0
1
00:00:02,030 --> 00:00:02,430
All right.
1

2
00:00:02,440 --> 00:00:03,880
Last one and then we're done.
2

3
00:00:03,880 --> 00:00:05,250
All right Vlad.
3

4
00:00:05,470 --> 00:00:10,390
Thanks for waiting so we're on PHP FPM now another PSP user.
4

5
00:00:10,810 --> 00:00:20,510
We're doing PHP FPM Alpine so this is a small image to start and then they load FPM on top of it and
5

6
00:00:20,510 --> 00:00:22,610
then you're setting all of your versions up here.
6

7
00:00:22,610 --> 00:00:24,040
Great.
7

8
00:00:24,050 --> 00:00:28,700
Your directories setting up a bunch of different environment variables that you can override.
8

9
00:00:28,790 --> 00:00:32,890
And then as a whole this is exactly this is HP.
9

10
00:00:33,200 --> 00:00:35,290
This is this is true production HP.
10

11
00:00:35,300 --> 00:00:36,530
Good job on that.
11

12
00:00:36,530 --> 00:00:38,650
That's exactly how mine would look.
12

13
00:00:40,370 --> 00:00:48,140
Now yeah you're setting P2P dependencies and then you're doing a P.K. down here.
13

14
00:00:48,250 --> 00:00:54,190
Now I would probably do versions I would pin versions again down here and then
14

15
00:01:03,610 --> 00:01:04,980
lots and lots of dependencies.
15

16
00:01:04,990 --> 00:01:09,880
I would pin the ones that your app uses in production all of the little stuff like curl and get and
16

17
00:01:09,880 --> 00:01:10,570
stuff.
17

18
00:01:10,570 --> 00:01:16,700
You probably need a pen and you've got your pens down your PSP three so good for that good there.
18

19
00:01:16,870 --> 00:01:25,020
And then even more extensions we got people in there we got all the proper Docker PSP extension installer.
19

20
00:01:25,020 --> 00:01:29,120
So that's good then you're adding a composer user.
20

21
00:01:29,250 --> 00:01:30,340
Great.
21

22
00:01:30,570 --> 00:01:31,590
Right here.
22

23
00:01:31,590 --> 00:01:39,070
So it's important to remember for everyone else that oh look at that in Alpine we cannot pen versions
23

24
00:01:39,070 --> 00:01:39,930
of Alpine packages.
24

25
00:01:39,930 --> 00:01:41,780
I did not know that.
25

26
00:01:41,950 --> 00:01:48,340
That seems odd but I don't have enough information to to do that.
26

27
00:01:48,340 --> 00:01:49,950
So this is an interesting point.
27

28
00:01:50,020 --> 00:02:00,480
And if you watch my doctor over here I mentioned earlier and oh right here on the doctor talk so you
28

29
00:02:00,480 --> 00:02:09,960
can go over to Dockers Web site and watch my talk from Dockercon and I give a 10 minute argument on minimal
29

30
00:02:10,230 --> 00:02:11,850
images versus Alpine.
30

31
00:02:11,880 --> 00:02:15,810
So if you're an HP and you're using the HP
31

32
00:02:18,870 --> 00:02:32,670
images you might consider a Debian minimal image such as a slim uh.
32

33
00:02:32,770 --> 00:02:39,010
Now they just have Alpine so unfortunately they're not using a slim option but with node and some other
33

34
00:02:39,010 --> 00:02:44,500
ones there are slim options and sometimes I prefer those in fact most of them I prefer those over Alpine
34

35
00:02:44,800 --> 00:02:48,200
because the difference is like 50 meg overall total.
35

36
00:02:48,280 --> 00:02:50,610
So it's not I don't do it for size reasons.
36

37
00:02:50,620 --> 00:02:56,830
And then lately with alpine between Alpine security issues the fact that you can't see V scan it very
37

38
00:02:56,830 --> 00:02:57,360
well.
38

39
00:02:58,420 --> 00:03:01,030
And so it actually to me right now is a container.
39

40
00:03:01,030 --> 00:03:06,850
Alpine is less secure than the most current Debian release of a base image.
40

41
00:03:06,850 --> 00:03:07,630
That's my opinion.
41

42
00:03:07,810 --> 00:03:09,790
But I talk more about that and my Dockercon talk.
42

43
00:03:10,090 --> 00:03:15,520
So scroll up in the chat or I'll throw it back in here one more time for those of you that have just
43

44
00:03:15,520 --> 00:03:17,430
joined.
44

45
00:03:17,890 --> 00:03:22,930
But I'm I might consider that if I can't pin versions because I'm just so paranoid because I've had
45

46
00:03:22,930 --> 00:03:27,820
production outages due to version incompatibilities that I pin everything
46

47
00:03:31,370 --> 00:03:31,690
all right.
47

48
00:03:31,820 --> 00:03:34,730
So yeah I got composer and you're changing chose down here.
48

49
00:03:34,740 --> 00:03:35,270
Yep.
49

50
00:03:35,270 --> 00:03:36,260
Change and permissions.
50

51
00:03:36,260 --> 00:03:38,030
Like you need to.
51

52
00:03:38,030 --> 00:03:38,420
Good.
52

53
00:03:38,420 --> 00:03:45,870
Good good you're copying over some of some stuff here.
53

54
00:03:45,880 --> 00:03:49,550
Now you know you can actually make this.
54

55
00:03:50,220 --> 00:03:54,680
Excuse me a copy command and actually now.
55

56
00:03:54,730 --> 00:03:55,240
Never mind.
56

57
00:03:55,240 --> 00:03:57,190
Don't do that.
57

58
00:03:57,220 --> 00:03:58,820
Keep it like it is.
58

59
00:04:00,710 --> 00:04:02,390
You're echoing in a memory limit.
59

60
00:04:02,390 --> 00:04:02,990
That's good.
60

61
00:04:03,020 --> 00:04:03,700
You can overstep.
61

62
00:04:03,710 --> 00:04:06,020
You can set that and override that.
62

63
00:04:06,020 --> 00:04:08,340
You've got a lot of copy files here.
63

64
00:04:08,360 --> 00:04:13,670
It doesn't look like there's a whole lot of opportunity here for consolidating but you could consolidate
64

65
00:04:13,670 --> 00:04:19,760
if they were going to the same directory like right.
65

66
00:04:19,890 --> 00:04:24,900
Well you got some different names on the outs so yeah if you clean it up pretty well.
66

67
00:04:24,960 --> 00:04:34,300
I was just looking for some opportunity to reduce the number of copies of they were doing setup for
67

68
00:04:34,300 --> 00:04:35,520
Cron.
68

69
00:04:35,590 --> 00:04:44,610
Looks like and then doing some last minute shown for the incoming app files.
69

70
00:04:44,610 --> 00:04:45,370
And what did we do.
70

71
00:04:45,380 --> 00:04:53,000
Do we do or copy our our code worked or at folder.
71

72
00:04:53,000 --> 00:04:56,440
You've got a health check in your doctor file a plus.
72

73
00:04:56,450 --> 00:04:59,020
So I love seeing health checks and Doctor files.
73

74
00:04:59,060 --> 00:05:03,770
Cuba daddies won't take advantage of them but doctors swarm will and that way you don't have to depend
74

75
00:05:03,770 --> 00:05:06,440
on an operations person to figure those out later.
75

76
00:05:06,470 --> 00:05:09,770
So yeah at some oh it's a core image.
76

77
00:05:09,800 --> 00:05:13,240
OK so you wouldn't be cupping your ear code into there.
77

78
00:05:13,250 --> 00:05:14,150
Great.
78

79
00:05:14,300 --> 00:05:15,170
And then yeah.
79

80
00:05:15,170 --> 00:05:19,640
And so it looks like what you're doing here is it's up also something that I've used in production many
80

81
00:05:19,640 --> 00:05:27,190
times especially with BHP is you're using supervisor as the base process and it's acting as the unit
81

82
00:05:27,500 --> 00:05:30,290
and then it's launching the various things you need to launch.
82

83
00:05:30,440 --> 00:05:39,350
And I'm assuming that since you have cron and you're gonna have PPF p.m. so there's probably some other
83

84
00:05:39,350 --> 00:05:43,100
stuff here you're doing with supervisor engine X is gonna need to run.
84

85
00:05:43,400 --> 00:05:48,350
So you're going to have at least three processes I bet as some processes and that's the right way to
85

86
00:05:48,350 --> 00:05:48,650
do it.
86

87
00:05:48,650 --> 00:05:50,880
To add sub supervisor in there.
87

88
00:05:51,050 --> 00:05:55,480
Now if you were trying to avoid supervisor it does get tricky on swarm.
88

89
00:05:55,490 --> 00:05:56,690
Oh and this log.
89

90
00:05:56,690 --> 00:05:56,940
Yeah.
90

91
00:05:56,970 --> 00:06:00,670
So I'm not sure why you're running the slug.
91

92
00:06:01,180 --> 00:06:07,160
And you're not using Docker drivers but I'm sure you have a good reason.
92

93
00:06:08,170 --> 00:06:14,660
But if you're using this log to then get them out into Dockers logging using FTD out.
93

94
00:06:14,680 --> 00:06:20,020
That's really what you should be using FTD Aristide's out for logging and ideally not using this log
94

95
00:06:20,020 --> 00:06:26,430
inside the image but that's the recommended best practice.
95

96
00:06:26,450 --> 00:06:33,740
There are a few cases I've seen where it's necessary to use our source log but yeah overall it's a very
96

97
00:06:33,740 --> 00:06:34,270
standard.
97

98
00:06:34,280 --> 00:06:38,340
I think BHP file and it's looking like you're doing a lot of good stuff here.
98

99
00:06:38,360 --> 00:06:42,200
What one thing you could really do if this is a base image.
99

100
00:06:42,440 --> 00:06:51,390
One thing that you could maybe optimize is figuring out a way to combine your base image with your regular
100

101
00:06:51,390 --> 00:07:00,290
images or use multi-stage to reduce any of this unnecessary if you don't need all this stuff right because
101

102
00:07:00,300 --> 00:07:05,610
there's a lot of stuff you're adding in here and it's I don't know whether this stuff is really needed
102

103
00:07:05,610 --> 00:07:10,410
by the production app or if it's just needed to do stuff before the app starts.
103

104
00:07:10,410 --> 00:07:11,000
Right.
104

105
00:07:11,010 --> 00:07:15,810
So now that we have multi-stage and maybe you're using that in your other files in your files that it's
105

106
00:07:15,810 --> 00:07:20,760
coming from the other Docker files but now that we have multi-stage I really like to break out the build
106

107
00:07:20,760 --> 00:07:27,090
dependencies from the production dependencies and basically at the very top I only put in the production
107

108
00:07:28,020 --> 00:07:33,450
and that's the image that I will use at the end and it sort of flows in that way in case of your if
108

109
00:07:33,450 --> 00:07:34,790
you're looking for an example.
109

110
00:07:34,890 --> 00:07:40,770
Again going back to my Docker can't talk I go through a walking example in that Docker can talk here
110

111
00:07:42,720 --> 00:07:48,570
on how you might break out and I think I had four or five stages where I had an initial stage is just
111

112
00:07:48,570 --> 00:07:54,270
for production and sort of the base layer and then I have a dev dependency layer but that layer is never
112

113
00:07:54,270 --> 00:07:55,950
used for production.
113

114
00:07:56,190 --> 00:08:01,020
It's only used for development and all that stuff and then I have a build phase and have a testing phase
114

115
00:08:01,740 --> 00:08:04,380
or stage and then I have the final production stage.
115

116
00:08:04,380 --> 00:08:10,620
So it certainly makes the files look more complicated but I think at the end it reduces complexity of
116

117
00:08:10,620 --> 00:08:17,040
your final image it reduces to attack surface so you maybe are already doing that in your other projects
117

118
00:08:18,120 --> 00:08:27,290
and oh you're saying supervisor has problems with logging so interesting other than that I think it
118

119
00:08:27,350 --> 00:08:28,910
looks like a great Docker file.
119

120
00:08:28,910 --> 00:08:34,280
I think you've got to stop signal in there which I see very rarely I thought cig term was a default
120

121
00:08:34,580 --> 00:08:38,050
but nothing wrong with setting a default in there just to be more literal.
121

122
00:08:38,570 --> 00:08:44,690
But it is pretty cool that you know about that because I rarely see in Docker files because people don't
122

123
00:08:44,690 --> 00:08:50,660
realise that that stop signal option exists so good for you on that because I think a lot is something
123

124
00:08:50,660 --> 00:08:51,620
that my Docker can talk.
124

125
00:08:51,620 --> 00:08:54,560
I talked a lot about was shutting down apps.
125

126
00:08:54,650 --> 00:08:58,570
So the importance of properly shutting down connections.
126

127
00:08:58,700 --> 00:09:02,240
I'm sure it looks like something that you've already known about and learned about in our handling in
127

128
00:09:02,240 --> 00:09:02,660
your app.
128

129
00:09:02,660 --> 00:09:04,100
So good stuff.