1
00:00:04,820 --> 00:00:11,510
‫This is going to be the first lecture where we're actually going to use multiple hosts or nodes,

2
00:00:11,570 --> 00:00:17,900
‫or instances, or whatever you want...multiple OS's. We're going to set up a 3-node swarm across

3
00:00:17,900 --> 00:00:19,630
‫all three of those nodes.

4
00:00:19,670 --> 00:00:24,620
‫We unfortunately can't do this part with the built-in Docker on your machine because that's only going to

5
00:00:24,620 --> 00:00:31,150
‫provide us one OS. We need three Linux OS's to play with this next set of features.

6
00:00:31,190 --> 00:00:35,600
‫The first option you're going to have is using play-with-docker.com.

7
00:00:35,630 --> 00:00:37,900
‫It's absolutely the easiest way to get started.

8
00:00:37,910 --> 00:00:43,400
‫It comes with Docker pre-installed. It takes you seconds to provision three different machines, and it requires

9
00:00:43,430 --> 00:00:50,160
‫no investment on your end. But it does mean that every four hours it will wipe all the work you've done.

10
00:00:50,180 --> 00:00:54,590
‫So if you think you're going to do the rest of this section over the course of days, then this may not

11
00:00:54,590 --> 00:00:59,210
‫be the best option for you. But if you're going to focus for a few hours, you can definitely get through

12
00:00:59,210 --> 00:01:05,030
‫most, if not all, of the multi node swarm stuff using Play With Docker in one sitting.

13
00:01:05,030 --> 00:01:09,380
‫Another option you have is you can use docker-machine, which is not something we've talked about yet,

14
00:01:09,530 --> 00:01:14,570
‫but it's a command line tool that actually gets installed with Docker for Windows and Docker for Mac.

15
00:01:14,720 --> 00:01:22,040
‫You can download it for Linux, and it is basically an automation tool for provisioning virtual machines

16
00:01:22,040 --> 00:01:26,960
‫locally, and on the Internet, and then automatically setting up Docker for you.

17
00:01:26,960 --> 00:01:31,850
‫The default way that it works is with VirtualBox, which you can install free. But it does mean that if

18
00:01:31,850 --> 00:01:37,340
‫you're going to run three virtual machines locally, each one of them probably needs a gig of RAM or so

19
00:01:37,340 --> 00:01:42,440
‫to run through a lot of these examples. So you'll need a machine with a good amount of RAM in it. Another

20
00:01:42,530 --> 00:01:48,350
‫option, and the one that I'm going to default to for a lot of these demos and walk throughs, is DigitalOcean.

21
00:01:48,350 --> 00:01:52,910
‫I just recommend DigitalOcean just because they have the cheapest and easiest service to

22
00:01:52,910 --> 00:01:58,760
‫get started, and they run everything on SSD so it's nice and fast. You can actually use this in a

23
00:01:58,760 --> 00:02:04,130
‫very, very similar set up to what production would be because you're using three actual virtual machines,

24
00:02:04,190 --> 00:02:07,810
‫in the cloud, and it takes about a minute to launch each virtual machine.

25
00:02:07,910 --> 00:02:12,050
‫If you're thinking that that might be an option, you can use my referral code and the references to

26
00:02:12,050 --> 00:02:15,630
‫this section to get your first $10 free.

27
00:02:15,650 --> 00:02:21,740
‫If you're only going to take a few days to get through this swarm stuff, you can provision three, $10 a

28
00:02:21,740 --> 00:02:25,480
‫month nodes and have them up for days or a week.

29
00:02:25,520 --> 00:02:30,280
‫And that might fit within the $10 budget so technically wouldn't cost you anything.

30
00:02:30,290 --> 00:02:35,270
‫And then lastly, there are so many ways to get Docker installed, and that's really all you need is three

31
00:02:35,270 --> 00:02:39,900
‫machines with Docker installed on them. You can use Docker Machine.

32
00:02:40,010 --> 00:02:47,120
‫It has built in drivers that allow you to use it to provision Amazon instances, or Azure instances, or

33
00:02:47,120 --> 00:02:54,170
‫DigitalOcean Droplets, or Google Compute nodes or any place you can get a Linux virtual machine, you

34
00:02:54,170 --> 00:02:58,460
‫can actually just install Docker with an automated script from get.docker.com.

35
00:02:58,520 --> 00:03:04,640
‫Frankly, I prefer the get.docker.com version only because Docker Machine may save you a few steps,

36
00:03:04,640 --> 00:03:09,230
‫especially for locations like Amazon and Azure that require you to provision a lot of different things

37
00:03:09,230 --> 00:03:13,760
‫before you can get to just provisioning a virtual machine. But you're probably not going to be using

38
00:03:13,760 --> 00:03:15,580
‫Docker Machine in production.

39
00:03:15,590 --> 00:03:20,960
‫It is a tool to simply automate dev and test environments. It was never really designed to set up all

40
00:03:20,960 --> 00:03:24,500
‫of the production settings you might need for a multi-node swarm.

41
00:03:24,530 --> 00:03:30,170
‫There's nothing inherently wrong with it, but typically get.docker.com works on all the major

42
00:03:30,170 --> 00:03:35,140
‫Linux distributions and doesn't take any longer than actually using Docker Machine itself.

43
00:03:35,210 --> 00:03:41,390
‫I'm going to give you a really quick example of just getting any of these scenarios to the point

44
00:03:41,390 --> 00:03:46,540
‫where Docker is installed. Then they all will be very similar in the way we actually create the

45
00:03:46,540 --> 00:03:48,850
‫swarm and do our services.

46
00:03:48,860 --> 00:03:52,250
‫So don't think that you have to do it the way I do it.

47
00:03:52,250 --> 00:03:56,600
‫Really we're just trying to get Docker info to actually bring back information, and then just to make

48
00:03:56,600 --> 00:04:01,400
‫sure that the three nodes have networking access to each other and they're on a well connected network.

49
00:04:01,400 --> 00:04:04,630
‫So ideally they should be in the same network.

50
00:04:04,670 --> 00:04:09,560
‫If you were going to use Play With Docker, you actually just go to play-with-docker.com and check

51
00:04:09,560 --> 00:04:11,430
‫on the box for I'm not a robot.

52
00:04:11,540 --> 00:04:15,580
‫And if it gives you the CAPTCHA screen, just follow along.

53
00:04:17,720 --> 00:04:20,290
‫All right. You'll see that it's counting down by four hours.

54
00:04:20,310 --> 00:04:26,130
‫But if I just click the creating instance, and then I click it two more times, it'll create three virtual

55
00:04:26,130 --> 00:04:27,480
‫environments very quickly.

56
00:04:27,570 --> 00:04:31,680
‫As a side note, this is actually using Docker in Docker, which is a pretty neat technology where we're

57
00:04:31,680 --> 00:04:34,480
‫actually running Docker inside a Docker container.

58
00:04:34,500 --> 00:04:39,870
‫But in this set up with Play With Docker, which was created by other Docker captains and the Docker team,

59
00:04:40,020 --> 00:04:41,450
‫it's really hard to tell a difference.

60
00:04:41,580 --> 00:04:51,240
‫If I go to node1, I can already see that Docker info gives me the latest version. And if I even tried

61
00:04:51,240 --> 00:04:58,200
‫to ping node2, it actually recognizes friendly names, so it will make it easier to communicate back and

62
00:04:58,200 --> 00:04:59,220
‫forth between them.

63
00:04:59,260 --> 00:05:03,990
‫So if this is something you want to try, it's a great and easy way on either low-powered machines

64
00:05:04,020 --> 00:05:07,290
‫or if you just don't want to take the time to set up an environment for the next four hours.

65
00:05:08,180 --> 00:05:15,290
‫Option two is to use docker-machine. If you have it installed, it should be at least version 10

66
00:05:15,380 --> 00:05:16,610
‫if not newer.

67
00:05:16,670 --> 00:05:20,450
‫You'll know if it's installed because you'll type the docker-machine command from your command line and

68
00:05:20,450 --> 00:05:22,340
‫it'll either work or it won't.

69
00:05:22,430 --> 00:05:28,370
‫If you need to download it for your particular Linux OS, then you can hop over to the Docker Documentation,

70
00:05:28,370 --> 00:05:31,580
‫which will actually give you a quick curl line to download it.

71
00:05:31,730 --> 00:05:37,840
‫If you're using the ToolBox on Windows 7 and 8, or Docker for Windows on Windows 10, or Docker for

72
00:05:37,840 --> 00:05:42,040
‫Mac on Mac OS, it's already going to be up to date on those OS's for you.

73
00:05:42,260 --> 00:05:48,260
‫If you want to use the VirtualBox option, which is the default, you just need to jump over to

74
00:05:48,260 --> 00:05:54,170
‫virtualbox.org and download the latest version for your OS. And you don't even have to do anything with

75
00:05:54,170 --> 00:05:54,500
‫VirtualBox.

76
00:05:54,500 --> 00:06:02,150
‫It just needs to be installed. And then from your command line, you can just type docker-machine create

77
00:06:02,510 --> 00:06:09,530
‫node1, and this will actually run through creating a virtual machine with BusyBox, which is a very

78
00:06:09,530 --> 00:06:12,430
‫lightweight Linux distribution.

79
00:06:12,640 --> 00:06:18,160
‫This will actually run through creating a very lightweight Linux machine, and you'll just repeat those

80
00:06:18,160 --> 00:06:20,920
‫two commands for node2 and 3.

81
00:06:21,070 --> 00:06:26,590
‫If you need more information on a Docker Machine, you can actually look at my reference section

82
00:06:26,620 --> 00:06:30,130
‫for tutorial links.

83
00:06:30,390 --> 00:06:34,400
‫Once your machines are created, you really have two options for how you can access those machines.

84
00:06:34,410 --> 00:06:41,580
‫You can either use docker-machine ssh, and then the name of the node you created and that'll hop you right

85
00:06:41,660 --> 00:06:42,630
‫into that machine.

86
00:06:42,960 --> 00:06:45,870
‫Or you can use the environment command.

87
00:06:45,960 --> 00:06:54,030
‫So the command docker-machine env, and then the name of the machine you want, will give you some output,

88
00:06:54,180 --> 00:06:59,760
‫and you can either cut and paste these options to your command line or you can copy and paste the bottom

89
00:06:59,760 --> 00:07:05,280
‫option, and that will actually create the environment variable so that when I run my docker command,

90
00:07:05,640 --> 00:07:10,830
‫because again, the Docker CLI just needs to know what Docker host to talk to. And by default, it's talking

91
00:07:10,830 --> 00:07:11,850
‫to your local machine.

92
00:07:11,880 --> 00:07:14,530
‫But that just reprogrammed it to talk to the Docker Machine

93
00:07:14,540 --> 00:07:15,560
‫node1.

94
00:07:15,630 --> 00:07:20,550
‫If I do docker info, you'll see that its name is node1 because that's the one I'm actually now operating

95
00:07:20,550 --> 00:07:20,920
‫on.

96
00:07:22,460 --> 00:07:27,010
‫For the last example, I'm going to use DigitalOcean, which I have created an account on.

97
00:07:27,080 --> 00:07:33,830
‫Here, they call servers Droplets. I recommend if you don't have a specific reason to run another

98
00:07:33,830 --> 00:07:41,900
‫OS, to start with Ubuntu 16.04 because that tends to be the first Linux distribution to support the

99
00:07:41,900 --> 00:07:43,430
‫newer Docker features.

100
00:07:43,430 --> 00:07:44,870
‫So that's the default there.

101
00:07:44,870 --> 00:07:50,990
‫Then I would recommend for the remainder of this course for us to use the $10 a month server. You

102
00:07:50,990 --> 00:07:55,850
‫might be able to get away with the $5 a month server, but it might get a little bit slow during some of

103
00:07:55,850 --> 00:08:01,860
‫the later demos when we do monitoring and logging containers on top of our applications.

104
00:08:01,910 --> 00:08:06,860
‫Then you can pick the data center closest to you. Then the only other option is to add an SSH key.

105
00:08:07,430 --> 00:08:12,690
‫If you're not familiar with how to create your own SSH key locally and then upload the public into

106
00:08:12,710 --> 00:08:13,830
‫DigitalOcean,

107
00:08:13,830 --> 00:08:17,930
‫there's a great tutorial that they have for it that you can find in the resources of this section.

108
00:08:19,140 --> 00:08:27,320
‫Down at the bottom I'm going to create three machines, and I'm going to call them node1, 2 and 3.

109
00:08:27,520 --> 00:08:34,570
‫Then I need to wait probably about 60 seconds and they'll be ready for me.

110
00:08:34,620 --> 00:08:38,040
‫You could just use the IP addresses here...jump on these machines.

111
00:08:38,090 --> 00:08:43,440
‫Clicking the copy button and then ssh root @ the IP address.

112
00:08:43,560 --> 00:08:49,230
‫And if your SSH key is properly set up on your machine, then it'll automatically log in with that

113
00:08:49,230 --> 00:08:49,870
‫key.

114
00:08:49,950 --> 00:08:56,070
‫But in the resources section, I've given you an example of how to set up a .ssh config file that will

115
00:08:56,070 --> 00:09:00,800
‫allow you to have these three nodes in a config so you can just simply call them node1, node2,

116
00:09:00,830 --> 00:09:04,980
‫node3, instead of having to remember their IP addresses, in case you're going to be doing this over multiple

117
00:09:04,980 --> 00:09:10,650
‫days. That's really an easy way to get in and out of these servers.

118
00:09:10,660 --> 00:09:15,550
‫If you've used Docker Machine, or if you used Play With Docker, then Docker is already installed

119
00:09:15,850 --> 00:09:17,760
‫and you're getting ready to create your swarm.

120
00:09:17,830 --> 00:09:22,400
‫But if you use DigitalOcean, or you roll your own VMs, then you need to get Docker installed.

121
00:09:22,540 --> 00:09:28,180
‫The easiest way to do that, that I like to use, is to use the script over at get.docker.com.

122
00:09:28,660 --> 00:09:33,520
‫You can see examples up here of how to use this script to get the current production release of

123
00:09:33,520 --> 00:09:34,850
‫Docker on your machine.

124
00:09:39,470 --> 00:09:45,020
‫Now, all three of my servers have the latest version of Docker installed and we're ready to set

125
00:09:45,020 --> 00:09:46,400
‫up our swarm.

126
00:09:46,400 --> 00:09:50,960
‫Again, all that's required is that these three nodes have solid networking to each other and specific

127
00:09:50,960 --> 00:09:51,760
‫ports open.

128
00:09:51,830 --> 00:09:56,360
‫If you're curious about the ports of swarm, look in the resources of this section where I actually

129
00:09:56,360 --> 00:10:01,220
‫have a link to my own list of the firewall ports necessary for Swarm to work.

130
00:10:01,220 --> 00:10:08,630
‫If you remember before, we did docker swarm init. Now, in this case, which is pretty common in cloud

131
00:10:08,630 --> 00:10:14,510
‫servers, it wants us to specify an IP address to advertise the Swarm service on.

132
00:10:14,510 --> 00:10:19,240
‫You want to use an IP address that is accessible from the other servers.

133
00:10:19,250 --> 00:10:23,570
‫In this case, I'm going to use the public IP address for my DigitalOcean

134
00:10:23,580 --> 00:10:24,050
‫host.

135
00:10:29,750 --> 00:10:30,650
‫And there we go.

136
00:10:30,700 --> 00:10:33,100
‫So far, this is just like on our local machine.

137
00:10:33,170 --> 00:10:39,530
‫This time, though, I'm going to copy the swarm join command and go over to node2 and add it in.

138
00:10:39,890 --> 00:10:40,720
‫And there we go.

139
00:10:41,000 --> 00:10:42,940
‫This node is now a part of the swarm.

140
00:10:43,010 --> 00:10:50,300
‫So we go back to node1, docker node ls, and now we have two.

141
00:10:50,510 --> 00:10:58,070
‫You'll notice the second node is only a worker. Because as it says, if we wanted it to be a manager,

142
00:10:58,070 --> 00:11:05,090
‫we should have gotten this manager token. But no worries. We can actually update this worker to promote

143
00:11:05,090 --> 00:11:06,160
‫it to being a manager.

144
00:11:06,200 --> 00:11:15,500
‫But if we're over on node2, notice that I can't use Swarm commands. Because workers aren't really privileged.

145
00:11:15,860 --> 00:11:18,720
‫They don't have access to control the swarm.

146
00:11:18,830 --> 00:11:23,510
‫So they're not going to be able to use any of the Swarm commands that we're used to using on the managers.

147
00:11:23,510 --> 00:11:28,280
‫Back on node1 and the manager, I'm going to do a docker node update...

148
00:11:37,470 --> 00:11:42,120
‫What I was doing there was hitting Tab for tab completion, which is always a fantastic feature in

149
00:11:42,120 --> 00:11:47,520
‫Docker, especially with so many features. It was walking me through each step of how to use that update

150
00:11:47,520 --> 00:11:48,240
‫command.

151
00:11:48,240 --> 00:11:53,390
‫Now I've got this final update finished and it has changed node2.

152
00:11:53,430 --> 00:11:59,930
‫Now if I do my docker node ls, you'll see that now it's considered reachable.

153
00:12:00,030 --> 00:12:04,920
‫But the original node is still the leader. By the way, the little asterisk? That just means that's the

154
00:12:04,920 --> 00:12:06,770
‫node you're currently talking to.

155
00:12:07,020 --> 00:12:10,750
‫For node3, let's add it as a manager by default.

156
00:12:10,890 --> 00:12:18,130
‫We need to go back to our original command of docker swarm, and then we need to get the join-token

157
00:12:20,740 --> 00:12:22,400
‫manager.

158
00:12:22,450 --> 00:12:25,430
‫You can get these tokens at any time.

159
00:12:25,510 --> 00:12:27,500
‫You don't have to write them down or save them.

160
00:12:27,520 --> 00:12:33,100
‫They're part of the swarm configuration and stored encrypted on disk, so don't worry about that.

161
00:12:33,220 --> 00:12:38,620
‫You can also change these in case they possibly get exposed or you maybe get a server that might have

162
00:12:38,620 --> 00:12:43,900
‫had a vulnerability on it, and you want to make sure that no nodes can join the swarm from the old key,

163
00:12:43,900 --> 00:12:45,790
‫you can actually rotate these keys.

164
00:12:45,910 --> 00:12:50,270
‫I'm going to copy this, paste it into node3.

165
00:12:50,770 --> 00:12:53,820
‫And there we go. Back on node1.

166
00:12:53,830 --> 00:13:00,670
‫If I do a docker node ls again, you'll see I've got three nodes. And they all have a manager status indicating

167
00:13:00,670 --> 00:13:03,540
‫that they are managers.

168
00:13:03,560 --> 00:13:11,600
‫So now we have a 3-node, redundant swarm, with three managers. Let's just do the same thing

169
00:13:11,600 --> 00:13:16,540
‫we did earlier, docker service create alpine

170
00:13:19,150 --> 00:13:22,420
‫ping 8.8.8.8

171
00:13:22,690 --> 00:13:23,770
‫Only this time,

172
00:13:27,350 --> 00:13:29,040
‫let's give it three replicas.

173
00:13:31,690 --> 00:13:33,920
‫And then do a docker service

174
00:13:34,010 --> 00:13:36,820
‫ls, and we'll see

175
00:13:36,820 --> 00:13:39,350
‫we've got three of three already running.

176
00:13:39,760 --> 00:13:50,050
‫If I did a docker node ps, I can actually see my local node is actually running this task or container.

177
00:13:50,320 --> 00:13:59,740
‫I can actually specify the node, node2, and it'll give me just the containers running on node2.

178
00:13:59,980 --> 00:14:07,900
‫If I wanted to get a full list, I could do a docker service ps, and then sleepy brown is the name that

179
00:14:07,900 --> 00:14:13,350
‫it gave our service. And you'll see here..let me expand the screen a little bit.

180
00:14:13,510 --> 00:14:17,210
‫You can see here that it's listing all three of our tasks.

181
00:14:17,470 --> 00:14:22,450
‫Once you've got this swarm created, you normally don't need to be typing commands into all the different

182
00:14:22,450 --> 00:14:26,910
‫nodes. You can really operate the whole swarm, for most things, from node1.

183
00:14:26,950 --> 00:14:33,250
‫So we'll be doing a lot of our work from node1 in future lectures, but for now, pat yourself on the back

184
00:14:33,280 --> 00:14:36,420
‫because you have a fully operational swarm cluster.