Docker - 用Flannel跨主机

试了下比较流行的几种SDN，感觉flannel还是比较好用，这里简单记录一下。

用的是virtualbox，3个机器，分别为:

• genesis : inet 192.168.99.103/24 brd 192.168.99.255 scope global dynamic enp0s3
• exodus : inet 192.168.99.105/24 brd 192.168.99.255 scope global dynamic enp0s3
• leviticus : inet 192.168.99.106/24 brd 192.168.99.255 scope global dynamic enp0s3

虚机信息如下

[root@localhost yum.repos.d]# uname -mars
Linux localhost.localdomain 3.10.0-229.el7.x86_64 #1 SMP Fri Mar 6 11:36:42 UTC 2015 x86_64 x86_64 x86_64 GNU/Linux

[root@localhost yum.repos.d]# cat /etc/*-release
CentOS Linux release 7.3.1611 (Core)
NAME="CentOS Linux"
VERSION="7 (Core)"
ID="centos"
ID_LIKE="rhel fedora"
VERSION_ID="7"
PRETTY_NAME="CentOS Linux 7 (Core)"
ANSI_COLOR="0;31"
CPE_NAME="cpe:/o:centos:centos:7"
HOME_URL="https://www.centos.org/"
BUG_REPORT_URL="https://bugs.centos.org/"

CENTOS_MANTISBT_PROJECT="CentOS-7"
CENTOS_MANTISBT_PROJECT_VERSION="7"
REDHAT_SUPPORT_PRODUCT="centos"
REDHAT_SUPPORT_PRODUCT_VERSION="7"

CentOS Linux release 7.3.1611 (Core)
CentOS Linux release 7.3.1611 (Core)

[root@localhost yum.repos.d]# docker version
Client:
 Version:      1.12.5
 API version:  1.24
 Go version:   go1.6.4
 Git commit:   7392c3b
 Built:        Fri Dec 16 02:23:59 2016
 OS/Arch:      linux/amd64

随便选择两台机器run一下，在容器中ifconfig:

[root@localhost ~]# docker run -it busybox
/ # ifconfig
eth0      Link encap:Ethernet  HWaddr 02:42:AC:11:00:02
          inet addr:172.17.0.2  Bcast:0.0.0.0  Mask:255.255.0.0
          inet6 addr: fe80::42:acff:fe11:2/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:12 errors:0 dropped:0 overruns:0 frame:0
          TX packets:6 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:1016 (1016.0 B)  TX bytes:508 (508.0 B)

lo        Link encap:Local Loopback
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:65536  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:0 (0.0 B)  TX bytes:0 (0.0 B)
    

发现参数完全相同，单纯bridge模式下并没有跨主机互通，而host模式是并不建议使用的。

install

先yum install -y etcd flannel，如果没问题是再好不过了。

etcd 3.x支持--config-file参数，需要的话可以从源代码install(需要golang 1.6+)。

先从etcd开始，简单说就是"distributed key value store"。

etcd集群的3种方式:

• static
• etcd discovery
• DNS discovery

DNS discovery主要是用srv record，这里先不搞DNS服务，下面对static和etcd discovery两种方式简单说明一下。

static

参数可以在启动时置顶，或者写到配置文件中，默认配置文件为/etc/etcd/etcd.conf。

genesis的配置如下:

ETCD_NAME=genesis
ETCD_DATA_DIR="/var/lib/etcd/genesis"
ETCD_LISTEN_PEER_URLS="http://192.168.99.103:2380"
ETCD_LISTEN_CLIENT_URLS="http://192.168.99.103:2379,http://127.0.0.1:2379"
ETCD_INITIAL_ADVERTISE_PEER_URLS="http://192.168.99.103:2380"
ETCD_ADVERTISE_CLIENT_URLS="http://192.168.99.103:2379"
ETCD_INITIAL_CLUSTER_STATE="new"
ETCD_INITIAL_CLUSTER_TOKEN="etct-fantasy"
ETCD_INITIAL_CLUSTER="exodus=http://192.168.99.105:2380,genesis=http://192.168.99.103:2380"

exodus的配置如下:

ETCD_NAME=exodus
ETCD_DATA_DIR="/var/lib/etcd/exodus"
ETCD_LISTEN_PEER_URLS="http://192.168.99.105:2380"
ETCD_LISTEN_CLIENT_URLS="http://192.168.99.105:2379,http://127.0.0.1:2379"
ETCD_INITIAL_ADVERTISE_PEER_URLS="http://192.168.99.105:2380"
ETCD_ADVERTISE_CLIENT_URLS="http://192.168.99.105:2379"
ETCD_INITIAL_CLUSTER_STATE="new"
ETCD_INITIAL_CLUSTER_TOKEN="etct-fantasy"
ETCD_INITIAL_CLUSTER="exodus=http://192.168.99.105:2380,genesis=http://192.168.99.103:2380"

启动方式看自己喜好，如果打算用systemctl启动的话，注意/usr/lib/systemd/system/etcd.service中的内容可能不会如你所愿。

启动后，检查一下集群状态，顺便也看看有哪些成员:

[root@localhost etcd]# etcdctl cluster-health
member 7a4f27f78a05e755 is healthy: got healthy result from http://192.168.99.103:2379
failed to check the health of member 8e8718b335c6c9a2 on http://192.168.99.105:2379: Get http://192.168.99.105:2379/health: dial tcp 192.168.99.105:2379: i/o timeout
member 8e8718b335c6c9a2 is unreachable: [http://192.168.99.105:2379] are all unreachable
cluster is healthy

提示"member unreachable"，看来是被exodus的防火墙拦住了，我们先粗暴一点。

[root@localhost etcd]# systemctl stop firewalld
[root@localhost etcd]# etcdctl cluster-health
member 7a4f27f78a05e755 is healthy: got healthy result from http://192.168.99.103:2379
member 8e8718b335c6c9a2 is healthy: got healthy result from http://192.168.99.105:2379
cluster is healthy

etcd discovery

当然，这样配置的前提是已经知道各个节点的信息。

但实际场景中可能无法预知各个member，所以我们需要让etcd自己去发现(discovery)。

首先，etcd提供了一个public discovery service - discovery.etcd.io，我们用它来生成一个discovery token，并在genesis创建目录:

[root@localhost etcd]# curl https://discovery.etcd.io/new?size=3
https://discovery.etcd.io/6321c0706046c91f2b2598206ffa3272
[root@localhost etcd]# etcdctl set /discovery/6321c0706046c91f2b2598206ffa3272/_config/size 3

修改exodus的配置，用discovery代替之前的cluster:

ETCD_NAME=exodus
ETCD_DATA_DIR="/var/lib/etcd/exodus"
ETCD_LISTEN_PEER_URLS="http://192.168.99.105:2380"
ETCD_LISTEN_CLIENT_URLS="http://192.168.99.105:2379,http://127.0.0.1:2379"
ETCD_INITIAL_ADVERTISE_PEER_URLS="http://192.168.99.105:2380"
ETCD_ADVERTISE_CLIENT_URLS="http://192.168.99.105:2379"
ETCD_DISCOVERY=http://192.168.99.103:2379/v2/keys/discovery/98a976dac265a218f1a1959eb8dde57f

如果启动后一直显示如下错误(参考: raft election):

rafthttp: the clock difference against peer ?????? is too high [??????s > 1s]

简单的解决方法是通过ntp:

[root@localhost etcd]yum install ntp -y
[root@localhost etcd]# systemctl enable ntpd
Created symlink from /etc/systemd/system/multi-user.target.wants/ntpd.service to /usr/lib/systemd/system/ntpd.service.
[root@localhost etcd]# systemctl start ntpd

flannel

set一个路径给flannel用:

etcdctl set /coreos.com/network/config '{ "Network": "10.1.0.0/16" }'

以systemctl start flanneld方式启动时，如果出现以下错误

network.go:53] Failed to retrieve network config: 100: Key not found (/coreos.net) [9]  

注意/etc/sysconfig/flanneld中的内容，FLANNEL_ETCD_PREFIX很可能是/atomic.io/network，将其改为/coreos.com/network。
或者也可以通过-etcd-prefix指定。

启动成功后，查看subnet:

[root@localhost etcd]# etcdctl ls /coreos.com/network/subnets
/coreos.com/network/subnets/10.1.90.0-24
/coreos.com/network/subnets/10.1.30.0-24
/coreos.com/network/subnets/10.1.18.0-24

flannel启动成功后会生成/run/flannel/docker，内容如下:

DOCKER_OPT_BIP="--bip=10.1.30.1/24"
DOCKER_OPT_IPMASQ="--ip-masq=true"
DOCKER_OPT_MTU="--mtu=1450"
DOCKER_NETWORK_OPTIONS=" --bip=10.1.30.1/24 --ip-masq=true --mtu=1450 "

用以下方式启动docker:

[root@localhost etcd]# source /run/flannel/docker
[root@localhost etcd]# docker daemon ${DOCKER_NETWORK_OPTIONS} >> /dev/null 2>&1 &

／run/flannel/docker是怎么来的？

参考flanneld的两个启动参数，-subnet-dir和-subnet-file。

在genesis进入容器看看效果:

[root@localhost etcd]# docker run -it busybox
/ # ifconfig
eth0      Link encap:Ethernet  HWaddr 02:42:0A:01:5A:02
          inet addr:10.1.90.2  Bcast:0.0.0.0  Mask:255.255.255.0
          inet6 addr: fe80::42:aff:fe01:5a02/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1450  Metric:1
          RX packets:6 errors:0 dropped:0 overruns:0 frame:0
          TX packets:6 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:508 (508.0 B)  TX bytes:508 (508.0 B)

lo        Link encap:Local Loopback
          inet addr:127.0.0.1  Mask:255.0.0.0
          inet6 addr: ::1/128 Scope:Host
          UP LOOPBACK RUNNING  MTU:65536  Metric:1
          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1
          RX bytes:0 (0.0 B)  TX bytes:0 (0.0 B)

在exodus也做类似操作，在exodus中ping一下10.1.90.2，发现是通的。
并在各自容器中也ping一下，检查是否跨主机互通。