配给每个Docker的虚拟IP地址段。
[[email protected] ~]# etcdctl set /coreos.com/network/config ‘{ "Network": "172.16.0.0/16" }‘
{ "Network": "172.16.0.0/16" }
由于Flannel将覆盖docker0网桥,所以如果Docker服务已启动,则需要停止Docker服务。
4)启动Flannel服务
systemctl daemon-reload
systemctl restart flanneld
5)重新启动Docker服务
systemctl daemon-reload
systemctl restart docker
6)设置docker0网桥的IP地址
mk-docker-opts.sh -i
source /run/flannel/subnet.env
ifconfig docker0 ${FLANNEL_SUBNET}
完成后确认网络接口docker0的IP属于flannel0的子网:
[[email protected] system]# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: enp0s3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 08:00:27:9f:89:14 brd ff:ff:ff:ff:ff:ff
inet 10.0.2.4/24 brd 10.0.2.255 scope global noprefixroute dynamic enp0s3
valid_lft 993sec preferred_lft 993sec
inet6 fe80::a00:27ff:fe9f:8914/64 scope link
valid_lft forever preferred_lft forever
3: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default
link/ether 02:42:c9:52:3d:15 brd ff:ff:ff:ff:ff:ff
inet 172.16.70.1/24 brd 172.16.70.255 scope global docker0
valid_lft forever preferred_lft forever
inet6 fe80::42:c9ff:fe52:3d15/64 scope link
valid_lft forever preferred_lft forever
6: flannel0: <POINTOPOINT,MULTICAST,NOARP,UP,LOWER_UP> mtu 1472 qdisc pfifo_fast state UNKNOWN group default qlen 500
link/none
inet 172.16.70.0/16 scope global flannel0
valid_lft forever preferred_lft forever
inet6 fe80::4b31:c92f:8cc9:3a22/64 scope link flags 800
valid_lft forever preferred_lft forever
[[email protected] system]#
至此,就完成了Flannel覆盖网络的设置。
使用ping命令验证各Node上docker0之间的相互访问。例如在Node1(docker0 IP=172.16.70.1)机器上ping Node2的docker0(docker0 IP=172.16.13.1),通过Flannel能够成功连接到其他物理机的Docker网络:
[[email protected] system]# ifconfig flannel0
flannel0: flags=4305<UP,POINTOPOINT,RUNNING,NOARP,MULTICAST> mtu 1472
inet 172.16.70.0 netmask 255.255.0.0 destination 172.16.70.0
inet6 fe80::524a:4b9c:3391:7514 prefixlen 64 scopeid 0x20<link>
unspec 00-00-00-00-00-00-00-00-00-00-00-00-00-00-00-00 txqueuelen 500 (UNSPEC)
RX packets 5 bytes 420 (420.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 8 bytes 564 (564.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
[[email protected] system]# ifconfig docker0
docker0: flags=4099<UP,BROADCAST,MULTICAST> mtu 1500
inet 172.16.70.1 netmask 255.255.255.0 broadcast 172.16.70.255
inet6 fe80::42:c9ff:fe52:3d15 prefixlen 64 scopeid 0x20<link>
ether 02:42:c9:52:3d:15 txqueuelen 0 (Ethernet)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 8 bytes 648 (648.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
[[email protected] system]# ping 172.16.13.1
PING 172.16.13.1 (172.16.13.1) 56(84) bytes of data.
64 bytes from 172.16.13.1: icmp_seq=1 ttl=62 time=1.63 ms
64 bytes from 172.16.13.1: icmp_seq=2 ttl=62 time=1.55 ms
^C
--- 172.16.13.1 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1002ms
rtt min/avg/max/mdev = 1.554/1.595/1.637/0.057 ms
我们可以在etcd中查看到Flannel设置的flannel0地址与物理机IP地址的对应规则:
[[email protected] etcd]# etcdctl ls /coreos.com/network/subnets
/coreos.com/network/subnets/172.16.70.0-24
/coreos.com/network/subnets/172.16.13.0-24
[[email protected] etcd]# etcdctl get /coreos.com/network/subnets/172.16.70.0-24
{"PublicIP":"10.0.2.4"}
[[email protected] etcd]# etcdctl get /coreos.com/network/subnets/172.16.13.0-24
{"PublicIP":"10.0.2.5"}
2 Open vSwitch
2.1 基本原理
Open vSwitch是一个开源的虚拟交换软件,有点儿像Linux中的bridge,但是功能要复杂得多。Open vSwitch的网桥可以直接建立多种通信通道(隧道),例如Open vSwitch with GRE/VxLAN。这些通道的建立可以很容易地通过OVS的配置命令实现。在k8s、Docker场景下,我们主要是建立L3到L3的隧道,例如下面样子的网络架构。
首先,为了避免Docker创建的docker0地址产生冲突,我们需要手动配置和指定下各个Node节点上docker0网桥的地址段分布。
其次,建立Open vSwitch的网桥ovs,然后使用ovs-vsctl命令给ovs网桥增加gre端口,添加gre端口时要将目标连接的NodeIP地址设置为对端的IP地址。对每一个对端IP地址都需要这么操作(对于大型网络,需要做自动化脚本来完成)。
最后,将ovs的网桥作为网络接口,加入Docker的网桥上。重启ovs网桥和Docker的网桥,并添加一个Docker的地址段到Docker网桥的路由规则项,就可以将两个容器的网络连接起来了。
2.2 网络通信过程
当容器内的应用访问另一个容器的地址时,数据包会通过容器内的默认路由发送给docker0网桥。ovs的网桥是作为docker0网桥的端口存在的,安会将数据发送给ovs网桥。ovs网络已经通过配置建立了和其他ovs网桥的GRE/VxLAN隧道,自然能将数据送达对端的Node,并送往docker0及Pod。
通过新增的路由项,使用得Node节点本身的应用的数据也路由到docker0网桥上,和刚才的通信过程一样,自然也可以访问其他Node上的Pod。
2.3 OVS with GRE/VxLAN组网方式的特点
OVS的优势是,作为开源虚拟交换机软件,它相对成熟和稳定,支持各类网络隧道协议,经过了OpenStack等项目的考验。
另一方面,相对于Flannel不但可以建立OverlayNetwork,实现Pod到Pod的通信,还和k8s、Docker架构体系紧密结合,感知k8s的Service,动态维护自己的路由表,还通过etcd来协助Docker对整个k8s集群中的docker0的子网地址进行分配。使用OVS时,很多事情就需要手工完成了。
此外,无外是OVS,还是Flannel,通过建立Overlay Network,实现Pod到Pod的通信,都会引入一些额外的通信开销。如果是对网络依赖特别重的应用,则需要评估对业务的影响。
2.4 Open vSwitch的安装与配置
以两个Node为例,目标网络拓扑如下图所示。
1)在两个Node上安装ovs
需要确认下关闭了Node节点上的selinux。
同时在两个Node节点上:
yum -y install openvswitch
查看Open vSwitch服务状态,需要有ovsdb-server与ovs-vswitchd两个进程。
[[email protected] system]# systemctl start openvswitch
[[email protected] system]# systemctl status openvswitch
● openvswitch.service - Open vSwitch
Loaded: loaded (/usr/lib/systemd/system/openvswitch.service; disabled; vendor preset: disabled)
Active: active (exited) since Sun 2018-06-10 17:06:40 CST; 6s ago
Process: 8368 ExecStart=/bin/true (code=exited, status=0/SUCCESS)
Main PID: 8368 (code=exited, status=0/SUCCESS)
Jun 10 17:06:40 k8s-node2.test.com systemd[1]: Starting Open vSwitch...
Jun 10 17:06:40 k8s-node2.test.com systemd[1]: Started Open vSwitch.
[[email protected] system]# ps -ef|grep ovs
root 8352 1 0 17:06 ? 00:00:00 ovsdb-server: monitoring pid 8353 (healthy)
root 8353 8352 0 17:06 ? 00:00:00 ovsdb-server /etc/openvswitch/conf.db -vconsole:emer -vsyslog:err -vfile:info --remote=punix:/var/run/openvswitch/db.sock --private-key=db:Open_vSwitch,SSL,private_key --certificate=db:Open_vSwitch,SSL,certificate --bootstrap-ca-cert=db:Open_vSwitch,SSL,ca_cert --no-chdir --log-file=/var/log/openvswitch/ovsdb-server.log --pidfile=/var/run/openvswitch/ovsdb-server.pid --detach --monitor
root 8364 1 0 17:06 ? 00:00:00 ovs-vswitchd: monitoring pid 8365 (healthy)
root 8365 8364 0 17:06 ? 00:00:00 ovs-vswitchd unix:/var/run/openvswitch/db.sock -vconsole:emer -vsyslog:err -vfile:info --mlockall --no-chdir --log-file=/var/log/openvswitch/ovs-vswitchd.log --pidfile=/var/run/openvswitch/ovs-vswitchd.pid --detach --monitor
2)创建网桥和GRE隧道
接下来需要在每个Node上建立ovs的网桥br0,然后在网桥上创建一个GRE隧道连接对端的网桥,最后把ovs的网桥br0作为一个端口连接到docker0这个Linux网桥上。
这样一来,两个节点机器上的docker0网段就能互通了。
以Node1节点为例,具体操作步骤如下:
(1)创建ovs网桥
[[email protected] system]# ovs-vsctl add-br br0
(2)创建GRE隧道连接对端,remote_ip为对端的eth0网卡地址
[[email protected] system]# ovs-vsctl add-port br0 gre1 -- set interface gre1 type=gre option:remote_ip=10.0.2.5
(3)添加br0到本地docker0,使得容器流量通过OVS流经tunnel
[[email protected] system]# brctl addif docker0 br0
(4)启动br0与docker0网桥
[[email protected] system]# ip link set dev br0 up
[[email protected] system]# ip link set dev docker0 up
(5)添加路由规则
由于10.0.2.5与10.0.2.4的docker0网段分别为172.16.20.0/24与172.16.10.0/24,这两个网段的路由都需要经过本机的docker0网桥路由,其中一个24网段是通过OVS的GRE隧道到达对端的。因此需要在每个Node上添加通过docker0网桥转发的172.16.0.0/16的路由规则:
[[email protected] system]# ip route add 172.16.0.0/16 dev docker0
(6)清空Docker自带的iptables规则及Linux的规则,后者存在拒绝icmp报文通过防火墙的规则
[[email protected] system]# iptables -t nat -F
[[email protected] system]# iptables -F
在Node1节点上完成以上操作后,在Node2节点上进行相同的配置。
配置完成后,Node1节点的IP地址、docker0的IP地址及路由等重要信息显示如下:
[[email protected] system]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: enp0s3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP group default qlen 1000
link/ether 08:00:27:9f:89:14 brd ff:ff:ff:ff:ff:ff
inet 10.0.2.4/24 brd 10.0.2.255 scope global noprefixroute dynamic enp0s3
valid_lft 842sec preferred_lft 842sec
inet6 fe80::a00:27ff:fe9f:8914/64 scope link
valid_lft forever preferred_lft forever
3: docker0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
link/ether 02:42:c9:52:3d:15 brd ff:ff:ff:ff:ff:ff
inet 172.16.10.1/24 brd 172.16.10.255 scope global docker0
valid_lft forever preferred_lft forever
inet6 fe80::42:c9ff:fe52:3d15/64 scope link
valid_lft forever preferred_lft forever
10: ovs-system: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN group default qlen 1000
link/ether 5e:a9:02:75:aa:98 brd ff:ff:ff:ff:ff:ff
11: br0: <BROADCAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master docker0 state UNKNOWN group default qlen 1000
link/ether 82:e3:9a:29:3c:46 brd ff:ff:ff:ff:ff:ff
inet6 fe80::a8de:24ff:fef4:f8ec/64 scope link
valid_lft forever preferred_lft forever
12: [email protected]: <NOARP> mtu 1476 qdisc noop state DOWN group default qlen 1000
link/gre 0.0.0.0 brd 0.0.0.0
13: [email protected]: <BROADCAST,MULTICAST> mtu 1462 qdisc noop state DOWN group default qlen 1000
link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
14: [email protected]: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 65490 qdisc pfifo_fast master ovs-system state UNKNOWN group default qlen 1000
link/ether 76:53:6f:11:e0:f8 brd ff:ff:ff:ff:ff:ff
inet6 fe80::7453:6fff:fe11:e0f8/64 scope link
valid_lft forever preferred_lft forever
[[email protected] system]#
[[email protected] system]# ip route
default via 10.0.2.1 dev enp0s3 proto dhcp metric 100
10.0.2.0/24 dev enp0s3 proto kernel scope link src 10.0.2.4 metric 100
172.16.0.0/16 dev docker0 scope link
172.16.10.0/24 dev docker0 proto kernel scope link src 172.16.10.1
3)两个Node上容器之间的互通测试
[[email protected] system]# ping 172.16.20.1
PING 172.16.20.1 (172.16.20.1) 56(84) bytes of data.
64 bytes from 172.16.20.1: icmp_seq=1 ttl=64 time=2.39 ms
64 bytes from 172.16.20.1: icmp_seq=2 ttl=64 time=3.36 ms
^C
--- 172.16.20.1 ping statistics ---
2 packets transmitted, 2 received, 0% packet loss, time 1004ms
rtt min/avg/max/mdev = 2.398/2.882/3.366/0.484 ms
[[email protected] system]#
在Node2上抓包观察:
[[email protected] system]# tcpdump -i docker0 -nnn
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on docker0, link-type EN10MB (Ethernet), capture size 262144 bytes
23:43:59.020039 IP 172.16.10.1 > 172.16.20.1: ICMP echo request, id 20831, seq 26, length 64
23:43:59.020096 IP 172.16.20.1 > 172.16.10.1: ICMP echo reply, id 20831, seq 26, length 64
23:44:00.020899 IP 172.16.10.1 > 172.16.20.1: ICMP echo request, id 20831, seq 27, length 64
23:44:00.020939 IP 172.16.20.1 > 172.16.10.1: ICMP echo reply, id 20831, seq 27, length 64
23:44:01.021706 IP 172.16.10.1 > 172.16.20.1: ICMP echo request, id 20831, seq 28, length 64
23:44:01.021750 IP 172.16.20.1 > 172.16.10.1: ICMP echo reply, id 20831, seq 28, length 64
接下来我们从前面曾做过的实验中找出来一份创建2实例的RC资源文件来,实际创建两个容器来测试下两个Pods间的网络通信:
[[email protected] ~]# more frontend-rc.yaml
apiVersion: v1
kind: ReplicationController
metadata:
name: frontend
labels:
name: frontend
spec:
replicas: 2
selector:
name: frontend
template:
metadata:
labels:
name: frontend
spec:
containers:
- name: php-redis
image: kubeguide/guestbook-php-frontend
ports:
- containerPort: 80
hostPort: 80
env:
- name: GET_HOSTS_FROM
value: env
[[email protected] ~]#
创建并观察下结果:
[[email protected] ~]# kubectl get rc
NAME DESIRED CURRENT READY AGE
frontend 2 2 2 33m
[[email protected] ~]# kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE
frontend-b6krg 1/1 Running 1 33m 172.16.20.2 10.0.2.5
frontend-qk6zc 1/1 Running 0 33m 172.16.10.2 10.0.2.4
我们继续登录进入Node1节点上的容器内部:
[[email protected] ~]# kubectl exec -it frontend-qk6zc -c php-redis /bin/bash
[email protected]:/var/www/html# ip a
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
2: [email protected]: <NOARP> mtu 1476 qdisc noop state DOWN group default qlen 1000
link/gre 0.0.0.0 brd 0.0.0.0
3: [email protected]: <BROADCAST,MULTICAST> mtu 1462 qdisc noop state DOWN group default qlen 1000
link/ether 00:00:00:00:00:00 brd ff:ff:ff:ff:ff:ff
22: [email protected]: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default
link/ether 02:42:ac:10:0a:02 brd ff:ff:ff:ff:ff:ff
inet 172.16.10.2/24 brd 172.16.10.255 scope global eth0
valid_lft forever preferred_lft forever
[email protected]:/var/www/html#
从Node1上运行的Pod中ping一个Nod2上运行的Pod的地址:
[email protected]:/var/www/html# ping 172.16.20.2
PING 172.16.20.2 (172.16.20.2): 56 data bytes
64 bytes from 172.16.20.2: icmp_seq=0 ttl=63 time=2017.587 ms
64 bytes from 172.16.20.2: icmp_seq=1 ttl=63 time=1014.193 ms
64 bytes from 172.16.20.2: icmp_seq=2 ttl=63 time=13.232 ms
64 bytes from 172.16.20.2: icmp_seq=3 ttl=63 time=1.122 ms
64 bytes from 172.16.20.2: icmp_seq=4 ttl=63 time=1.379 ms
64 bytes from 172.16.20.2: icmp_seq=5 ttl=63 time=1.474 ms
64 bytes from 172.16.20.2: icmp_seq=6 ttl=63 time=1.371 ms
64 bytes from 172.16.20.2: icmp_seq=7 ttl=63 time=1.583 ms
^C--- 172.16.20.2 ping statistics ---
8 packets transmitted, 8 packets received, 0% packet loss
round-trip min/avg/max/stddev = 1.122/381.493/2017.587/701.350 ms
[email protected]:/var/www/html#
在Node2节点上抓包看到数据包交互:
[[email protected] system]# tcpdump -i docker0 -nnn
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on docker0, link-type EN10MB (Ethernet), capture size 262144 bytes
00:13:18.601908 IP 172.16.10.2 > 172.16.20.2: ICMP echo request, id 38, seq 4, length 64
00:13:18.601947 IP 172.16.20.2 > 172.16.10.2: ICMP echo reply, id 38, seq 4, length 64
00:13:18.601956 IP 172.16.20.2 > 172.16.10.2: ICMP echo reply, id 38, seq 4, length 64
00:13:28.609109 IP 172.16.10.2 > 172.16.20.2: ICMP echo request, id 38, seq 5, length 64
00:13:28.609165 IP 172.16.20.2 > 172.16.10.2: ICMP echo reply, id 38, seq 5, length 64
00:13:28.609179 IP 172.16.20.2 > 172.16.10.2: ICMP echo reply, id 38, seq 5, length 64
00:13:29.612564 IP 172.16.10.2 > 172.16.20.2: ICMP echo request, id 38, seq 6, length 64
注:如果以上网络通信测试没有完全成功,不妨检查下Node节点上的firewalld防火墙配置。
至此,基于OVS的网络搭建成功,由于GRE是点对点的隧道通信方式,所以如果有多个Node,则需要建立N*(N-1)条GRE隧道,即所有Node组成一个网状的网络,以实现全网互通。
3 直接路由
我们在前几节的实验中已经测试过通过直接手动写路由的方式,实现Node之间的网络通信功能了,配置方法不再讨论。该直接路由配置方法的问题是,在集群节点发生变化时,需要手动去维护每个Node上的路由表信息,效率很低。为了有效管理这些动态变化的网络路由信息,动态地让其他Node都感知到,就需要使用动态路由发现协议来同步这些变化。
在实现这些动态路由发现协议的开源软件中,常用的有Quagga、Zebra等。
下面简单介绍下配置步骤和注意事项。
(1)仍然需要手动分配每个Node节点上的Docker bridge的地址段
无论是修改默认的docker0使用的地址段,还是另建一个bridge并使用--bridge=XX来指定使用的网桥,都需要确保每个Node上Docker网桥使用的地址段不能重叠。
(2)然后在每个Node上运行Quagga
既可以选择在每台服务器上安装Quagga软件并启动,也可以使用Quagga容器来运行。在每台Node上下载Docker镜像:
# docker pull georce/router
在每台Node上启动Quagga容器,需要说明的是,Quagga需要以--privileged特权模式运行,并且指定--net=host,表示直接使用物理机的网络。
# docker run -itd --name=router --privileged --net=host georce/router
启动成功后,各Node上的Quagga会相互学习来完成到其他机器的docker0路由规则的添加。
至此,所有Node上的docker0都可以互联互通了。
注:如果集群规模在数千台Node以上,则需要测试和评估路由表的效率问题。
4 Calico容器网络和网络策略
4.1 Calico简介
Calico 是容器网络的又一种解决方案,和其他虚拟网络最大的不同是,它没有采用 overlay 网络做报文的转发,提供了纯 3 层的网络模型。三层通信模型表示每个容器都通过 IP 直接通信,中间通过路由转发找到对方。在这个过程中,容器所在的节点类似于传统的路由器,提供了路由查找的功能。要想路由工作能够正常,每个虚拟路由器(容器所在的主机节点)必须有某种方法知道整个集群的路由信息,calico 采用的是 BGP 路由协议,全称是 Border Gateway Protocol。除了能用于 容器集群平台 kubernetes、共有云平台 AWS、GCE 等, 也能很容易地集成到 openstack 等 Iaas 平台。
Calico在每个计算节点利用Linux Kernel实现了一个高效的vRouter来负责数据转发。每个vRouter通过BGP协议把在本节点上运行的容器的路由信息向整个Calico网络广播,并自动设置到达其他节点的路由转发规则。Calico保证所有容器之间的数据流量都是通过IP路由的方式完成互联互通的。Calico节点组网可以直接利用数据中心的网络结构(L2或者L3),不需要额外的NAT、隧道或者Overlay Network,没有额外的封包解包,能够节约CPU运算,提高网络通信效率。Calico的数据包结构示意图如下。
Calico在小规模集群中可以直接互联,在大规模集群中可以通过额外的BGP route reflector来完成。
此外,Calico基于iptables还提供了丰富的网络策略,实现了k8s的Network Policy策略,提供容器间网络可达性限制的功能。
Calico的主要组件如下:
Felix:Calico Agent,运行在每台Node上,负责为容器设置网络源(IP地址、路由规则、iptables规则等),保证主机容器网络互通。
etcd:Calico使用的存储后端。
BGP Client(BIRD):负责把Felix在各Node上设置的路由信息通过BGP协议广播到Calico网络。
BGP Route Reflector(BIRD):通过一个或者多个BGP Route Reflector来完成大规模集群的分级路由分发。
calicoctl:Calico命令行管理工具。
4.2 部署Calico服务
在k8s中部署Calico的主要步骤包括两部分。
4.2.1 修改kubernetes服务的启动参数,并重启服务
设置Master上kube-apiserver服务的启动参数:--allow-privileged=true(因为Calico-node需要以特权模式运行在各Node上)。
设置各Node上kubelet服务的启动参数:--network-plugin=cni(使用CNI网络插件), --allow-privileged=true
本例中的K8s集群包括两台Node:Node1(10.0.2.4)和Node2(10.0.2.5)
4.2.2 创建Calico服务,主要包括Calico-node和Calico policy controller
需要创建出以下的资源对象:
创建ConfigMap calico-config,包含Calico所需的配置参数。
创建Secret calico-etcd-secrets,用于使用TLS方式连接etcd。
在每个Node上运行calico/node容器,部署为DaemonSet。
在每个Node上安装Calico CNI二进制文件和网络配置参数(由install-cni容器完成)。
部署一个名为calico/kube-policy-controller的Deployment,以对接k8s集群中为Pod设置的Network Policy。
4.2.3 Calico服务安装与配置的详细说明
从Calico官网下载Calico的yaml配置文件,下载地址为https://docs.projectcalico.org/v2.1/getting-started/kubernetes/installation/hosted/calico.yaml 。
该配置文件中包括了启动Calico所需的全部资源对象的定义。下面对其逐个进行说明。
(1)Calico所需的配置以ConfigMap对象进行创建,如下所示
# Calico Version v2.1.5
# https://docs.projectcalico.org/v2.1/releases#v2.1.5
# This manifest includes the following component versions:
# calico/node:v1.1.3
# calico/cni:v1.8.0
# calico/kube-policy-controller:v0.5.4
# This ConfigMap is used to configure a self-hosted Calico installation.
kind: ConfigMap
apiVersion: v1
metadata:
name: calico-config
namespace: kube-system
data:
# Configure this with the location of your etcd cluster.
etcd_endpoints: "http://10.0.2.15:2379"
# Configure the Calico backend to use.
calico_backend: "bird"
# The CNI network configuration to install on each node.
cni_network_config: |-
{
"name": "k8s-pod-network",
"type": "calico",
"etcd_endpoints": "__ETCD_ENDPOINTS__",
"etcd_key_file": "__ETCD_KEY_FILE__",
"etcd_cert_file": "__ETCD_CERT_FILE__",
"etcd_ca_cert_file": "__ETCD_CA_CERT_FILE__",
"log_level": "info",
"ipam": {
"type": "calico-ipam"
},
"policy": {
"type": "k8s",
"k8s_api_root": "https://__KUBERNETES_SERVICE_HOST__:__KUBERNETES_SERVICE_PORT__",
"k8s_auth_token": "__SERVICEACCOUNT_TOKEN__"
},
"kubernetes": {
"kubeconfig": "__KUBECONFIG_FILEPATH__"
}
}
# If you‘re using TLS enabled etcd uncomment the following.
# You must also populate the Secret below with these files.
etcd_ca: "" # "/calico-secrets/etcd-ca"
etcd_cert: "" # "/calico-secrets/etcd-cert"
etcd_key: "" # "/calico-secrets/etcd-key"
主要参数如下:
etcd_endpoints:Calico使用etcd来保存网络拓扑和状态,该参数指定etcd的地址,可以使用k8s Master所用的etcd,也可以另外搭建。
calico_backend:Calico的后端,默认为bird。
cni_network_config:符合CNI规范的网络配置。其中type=calico表示kubelet将从/opt/cni/bin目录下搜索名为“Calico”的可执行文件,并调用它完成容器网络的设置。ipam中type=calico-ipam表示kubelet将在/opt/cni/bin目录下搜索名为"calico-ipam"的可执行文件,用于完成容器IP地址的分配。
etcd如果配置了TLS安全认证,则还需要指定相应的ca、cert、key等文件。
(2)访问etcd所需的secret,对于无TLS的etcd服务,将data设置为空即可
# The following contains k8s Secrets for use with a TLS enabled etcd cluster.
# For information on populating Secrets, see http://kubernetes.io/docs/user-guide/secrets/
apiVersion: v1
kind: Secret
type: Opaque
metadata:
name: calico-etcd-secrets
namespace: kube-system
data:
# Populate the following files with etcd TLS configuration if desired, but leave blank if
# not using TLS for etcd.
# This self-hosted install expects three files with the following names. The values
# should be base64 encoded strings of the entire contents of each file.
# etcd-key: null
# etcd-cert: null
# etcd-ca: null
(3)calico-node,以Daemonset形式在每台Node上运行一个calico-node服务和一个install-cni服务
# This manifest installs the calico/node container, as well
# as the Calico CNI plugins and network config on
# each master and worker node in a Kubernetes cluster.
kind: DaemonSet
apiVersion: extensions/v1beta1
metadata:
name: calico-node
namespace: kube-system
labels:
k8s-app: calico-node
spec:
selector:
matchLabels:
k8s-app: calico-node
template:
metadata:
labels:
k8s-app: calico-node
annotations:
scheduler.alpha.kubernetes.io/critical-pod: ‘‘
scheduler.alpha.kubernetes.io/tolerations: |
[{"key": "dedicated", "value": "master", "effect": "NoSchedule" },
{"key":"CriticalAddonsOnly", "operator":"Exists"}]
spec:
hostNetwork: true
containers:
# Runs calico/node container on each Kubernetes node. This
# container programs network policy and routes on each
# host.
- name: calico-node
image: quay.io/calico/node:v1.1.3
env:
# The location of the Calico etcd cluster.
- name: ETCD_ENDPOINTS
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_endpoints
# Choose the backend to use.
- name: CALICO_NETWORKING_BACKEND
valueFrom:
configMapKeyRef:
name: calico-config
key: calico_backend
# Disable file logging so `kubectl logs` works.
- name: CALICO_DISABLE_FILE_LOGGING
value: "true"
# Set Felix endpoint to host default action to ACCEPT.
- name: FELIX_DEFAULTENDPOINTTOHOSTACTION
value: "ACCEPT"
# Configure the IP Pool from which Pod IPs will be chosen.
- name: CALICO_IPV4POOL_CIDR
value: "192.168.0.0/16"
- name: CALICO_IPV4POOL_IPIP
value: "always"
# Disable IPv6 on Kubernetes.
- name: FELIX_IPV6SUPPORT
value: "false"
# Set Felix logging to "info"
- name: FELIX_LOGSEVERITYSCREEN
value: "info"
# Location of the CA certificate for etcd.
- name: ETCD_CA_CERT_FILE
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_ca
# Location of the client key for etcd.
- name: ETCD_KEY_FILE
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_key
# Location of the client certificate for etcd.
- name: ETCD_CERT_FILE
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_cert
# Auto-detect the BGP IP address.
- name: IP
value: ""
securityContext:
privileged: true
resources:
requests:
cpu: 250m
volumeMounts:
- mountPath: /lib/modules
name: lib-modules
readOnly: true
- mountPath: /var/run/calico
name: var-run-calico
readOnly: false
- mountPath: /calico-secrets
name: etcd-certs
# This container installs the Calico CNI binaries
# and CNI network config file on each node.
- name: install-cni
image: quay.io/calico/cni:v1.8.0
command: ["/install-cni.sh"]
env:
# The location of the Calico etcd cluster.
- name: ETCD_ENDPOINTS
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_endpoints
# The CNI network config to install on each node.
- name: CNI_NETWORK_CONFIG
valueFrom:
configMapKeyRef:
name: calico-config
key: cni_network_config
volumeMounts:
- mountPath: /host/opt/cni/bin
name: cni-bin-dir
- mountPath: /host/etc/cni/net.d
name: cni-net-dir
- mountPath: /calico-secrets
name: etcd-certs
volumes:
# Used by calico/node.
- name: lib-modules
hostPath:
path: /lib/modules
- name: var-run-calico
hostPath:
path: /var/run/calico
# Used to install CNI.
- name: cni-bin-dir
hostPath:
path: /opt/cni/bin
- name: cni-net-dir
hostPath:
path: /etc/cni/net.d
# Mount in the etcd TLS secrets.
- name: etcd-certs
secret:
secretName: calico-etcd-secrets
该Pod中包括如下两个容器:
calico-node:Calico服务程序,用于设置Pod的网络资源,保证Pod的网络与各Node互联互通,它还需要以hostNetwork模式运行,直接使用宿主机网络。
install-cni:在各Node上安装CNI二进制文件到/opt/cni/bin目录下,并安装相应的网络配置文件到/etc/cni/net.d目录下。
calico-node服务的主要参数如下:
CALICO_IPV4POOL_CIDR:Calico IPAM的IP地址池,Pod的IP地址将从该池中进行分配。
CALICO_IPV4POOL_IPIP:是否启用IPIP模式。启用IPIP模式时,Calico将在Node上创建一个名为"tunl0"的虚拟隧道。
FELIX_IPV6SUPPORT:是否启用IPV6。
FELIX_LOGSEVERITYSCREEN:日志级别。
IP Pool可以使用两种模式:BGP或IPIP模式。
使用IPIP模式时,设置CALICO_IPV4POOL_IPIP=“always”,不使用IPIP模式时,设置CALICO_IPV4POOL_IPIP="off",此时将使用BGP模式。
IPIP是一种将各Node的路由之间做一个tunnel,再把两个网络连接起来的模式。启用IPIP模式时,Calico将在各Node上创建一个名为"tunl0"的虚拟网络接口。如下图所示。
BGP模式则直接使用物理机作为虚拟路由路(vRouter),不再创建额外的tunnel。
(4)calico-policy-controller容器
用于对接k8s集群中为Pod设置的Network Policy。
# This manifest deploys the Calico policy controller on Kubernetes.
# See https://github.com/projectcalico/k8s-policy
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
name: calico-policy-controller
namespace: kube-system
labels:
k8s-app: calico-policy
annotations:
scheduler.alpha.kubernetes.io/critical-pod: ‘‘
scheduler.alpha.kubernetes.io/tolerations: |
[{"key": "dedicated", "value": "master", "effect": "NoSchedule" },
{"key":"CriticalAddonsOnly", "operator":"Exists"}]
spec:
# The policy controller can only have a single active instance.
replicas: 1
strategy:
type: Recreate
template:
metadata:
name: calico-policy-controller
namespace: kube-system
labels:
k8s-app: calico-policy
spec:
# The policy controller must run in the host network namespace so that
# it isn‘t governed by policy that would prevent it from working.
hostNetwork: true
containers:
- name: calico-policy-controller
image: quay.io/calico/kube-policy-controller:v0.5.4
env:
# The location of the Calico etcd cluster.
- name: ETCD_ENDPOINTS
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_endpoints
# Location of the CA certificate for etcd.
- name: ETCD_CA_CERT_FILE
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_ca
# Location of the client key for etcd.
- name: ETCD_KEY_FILE
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_key
# Location of the client certificate for etcd.
- name: ETCD_CERT_FILE
valueFrom:
configMapKeyRef:
name: calico-config
key: etcd_cert
# The location of the Kubernetes API. Use the default Kubernetes
# service for API access.
- name: K8S_API
value: "https://kubernetes.default:443"
# Since we‘re running in the host namespace and might not have KubeDNS
# access, configure the container‘s /etc/hosts to resolve
# kubernetes.default to the correct service clusterIP.
- name: CONFIGURE_ETC_HOSTS
value: "true"
volumeMounts:
# Mount in the etcd TLS secrets.
- mountPath: /calico-secrets
name: etcd-certs
volumes:
# Mount in the etcd TLS secrets.
- name: etcd-certs
secret:
secretName: calico-etcd-secrets
用户在k8s集群中设置了Pod的Network Policy之后,calico-policy-controller就会自动通知各个Node上的calico-node服务,在宿主机上设置相应的iptables规则,完成Pod间网络访问策略的设置。
做好以上配置文件的准备工作后,就可以开始创建Calico的各资源对象了。
[[email protected] ~]# kubectl create -f calico.yaml
configmap "calico-config" created
secret "calico-etcd-secrets" created
daemonset "calico-node" created
deployment "calico-policy-controller" created
[[email protected] ~]#
确保各服务正确运行:
[[email protected] ~]# kubectl get pods --namespace=kube-system -o wide
NAME READY STATUS RESTARTS AGE IP NODE
calico-node-59n9j 2/2 Running 1 9h 10.0.2.5 10.0.2.5
calico-node-cksq5 2/2 Running 1 9h 10.0.2.4 10.0.2.4
calico-policy-controller-54dbfcd7c7-ctxzz 1/1 Running 0 9h 10.0.2.5 10.0.2.5
[[email protected] ~]#
[[email protected] ~]# kubectl get rs --namespace=kube-system
NAME DESIRED CURRENT READY AGE
calico-policy-controller-54dbfcd7c7 1 1 1 9h
[[email protected] ~]#
[[email protected] ~]# kubectl get deployment --namespace=kube-system
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
calico-policy-controller 1 1 1 1 9h
[[email protected] ~]# kubectl get secret --namespace=kube-system
NAME TYPE DATA AGE
calico-etcd-secrets Opaque 0 9h
[[email protected] ~]# kubectl get configmap --namespace=kube-system
NAME DATA AGE
calico-config 6 9h
[[email protected] ~]#
我们看下Node1上:
[[email protected] ~]# docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
dd431155ed2d quay.io/calico/cni "/install-cni.sh" 8 hours ago Up 8 hours k8s_install-cni_calico-node-cksq5_kube-system_e3ed0d80-6fe9-11e8-8a4a-080027800835_0
e7f20b684fc2 quay.io/calico/node "start_runit" 8 hours ago Up 8 hours k8s_calico-node_calico-node-cksq5_kube-system_e3ed0d80-6fe9-11e8-8a4a-080027800835_1
1c9010e4b661 gcr.io/google_containers/pause-amd64:3.0 "/pause" 8 hours ago Up 8 hours k8s_POD_calico-node-cksq5_kube-system_e3ed0d80-6fe9-11e8-8a4a-080027800835_1
[[email protected] ~]#
[[email protected] ~]# docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
cloudnil/pause-amd64 3.0 66c684b679d2 11 months ago 747kB
gcr.io/google_containers/pause-amd64 3.0 66c684b679d2 11 months ago 747kB
quay.io/calico/cni v1.8.0 8de7b24bd7ec 13 months ago 67MB
quay.io/calico/node v1.1.3 573ddcad1ff5 13 months ago 217MB
kubeguide/guestbook-php-frontend latest 47ee16830e89 23 months ago 510MB
Node2上多出一个Pod:calico-policy-controller
[[email protected] ~]# docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
ff4dbcd77892 quay.io/calico/kube-policy-controller "/dist/controller" 8 hours ago Up 8 hours k8s_calico-policy-controller_calico-policy-controller-54dbfcd7c7-ctxzz_kube-system_e3f067be-6fe9-11e8-8a4a-080027800835_0
60439cfbde00 quay.io/calico/cni "/install-cni.sh" 8 hours ago Up 8 hours k8s_install-cni_calico-node-59n9j_kube-system_e3efa53c-6fe9-11e8-8a4a-080027800835_1
c55f279ef3c1 quay.io/calico/node "start_runit" 8 hours ago Up 8 hours k8s_calico-node_calico-node-59n9j_kube-system_e3efa53c-6fe9-11e8-8a4a-080027800835_0
17d08ed5fd86 gcr.io/google_containers/pause-amd64:3.0 "/pause" 8 hours ago Up 8 hours k8s_POD_calico-node-59n9j_kube-system_e3efa53c-6fe9-11e8-8a4a-080027800835_1
aa85ee06190f gcr.io/google_containers/pause-amd64:3.0 "/pause" 8 hours ago Up 8 hours k8s_POD_calico-policy-controller-54dbfcd7c7-ctxzz_kube-system_e3f067be-6fe9-11e8-8a4a-080027800835_0
[[email protected] ~]#
[[email protected] ~]#
[[email protected] ~]# docker images
REPOSITORY TAG IMAGE ID CREATED SIZE
cloudnil/pause-amd64 3.0 66c684b679d2 11 months ago 747kB
gcr.io/google_containers/pause-amd64 3.0 66c684b679d2 11 months ago 747kB
quay.io/calico/cni v1.8.0 8de7b24bd7ec 13 months ago 67MB
quay.io/calico/node v1.1.3 573ddcad1ff5 13 months ago 217MB
quay.io/calico/kube-policy-controller v0.5.4 ac66b6e8f19e 14 months ago 22.6MB
kubeguide/guestbook-php-frontend latest 47ee16830e89 23 months ago 510MB
georce/router latest f3074d9a8369 3 years ago 190MB
[[email protected] ~]#
calico-node在正常运行之后,会根据CNI规范,在/etc/cni/net.d/目录下生成如下文件和目录,并在/opt/cni/bin目录下安装二进制文件calico和calico-ipam,供kubelet调用。
10-calico.conf:符合CNI规范的网络配置,其中type=calico表示该插件的二进制文件名为calico。
calico-kubeconfig:Calico所需的kubeconfig文件。
calico-tls目录:以TLS方式连接etcd的相关文件。
[[email protected] ~]# cd /etc/cni/net.d/
[[email protected] net.d]# ls
10-calico.conf calico-kubeconfig calico-tls
[[email protected] net.d]#
[[email protected] net.d]# ls /opt/cni/bin
calico calico-ipam flannel host-local loopback
[[email protected] net.d]#
查看k8s node1服务器的网络接口设置,可以看到一个新的名为"tunl0"的接口,并设置了网络地址为192.168.196.128
[[email protected] net.d]# ifconfig tunl0
tunl0: flags=193<UP,RUNNING,NOARP> mtu 1440
inet 192.168.196.128 netmask 255.255.255.255
tunnel txqueuelen 1000 (IPIP Tunnel)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 0 bytes 0 (0.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
查看k8s node2服务器的网络接口设置,可以看到一个新的名为"tunl0"的接口,并设置了网络地址为192.168.19.192
[[email protected] ~]# ifconfig tunl0
tunl0: flags=193<UP,RUNNING,NOARP> mtu 1440
inet 192.168.19.192 netmask 255.255.255.255
tunnel txqueuelen 1000 (IPIP Tunnel)
RX packets 0 bytes 0 (0.0 B)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 0 bytes 0 (0.0 B)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
这两个子网都是从calico-node的IP地址池192.168.0.0/16中进行分配的。同时,docker0对于k8s设置Pod的IP地址将不再起作用。
查看两台主机的路由表。可以看到node1服务器上有一条到node2的私网192.168.19.192的路由转发规则:
[[email protected] net.d]# ip route
default via 10.0.2.1 dev enp0s3 proto dhcp metric 100
10.0.2.0/24 dev enp0s3 proto kernel scope link src 10.0.2.4 metric 100
172.16.10.0/24 dev docker0 proto kernel scope link src 172.16.10.1
192.168.19.192/26 via 10.0.2.5 dev tunl0 proto bird onlink
blackhole 192.168.196.128/26 proto bird
[[email protected] net.d]#
然后查看node2服务器的路由表,也可以看到有一条到node1私网192.168.196.128的路由转发规则:
[[email protected] ~]# ip route
default via 10.0.2.1 dev enp0s3 proto dhcp metric 100
10.0.2.0/24 dev enp0s3 proto kernel scope link src 10.0.2.5 metric 100
172.16.20.0/24 dev docker0 proto kernel scope link src 172.16.20.1
blackhole 192.168.19.192/26 proto bird
192.168.196.128/26 via 10.0.2.4 dev tunl0 proto bird onlink
这样通过Calico就完成了Node间容器网络的设置。在后续的Pod创建过程中,kubelet将通过CNI接口调用Calico进行Pod网络的设置,包括IP地址、路由规则、iptables规则等。
如果设置CALICO_IPV4POOL_IPIP="off",即不使用IPIP模式,则Calico将不会创建tunl0网络接口,路由规则直接使用物理机网卡作为路由器进行转发。
4.3 使用网络策略实现Pod间的访问策略
Calico支持设置Pod间的访问策略,基本原理如下图所示。
下面以一个提供服务的Nginx Pod为例,为两个客户端Pod设置不同的网络访问权限,允许包含Label "role=nginxclient"的Pod访问Nginx容器,无此Label的其他容器则拒绝访问。
步骤1:
首先为需要设置网络隔离的Namespace进行标注,本例中的所有Pod都在Namespace default中,故对其进行默认网络隔离的设置:
# kubectl annotate ns default
"net.beta.kubernetes.io/network-policy={\"ingress\": {\"isolation\": \"DefaultDeny\"}}"
设置完成后,default内的各Pod之间的网络就无法连通了。
步骤2:创建Nginx Pod,并添加Label "app=nginx"
apiVersion: v1
kind: Pod
metadata:
name: nginx
labels:
app: nginx
spec:
containers:
name: nginx
image: nginx
步骤3:为Nginx设置准入访问 策略
networkpolicy-allow-nginxclient.yaml
kind: NetworkPolicy
apiVersion: extension/v1beta1
metadata:
name: allow-nginxclient
spec:
podSelector:
matchLabels:
app: nginx
ingress:
- from:
- podSelector:
matchLabels:
role: nginxclient
ports:
- protocol: TCP
port: 80
目标Pod应包含Label "app=nginx",允许访问的客户端Pod包含Label "role=nginxclient",并允许客户端访问mysql容器的80端口。
创建该NetworkPolicy资源对象:
# kubectl create -f networkpolicy-allow-nginxclient.yaml
步骤4:创建两个客户端Pod,一个包含Label "role=nginxclient",另一个无此Label。分别进入各Pod,访问Nginx容器,验证网络策略的效果。
client1.yaml
apiVersion: v1
kind: Pod
metadata:
name: client1
labels:
role: nginxclient
spec:
containers:
- name: client1
image: busybox
command: [ "sleep", "3600" ]
client2.yaml
apiVersion: v1
kind: Pod
metadata:
name: client2
spec:
containers:
- name: client2
image: busybox
command: [ "sleep", "3600" ]
创建以上两个Pods,并进入每个容器中进行服务访问的验证。
上面例子中的网络策略是由calico-policy-controller具体实现的,calico-poliey-controller持续监听k8s中NetworkPolicy的定义,与各Pod通过Label进行关联,将允许访问或拒绝访问的策略通知到各calico-node服务。
最终calico-node完成对Pod间网络访问的设置,实现应用的网络隔离。
参考资料:
https://blog.csdn.net/watermelonbig/article/details/80720378
http://cizixs.com/2017/10/19/docker-calico-network
原文地址:https://www.cnblogs.com/heboxiang/p/12183173.html