0:00

Hey, it's Ryan Seacrest. Life comes at

0:02

you fast, which is why it's important

0:04

to find some time to relax. A

0:06

little you time. Enter Chumba Casino with

0:08

no download required. You can jump on

0:11

any time, anywhere for the chance to

0:13

redeem some serious prizes. So treat yourself

0:15

with Chumba Casino and play over a

0:17

hundred online casino style games. All for

0:20

free. Go to Chumba Casino.com to collect

0:22

your free welcome bonus. Sponsored by Chumba

0:24

Casino. This

0:29

podcast is sponsored by TalkSpace. You know

0:31

when you're really stressed or not feeling

0:33

so great about your life or about

0:35

yourself? Talking to someone who understands can

0:37

really help. But who is that person?

0:40

How do you find them? Where do

0:42

you even start? TalkSpace. TalkSpace makes it

0:44

easy to get the support you need.

0:46

With TalkSpace, you can go online, answer

0:48

a few questions about your preferences, and

0:50

be matched with a therapist. And because

0:52

you'll meet your therapist online, you'll have

0:54

to take time off work or arrange

0:56

childcare. You'll meet on your schedule, wherever

0:58

you feel most at ease. If

1:01

you're depressed, stressed, struggling with

1:03

a relationship, or if you

1:05

want some counseling for you

1:07

and your partner, or just

1:09

need a little extra one-on-one

1:11

support, TalkSpace is here for

1:13

you. Plus, TalkSpace works with

1:15

most major insurers, and most

1:18

insured members have a $0

1:20

copay. No insurance? No problem.

1:22

Now get $80 off of

1:24

your first month with promo

1:26

code space 80 when you

1:28

go to talkspace.com. Match with

1:31

a licensed therapist

1:34

today at

1:36

talkspace.com. Save

1:38

$80 with code space

1:41

80 at

1:43

talkspace.com. And we have

1:46

a very special guest today.

1:48

Joseph, do you want to

1:50

introduce yourself and tell

1:52

everybody why you're Ruby

1:54

famous here? Yes. Hello

1:57

everyone. So I'm Joseph.

1:59

I am... freelance software

2:01

engineer. And apparently now

2:03

also a book author,

2:05

I wrote two books.

2:07

The first one was

2:09

called Deployment from Scratch,

2:11

and the second one

2:13

is called Camelhan Book.

2:15

And that's probably why

2:17

we are here today.

2:21

Yeah, you know that Kamala

2:23

is like the one of

2:25

the latest crazes with all

2:27

the new rails announcements that

2:29

came out of rails world

2:31

So it's kind of exciting

2:34

to have an expert here

2:36

or at least somebody who

2:38

has Doved into the trenches

2:40

and put together all the

2:42

details for us to You

2:44

know expose the the secrets

2:46

of how minimalistic it is.

2:48

I hope I I honestly

2:50

haven't had a chance to

2:52

play with Kamal, so I'm

2:54

looking forward to hearing your

2:56

take on how it all

2:58

comes together. Yeah, I did

3:00

a deploy or two before

3:02

the show just so I

3:05

know. So I guess we

3:07

can start maybe at a

3:09

high level like, you know,

3:11

what does the deploy process

3:13

like look like, you know,

3:15

maybe before Kamal and why

3:17

Kamal is even here? Yeah,

3:20

so maybe it would be

3:22

interesting to see it from

3:24

the point of 47 signals,

3:26

because there are the company

3:29

behind Kamo, right? Kamel, actually

3:31

started by David Halmer Hansen,

3:34

so out there of rails,

3:36

he also started Kamo. And

3:38

they were doing the deployment

3:41

with Kapistan, which is a

3:43

simple Ruby deploy tool. You

3:46

basically have all your little...

3:48

tasks predefined in Ruby and

3:50

then you simply run them

3:53

there on the server and

3:55

something happens you know copies

3:58

channel at the time didn't

4:00

do much assumptions about what

4:02

will be on the server

4:05

or so on you could

4:07

define your own tasks so

4:10

you couldn't drive the deployment

4:12

the way you wanted it

4:14

was simply a tool given

4:17

to you and then they

4:19

jumped as everybody else on

4:21

this Kuberneti strain right like

4:24

big deployment that have infinite

4:26

scale. And to me at

4:29

the time, I remember it,

4:31

I was a little bit

4:33

sad because I was like,

4:36

they were the heroes of

4:38

kind of marketing the simple

4:41

way of deploys, right? So

4:43

not everybody has to be

4:45

on Kubernetes. And then it

4:48

was announced that they are

4:50

making... the leaf from cloud,

4:53

right? The big saving journey

4:55

of leaving the cloud. And

4:57

I know they wanted to

5:00

make a burnities work. They

5:02

went to some vendors, you

5:05

know, from Sousa and so

5:07

on. And David was talking,

5:09

writing about that. But then

5:12

he was really thinking, okay,

5:14

maybe we can do something

5:17

on our own, which is

5:19

not silly, so unusual for

5:21

David, I would say. And

5:24

so they kind of. So

5:26

what state is the SSH

5:29

kid, which is like Ruby

5:31

library, so Kamal is design

5:33

around Docker from the ground

5:36

up, you know, and because

5:38

of that, it also made

5:40

some assumptions to make something

5:43

a little bit easier. So

5:45

what state is the SSH

5:48

kid, which is like Ruby

5:50

library that simply connects over

5:52

SSH to your server. But

5:55

instead of running. some just

5:57

pure brush cold. You

6:00

still run, run, you still

6:02

run tasks, but they are

6:04

usually around running ducar commands.

6:07

So, come on, can install

6:09

ducar, and then simply run

6:11

various ducar commands tasks. The

6:13

simplest way, how you can

6:16

think about it at the

6:18

simplest level, is simply that

6:20

the car will connect and

6:22

run ducar, run, name of

6:25

your application, you know. And

6:27

it doesn't even have a

6:29

specific ducer pool task, for

6:31

example, because that's already something

6:34

that ducer does itself. So

6:36

common really tries to be

6:38

as minimalistic as possible and

6:40

as simple as possible. There

6:43

are some limitations to that,

6:45

of course, and I would

6:47

say the simplicity may confuse

6:49

people sometimes. Because as you

6:52

said, so what is the

6:54

deploy process, right? And usually

6:56

I would say before the

6:58

deploy itself, you prepared the

7:01

server beforehand, which we can

7:03

call provisioning, you know, part

7:05

of that's that can be

7:08

even spinning up the virtual

7:10

machines. For example, it's what

7:12

most people would probably do,

7:14

unless you have physical servers,

7:17

and then installing some software

7:19

firewalls and so on. It's

7:21

just that we've come out,

7:23

we don't need anything else

7:26

for the come-out to war.

7:28

then docor. And even docor

7:30

can be installed by command.

7:32

So it simplifies it a

7:35

lot, but it's still like

7:37

one piece of the puzzle.

7:39

Maybe you need other tools.

7:41

You want to install some

7:44

monitoring or recup scripts or

7:46

whatever. So just one part

7:48

of the whole of the

7:50

whole process. Nice. So, um,

7:53

noob question straight off the

7:55

bat. Like, I assume this

7:57

will also be, maybe a

7:59

source of confusion for someone

8:02

who does... doesn't know a

8:04

whole lot about DACA or

8:06

just like the modern deployment

8:08

complexity and stuff, but can

8:11

you explain what the difference

8:13

between Kuveneti and Kamal is

8:15

for people who don't understand

8:17

either of them? Yeah, so

8:20

basically you can split this

8:22

kind of deploy systems to

8:24

two categories. One is doing

8:26

push. to the server. So

8:29

we are pushing your configuration

8:31

to the server, your changes

8:33

to the server. In terms

8:35

of Kamal, that means from

8:38

your local computer or the

8:40

CI server, we will run

8:42

Kamal the Poi and Kamal

8:44

will issue the concrete commands

8:47

that have to run on

8:49

all those servers that you

8:51

defined. And Kubernetes is the

8:53

other category, which we call

8:56

like a pool systems, where

8:58

usually you have... one host

9:00

that runs the main master

9:02

agent in so sort of

9:05

way and the other servers

9:07

run some lightweight minion agents

9:09

and so the central system

9:11

is that one master host

9:14

and all the systems around

9:16

it can pull the information

9:18

from it. So they constantly...

9:20

to fix the new information.

9:23

And if they say, okay,

9:25

there is a new version,

9:27

then they go and pull

9:29

the information themselves. Oh, cool.

9:32

That makes sense. So, like,

9:34

why would, so me as

9:36

an indie developer who is

9:38

building very small apps on

9:41

very small scale, and I'm

9:43

thinking, do I even need

9:45

Dhaka, like, what's your opinion

9:47

on that? Yeah, in terms

9:50

of what I was thinking

9:52

about the split between like

9:54

Kubernetes and the agent systems,

9:56

this is what I think

9:59

that you do. don't really

10:01

need for small scale, right?

10:03

Because you don't need extra

10:05

resources to run the agents.

10:08

You don't need the extra

10:10

master server that maybe also

10:12

need some kind of backup

10:14

to be available and so

10:17

on. But as you say,

10:19

there's still the question about

10:21

darker itself, like isn't even

10:23

darker overhead, right? I saw

10:26

some people commenting on. Ruby

10:28

subreddit and they were like

10:30

kind of criticizing Kamal for

10:32

using Docker when THH wants

10:35

rails to be no build,

10:37

as the no build, right,

10:39

on the front end, but

10:41

then introducing build process on

10:44

the, on the server side.

10:46

But, Booker still has several

10:48

advantages. First of all, we

10:50

really package. pre-package the application

10:53

in the container. So we

10:55

can be sure of what's

10:57

really inside and that at

10:59

this revision the application was

11:02

running, which is important when

11:04

you will be doing a

11:06

rollback, you know, when something

11:08

wrong happened and you want

11:11

to roll back, then you

11:13

know that the previous version

11:15

was actually running fine. The

11:17

problem with copy strano way

11:20

of doing that is that

11:22

suddenly you are trying to

11:24

switch back to Ruby version

11:26

that hopefully is there, but

11:29

maybe if there will some

11:31

system changes, you know, system

11:33

updates, system utilities, and a

11:35

lot of real applications also

11:38

shell out to call some

11:40

system system tools, right? So

11:42

suddenly you don't even know

11:44

what will happen because it

11:47

can go all wrong. Yeah,

11:50

for those that aren't aware

11:52

Kavanaugh is all like local

11:54

file system and just has

11:56

like a, you know, sim

11:58

link cutoff to switch. whatever

12:01

production version you're using right

12:03

behind the scenes. Yeah, like

12:05

TLDR of it. There's all

12:07

this stuff, right? Yeah, exactly

12:10

as you say, and of

12:12

course, yes, it's simply a

12:14

trade-off. You can't really say

12:16

it's so much better or

12:19

so much worse. I think

12:21

for modern deploys, it's it's

12:23

preferable to do it with

12:25

something like doker or different

12:27

container technology. But I do

12:30

sympathize somehow to the local

12:32

criticism because you now need

12:34

a doctor registry to save

12:36

your images, which by the

12:39

way David suggested that they

12:41

will get rid of in

12:43

some future versions of Kamal.

12:45

So hopefully that will go

12:48

away. But yeah, it's a

12:50

new thing to learn basically.

12:52

So I understand. especially because

12:54

when I was writing my

12:56

first book deployment from scratch,

12:59

I also basically did something

13:01

like Capist Channel, except I

13:03

didn't even use Ruby. I

13:05

simply do it with Bash

13:08

only, you know. So you

13:10

don't have to learn any

13:12

other tool and you can

13:14

still install and run. And

13:17

the biggest advantage there is

13:19

that the switch, the update,

13:21

can be really fast. If

13:23

you are not even installing

13:26

a new version of Ruby

13:28

Jam, for example, it's really,

13:30

it's really super fast. But

13:32

I think the other advantages

13:34

of the credibility really really

13:37

are better for most people,

13:39

for majority of people. And

13:41

when I saw Kamal as

13:43

a tool, I was, I

13:46

told myself, okay, this is

13:48

it. because since I had

13:50

this Bash script kind of

13:52

deploy, you know, very, very

13:55

straightforward. I was thinking myself

13:57

of doing some kind of

13:59

small tool, something like Kamal.

14:01

At the time, I was

14:03

thinking maybe using some components

14:06

like keddy for web server,

14:08

you know, or maybe podman

14:10

instead of docor, because my

14:12

background is in like federally

14:15

nukes and stuff. But I

14:17

know that my little tool,

14:19

again, be just for me,

14:21

you know. So I thought

14:24

like Kamal is the closest

14:26

to what I want. And

14:28

so that's how I basically

14:30

started to be interested and

14:32

started to look more into

14:35

it. And I really liked

14:37

Kamal's direction to make it

14:39

just easier to jump on

14:41

darker without having to like

14:44

necessarily worry about like how

14:46

you configure darker and like

14:48

connect all of the services

14:50

and things like that out

14:53

of the box anyway, right?

14:55

And then, because it sets

14:57

you up kind of like,

14:59

okay, if you needed to

15:01

scale, you could then move

15:04

to some other, you know,

15:06

provider like AWS or something

15:08

that could scale the nodes

15:10

for you pretty easily and

15:13

you can reuse all the

15:15

configuration stuff. And so it's

15:17

a little exciting in that

15:19

way. And it makes it,

15:22

it makes me think though,

15:24

like, like, so how does

15:26

this compare to like, you

15:28

know, tools like Chef or,

15:30

you know, some other configuration

15:33

management aspects of like, like,

15:35

like, you know, the system

15:37

deployment process, like where are

15:39

some advantages that Kamal could

15:42

lead us? Where you, you

15:44

know, maybe, I know Chef

15:46

is like a, it takes

15:48

some extreme getting used, there's

15:51

onboarding a ramp like pretty

15:53

steep learning curve. But like

15:55

where do you see Kamal

15:57

like kind of maybe simplifying

15:59

that or how does it

16:02

compare? Or does it not?

16:04

Yeah, I think there is

16:06

this confusion among the the

16:08

post systems because then you

16:11

have exactly as you say

16:13

chef or Ansible and they

16:15

they maybe started more like

16:17

configuration tools, but they also

16:20

offer you a way to

16:22

do a deploy, you know.

16:24

So they're not completely obsolete.

16:26

We've come out because you

16:28

can still use them provision

16:31

the servers and then you

16:33

come out for the deploy

16:35

part. In my case, I

16:37

find that I don't need

16:40

that much of the configuration.

16:42

I mean, come out itself

16:44

even bills even bills. It's

16:46

almost you don't need anything,

16:49

but that's not really true

16:51

and then people are disappointed.

16:53

But for me, I for

16:55

example, included a very small

16:57

Ruby script, which is just

17:00

a little bit wrapper around

17:02

some show commands. How to

17:04

have fun, anytime, anywhere, step

17:06

one, go to Chumba Casino.com.

17:09

Chumba Casino. This podcast is

17:11

sponsored by TalkSpace. You know

17:13

when you're really stressed or

17:15

not feeling so great about

17:18

your life or about yourself?

17:20

Talking to someone who understands

17:22

can really help. But who

17:24

is that person? How do

17:26

you find them? Where do

17:29

you even start? Talk space.

17:31

Talk space makes it easy

17:33

to get the support you

17:35

need. With Talk space, you

17:38

can go online, answer a

17:40

few questions about your preferences,

17:42

and be matched with a

17:44

therapist. And because you'll meet

17:47

your therapist online, you don't

17:49

have to take time off

17:51

work or arrange child care.

17:53

You'll meet on your schedule,

17:55

wherever you feel most at

17:58

ease. If you're depressed, stressed,

18:00

struggling with a relationship, or

18:02

if you want some counseling

18:04

for you and your partner,

18:07

or just need a little

18:09

extra one-on-one support, TalkSpace is

18:11

here for you. Plus, TalkSpace

18:13

works with most major insurers,

18:16

and most insured members have

18:18

a... zero dollar copay. No

18:20

insurance, no problem. Now get

18:22

$80 off of your first

18:24

month with promo code space

18:27

80 when you go to

18:29

talkspace.com. Match with a licensed

18:31

therapist today at talkspace.com. Save

18:33

$80 with code space 80

18:36

at talk space.com. To provision

18:38

some very basic things for

18:40

the server like firewalls, different

18:42

SSHC users for security reasons

18:45

and so on. And in

18:47

this way, you might find

18:49

out that you need so

18:51

little that you don't really

18:53

need a proper tool like

18:56

chef. But at the same

18:58

time, I wouldn't say that

19:00

Kamal makes it obsolete because

19:02

they complement each other. I

19:05

do wonder if David, for

19:07

example, still have in his

19:09

mind to release something complementing

19:11

Kamal because... Kamal is so

19:14

simple and it started as

19:16

a multi-server tool but at

19:18

the same time it cannot

19:20

deploy a load balancer like

19:22

in front of the servers

19:25

right you have a proxy

19:27

that does to get plus

19:29

depose on each of hosts

19:31

like each server but you

19:34

still need like third-party load

19:36

balancer so Kamal is like

19:38

really like a very specific

19:40

tool and I wonder if

19:43

simply Kamal will Maybe get

19:45

a little bit bigger and

19:47

better in this way to

19:49

handle these things or there

19:51

will be something else Maybe

19:54

Would you use Kamal for

19:56

your like on independent single

19:58

server applications as well? Or

20:00

would you use your bash

20:03

script? Whatever approach you thought

20:05

in in your first book?

20:07

So I switched everything to

20:09

Kamal for all my projects

20:12

and and also I was

20:14

working with some clients to

20:16

switch them to come out.

20:18

For me, although I actually

20:20

really used the best scripts

20:23

from from the book, they

20:25

were developed specifically to be

20:27

educational. I didn't say don't

20:29

use tools, don't use Kubernetes.

20:32

But I've write it to

20:34

pure bash to simply show

20:36

people that everything else is

20:38

an obstruction, right? For example,

20:41

when you have Ansible, it's

20:43

written in Python, it's a

20:45

lot of Python code, that

20:47

in the end prepares a

20:49

bash code, the remote server

20:52

and then it executed there,

20:54

right? So I basically skipped

20:56

the part where I have

20:58

to teach someone about different

21:01

tools because, to be honest,

21:03

the tools come and go.

21:05

So the idea for the

21:07

first book was really not

21:10

being about particular tool, but

21:12

more like how this actually

21:14

works. But since we have

21:16

come out, I think it's,

21:18

for me, it's like a

21:21

perfect tool. There is some

21:23

alternative, you can look into

21:25

docu, or maybe even simple

21:27

docu compose, but personally I

21:30

think Kamal has advantage to

21:32

them because they are, it's

21:34

designed for multi-server use from

21:36

the ground up. Basically the

21:39

single server use schedule is

21:41

like edit later, right, with

21:43

Kamal too. And compared to

21:45

docu, there's also a different

21:47

philosophy. It's not like pass.

21:50

It's more like really just

21:52

a deploy tool. But if

21:54

I'm deploying my own application

21:56

that I also develop, I

21:59

think Kamal fits best to

22:01

that use case. You know,

22:03

maybe if you want to

22:05

deploy some public application, maybe

22:08

it's easier to install something

22:10

like qualify or some. are

22:12

the tools that can one

22:14

click install them. But I

22:16

think if you are the

22:19

developer of the application. really

22:21

fits the fields of it

22:23

with the best. And for

22:25

me personally, I really switched

22:28

100%. Yeah, to add on,

22:30

I'm seeing there's a PR

22:32

open right now to add

22:34

a cloud native build text

22:37

to Kamo, which would be

22:39

interesting, where you can basically

22:41

build the app based off

22:43

of other ones pretty easily.

22:45

Yeah, so basically that's. That's

22:48

the single, I would say,

22:50

advantage of docu today, that

22:52

it can be a little

22:54

bit more magical. It somehow

22:57

will build it for you.

22:59

I didn't have to write

23:01

docor file. But lucky for

23:03

us, Wales actually counts of

23:06

docor file, so there is

23:08

no extra work. It is

23:10

not true for every framework.

23:12

But of course, if you

23:14

are adopting docor, I still

23:17

recommend looking into... how to

23:19

build a file, how to

23:21

optimize it. For example, before

23:23

I was trying to squeeze

23:26

the the profile for my

23:28

real applications, you know, I

23:30

was removing all the all

23:32

the files. I don't need

23:35

them to have small footprint

23:37

since you have to do

23:39

the image and pull and

23:41

push. So, so yeah. What

23:43

is like kind of like

23:46

the full process like let's

23:48

say we do like a

23:50

real's new app and we

23:52

wanted to like deploy it

23:55

with Kamal to like some

23:57

Linod or something right or

23:59

digital ocean droplet like what

24:01

is the steps of the

24:04

process there like To just

24:06

like real's new yeah I

24:08

have a you know so

24:10

you know so you so

24:12

let's say you do you

24:15

did a real's new you

24:17

have your doctor file so

24:19

that's one that's one prerequisite

24:21

you have to have for

24:24

the ducer file because maybe

24:26

you are not on rails.

24:28

So let's just mention that

24:30

you do need your server

24:33

to run on port 80.

24:35

You can change it, but

24:37

it's like default if you

24:39

want it to simply work.

24:41

That's the main thing for

24:44

the docor file. Then you

24:46

will need a docor registry.

24:48

So make an account, a

24:50

docor hub or maybe your

24:53

provider hers, it's own register

24:55

that you can use. And

24:57

you need to create like,

24:59

they call it, come out,

25:02

because it's still like registry

25:04

passport, but you should create

25:06

a token and note it

25:08

down. And then you will

25:10

need to create some virtual

25:13

machine. It can be headsner,

25:15

it can be digital ocean,

25:17

and a cloud provider like

25:19

that would work. And for

25:22

that you will need to

25:24

create SSH key pairs. So

25:26

you will create key pair,

25:28

that means you have two

25:31

keys, one private and one

25:33

public, right? So the private,

25:35

obviously, you have to protect

25:37

the registry password. And during

25:39

the creation of the virtual

25:42

machine, When you go like

25:44

create, you choose some resources,

25:46

you know, how big the

25:48

VM should be and so

25:51

on. And then it will

25:53

ask you to copy your

25:55

key there. So that's how

25:57

they will provision the VM

26:00

for you. And you should

26:02

disable the password login value

26:04

there just for security. And

26:06

so that's the... That's that's

26:08

what you need to that's

26:11

what you need to have

26:13

you know at the at

26:15

the minimum at the minimum

26:17

like the basic ones. Now

26:20

maybe you also want to

26:22

deploy to a domain name

26:24

so then obviously you have

26:26

to purchase a domain name

26:29

and you will need a

26:31

DSM record at the very

26:33

least type a to point

26:35

your domain to the IP

26:37

address to the public IP

26:40

address you will get for

26:42

the machine you just provisioned.

26:44

And if you have that

26:46

you can basically start with

26:49

Kamal. With Kamal in it,

26:51

there is some basic configuration

26:53

file. There are comments in

26:55

line, so even when you

26:58

never saw it, you can

27:00

usually guess a little bit

27:02

what you will have to

27:04

put there. So you will

27:06

name your application. Kamal calls

27:09

it service. I would also

27:11

say be careful of the

27:13

name because when you are

27:15

doing the duker, a local

27:18

private networking, it's also like

27:20

kind of a name that

27:22

you are using. So you

27:24

cannot like put any characters

27:27

you want there, you know,

27:29

just stick to normal letters

27:31

and dashes and then you

27:33

are fine. So, so we

27:35

are not surprised later that

27:38

it doesn't work. Then you

27:40

specify the image. which is

27:42

your container registries username like

27:44

slash the application name basically

27:47

and then you can put

27:49

the IP address among servers.

27:51

When you do that you

27:53

are configuring your application server

27:56

by default the role called

27:58

web and you can. return

28:00

just after this one, how

28:02

Kamal splits the application circuit.

28:04

roles and accessories. But for

28:07

now if you just put

28:09

there the IP address that's

28:11

enough and then you will

28:13

configure proxy you will say

28:16

okay I want SSL and

28:18

you will put their host

28:20

which is your domain name

28:22

or it can be also

28:25

sub domain. It can be

28:27

even more hosts if you

28:29

want. So very very flexible.

28:31

This is very nice. Then

28:33

you finish up with the

28:36

registry and then there's builder

28:38

configuration, but by default I

28:40

think there's some reasonable value.

28:42

And once you have all

28:45

this, then you only need

28:47

to provide the environment, which

28:49

is the secrets fee. We

28:51

talk about like common registry

28:54

password, but it can be

28:56

many more secrets that the

28:58

application have, obviously. So you

29:00

list them. There is basically

29:02

some environment variables you can

29:05

list directly, but the secret

29:07

ones you actually just put

29:09

under like end secret in

29:11

the amo file, and then

29:14

you only list them, but

29:16

they leave somewhere else. When

29:18

you did the Kamal in

29:20

it from before, Kamal also

29:23

created.coma slash secret file. That's

29:25

where the secret is. And

29:27

of course, then you can

29:29

also call password managers and

29:31

so on. But at the

29:34

very basic one, you would

29:36

just put the secrets there.

29:38

You will not check it

29:40

out. Check it in the

29:43

source control. And then your

29:45

configuration is done. You run

29:47

Kamal setup. And Kamal does

29:49

everything for you. It will

29:52

connect to the server. It

29:54

will install the occur. It

29:56

will install the... come out

29:58

proxy and then it will

30:00

book your application. So that's.

30:03

the most minimalistic way of

30:05

the point with Kamal. Gotcha.

30:07

So I'm walking through the

30:09

commands, like as you were

30:12

reading through that, and just

30:14

so like we can, for

30:16

the most people know, Heroku,

30:18

very like, Heroku new, and

30:21

then deploy, you know, get

30:23

push for Well, everything you

30:25

just said, right, which is

30:27

pretty wild. Where we have

30:29

like, come all in it,

30:32

creates all the configuration files,

30:34

come all config, like, you

30:36

know, starts to create your

30:38

initial configuration file to add

30:41

all of the stuff to,

30:43

and then you have come

30:45

all deploy that just, like,

30:47

push that up and proxy

30:50

to the new version, right?

30:52

of the app. Yeah, yeah,

30:54

basically, as you say, I

30:56

mean, the camera config is

30:58

already created if camera in

31:01

it and come out config

31:03

simply rejects the configuration. It's

31:05

good. That's actually good point

31:07

to run before the deploy.

31:10

Yeah, that's awesome. I mean,

31:12

that's where we want to

31:14

head, right? Like, it's like,

31:16

it's pretty close. I mean,

31:19

I guess like, like, uh,

31:22

Yeah, I mean so many

31:25

like so many hosting providers

31:27

now have that like one

31:29

clip process to just get

31:31

your VM up that It

31:33

almost doesn't even make sense

31:35

to abstract away that process,

31:37

right? To make it easier

31:40

because they already provide that

31:42

for the most part But

31:44

this is really cool. So

31:46

I'm trying to think like

31:50

We walk through a very

31:52

simple like example, right? And

31:55

I imagine like with all

31:57

of the push to solid

31:59

Q and solid X Y

32:01

Z that like this is

32:03

all running on a single

32:06

machine? How does all of

32:08

that work? Like is it

32:10

by default running the database

32:12

on the same web server?

32:15

Yeah, so that's a good

32:17

question for talking a little

32:19

bit about how Kama actually

32:21

splits your application and what

32:23

it actually does for you.

32:26

Because this was a simple

32:28

example. You have one VM,

32:30

one. one container, right? But

32:32

your application probably does a

32:34

little bit more. It has

32:37

to run jobs or some

32:39

kind of other scheduling or

32:41

maybe are running an API

32:43

server or some marketing website

32:46

and then all those needs

32:48

databases. So at the high

32:50

level, Kamel splits this into

32:52

servers and accessories. So servers

32:54

are your application servers. Those

32:57

all run your application image.

32:59

So you cannot build like

33:01

from multiple images. It has

33:03

to be one image. There's

33:06

just different command that it's

33:08

being drawn. But it comes

33:10

from the application. Of course

33:12

you can package whatever you

33:14

want in the container and

33:17

you for example can have

33:19

application. How to have fun,

33:21

anytime, anywhere. Step one, go

33:23

to Chumba Casino.com. Chumba Casino.com.

33:25

Got it. Step two, collect

33:28

your welcome bonus. Come to

33:30

Papa, welcome bonus. Step three,

33:32

play hundreds of casino-style games

33:34

for free. That's a lot

33:37

of games, all for free.

33:39

Step four, unleash your excitement.

33:41

Chumba Casino has been delivering

33:43

thrills for over a decade.

33:45

So claim your free welcome

33:48

bonus now and live the

33:50

Chumba Life. Visit Chumba. Ryan

33:56

Reynolds here from it mobile.

33:58

I don't know if you

34:00

knew this but anyone can

34:02

get the same premium wireless

34:04

for $15 a month plan

34:07

that I've been enjoying. It's

34:09

not just for celebrities, so

34:11

do like I did, and

34:13

have one of your assistance

34:15

assistance to switch you to

34:17

Mint Mobile today. I'm told

34:19

it's super easy to do

34:21

at MintMobile.com/switch. Up front payment

34:23

of $45 for three month

34:25

plan, equivalent to $15 per

34:27

month required. Intro rate for

34:30

three months only. Then full

34:32

price plan options available, taxes

34:34

and full taxes available, taxes

34:36

and fees extra. a jungle

34:38

application next to it, you

34:40

know, but you have to

34:42

put it to the image

34:44

because that's kind of a

34:46

constraint that come on puts

34:48

and you. And with this

34:50

server roles, it's important to

34:53

say that there is a

34:55

primary role by default. It's

34:57

called web, but you can

34:59

rename it. And the primary

35:01

role assumption is that this

35:03

is your main application server.

35:05

That's your rails S, basically,

35:07

or like your puma starting

35:09

up. And why is important?

35:11

Because it plays in a

35:13

role, how come I actually

35:16

does deploy on multiple servers

35:18

and also on each particular

35:20

server. Because when it puts

35:22

your application, it usually puts

35:24

your accessories, which are the

35:26

databases, all these low-running services

35:28

that are not necessarily, I

35:30

mean, they're connected with your

35:32

application, but it's... but it's

35:34

something else. You know, they're

35:36

not part of the Kamal

35:39

deployed process per side. They

35:41

should already be running and

35:43

you are simply taking use

35:45

of it. But the primary

35:47

was important because Kamal changed

35:49

a little bit regarding health

35:51

checks and how it ensures

35:53

that actually the container is

35:55

healthy and can serve traffic

35:57

or can run. before we've

35:59

tried. we had something like

36:02

a court file. I won't

36:04

really go into it because

36:06

it's already kind of obsolete.

36:08

But I will say that

36:10

the new way, there is

36:12

something called the barrier. Maybe

36:14

you won't find this name

36:16

in the dogs, but you

36:18

might find it in the

36:20

code or somewhere in the

36:22

issues. That's how common maintenance

36:25

call it. And the barrier

36:27

means that common weight first

36:29

for your that container to

36:31

boot up successfully. the new

36:33

one, for example, for the

36:35

new version. And once it's

36:37

healthy, then it assumes it

36:39

can boot everything else. So

36:41

all other roles, whether it's

36:43

jobs or marketing sites or

36:45

whatever you put there, will

36:48

also start without any further

36:50

health check, because the only

36:52

camel proxy checks is for

36:54

this primary role. So that's...

36:56

very important to realize, for

36:58

example, you might be deploying

37:00

front and back end, you

37:02

know, split. And if come

37:04

out, you can have front

37:06

and application as one application

37:08

and back end as one

37:11

application. But if you are

37:13

putting it together and you

37:15

are simply splitting it into

37:17

roles, then one of them

37:19

has to be the primary

37:21

that come out will wait

37:23

on and then it will

37:25

also boot the others. So

37:27

that's like. one important thing.

37:29

You can still have health

37:31

checks also for other containers,

37:34

but then it's just the

37:36

regular docor health checks that

37:38

you can add there, but

37:40

nothing to do with the

37:42

come out proxy process of

37:44

booting. Which docor registry would

37:46

you personally recommend people use?

37:48

I personally use the care

37:50

hub right now. I want

37:52

to support them because I

37:54

know they have... hard time

37:57

as a company to actually

37:59

make money out of Tucker.

38:01

And it's great that is

38:03

free and everything, you know,

38:05

but to support them, I

38:07

use them currently. Maybe the

38:09

smart answer is use the

38:11

registry with your provider because

38:13

it might be faster for

38:15

you, you know, might be

38:17

even cheaper because I think

38:20

they're increasing prices right now

38:22

with But I like the

38:24

idea of supporting them in

38:26

some way. So right now

38:28

I'm using the registry. Cool.

38:30

And so the thing with

38:32

this more of a broader

38:34

question around self hosting as

38:36

opposed to using up a

38:38

pass, the two things that

38:40

kind of scare me about

38:43

self hosting, which is why

38:45

I haven't actually deployed an

38:47

app by myself yet I

38:49

use render. The two things

38:51

that kind of scare me.

38:53

One is database backups for

38:55

like, I use postgras personally,

38:57

but I probably use SQLite

38:59

for the next project, which

39:01

would make life a bit

39:03

easier in that regard. But

39:06

database backups and OS upgrades.

39:08

So like for someone who's

39:10

only ever used like Heroku

39:12

or Ender never had to

39:14

worry about this kind of

39:16

stuff. What would you tell

39:18

them? You mentioned backups first,

39:20

right? So in the docor

39:22

way of doing backups is

39:24

likely to simply run a

39:26

docor container that does backups,

39:29

right? That's the docor way

39:31

because you have to run

39:33

everything in docor. You can

39:35

still do manual backups as

39:37

before. I actually mentioned it

39:39

in Kamalhan books, how to

39:41

do it. So you can

39:43

still do manual backup. You

39:45

can call, you know, the...

39:47

The command is a bit

39:50

larger because you have to

39:52

use the... It is actually

39:54

a good point. come out

39:56

to edit aliases so you

39:58

can prepare these commands and

40:00

then have like nice come

40:02

all backup and it will

40:04

do backup right so so

40:06

in a way it works

40:08

the same as normal self

40:10

hosting but it's definitely more

40:13

work than just using managed

40:15

database and I would say

40:17

it might be good for

40:19

people to first just move

40:21

with their server load like

40:23

application servers and wait with

40:25

the database just used to

40:27

manage the database. There's nothing

40:29

wrong with that. In fact,

40:31

I think most people will

40:33

run it like that will

40:36

take advantage of the managed

40:38

database because it saves you

40:40

a lot of hustle which

40:42

is true. But yeah, if

40:44

you want to do it

40:46

yourself, you will try on

40:48

a container there. Then you

40:50

can upload it, for example,

40:52

every day to S3 or

40:54

something like that and also

40:56

have a way to pull

40:59

it back if something happens.

41:01

I also recommend to always

41:03

turn on the automatic VM

41:05

backups, which means the whole

41:07

VM is backup like every

41:09

day on your cloud hosting

41:11

provider. This is especially cool

41:13

for like this project that,

41:15

for example, I deploy, you

41:17

know, they are not big

41:19

yet, they are one per

41:22

server. So I know I

41:24

have snapshot every day. database

41:26

and but also everything else

41:28

so I can just with

41:30

one click to return there

41:32

and continue which is really

41:34

really good but you know

41:36

as it is with backups

41:38

it's good to always have

41:40

like two different backups maybe

41:42

just to be sure so

41:45

I think that extra extra

41:47

assurance can be this virtual

41:49

machine backups as a whole

41:51

yeah And actually with

41:53

the system updates, I think

41:56

it's kind of similar because

41:58

if you have backup of

42:01

the VM and you can

42:03

return and return to

42:05

the to the process later.

42:08

And if Kamal, if you

42:10

have more hosts, you

42:12

can simply put the host

42:15

that you're upgrading like kind

42:17

of offline, you know, upgrade

42:20

the system there and then

42:22

you put it back to

42:25

the pool of servers. That's

42:27

ideal. If you have more

42:30

service and you can do

42:32

it, that's like a safe

42:35

way to to run

42:37

it. Does the current Kamal

42:39

proxy handle like low balancing

42:42

to do like slow

42:44

rollouts of versions of the

42:46

app or? Yeah, yeah, it

42:49

will, it does so-called blue

42:51

green deploy. So on every

42:54

host it will put a

42:56

new application, you're running also

42:59

the old one, then after

43:01

the house check, it switch,

43:04

but then also it switch,

43:06

but then also it

43:08

like goes and runs it.

43:11

across all hosts, right? So

43:13

yeah, there's some settings,

43:15

some little settings around that.

43:18

And it will do the

43:20

rollout for you. Yeah. That's

43:23

really cool. So I'd love

43:25

to shift gears a little

43:28

bit into this whole, like,

43:30

Kamal accessory stuff, because this

43:33

is something I wasn't immediately

43:35

obvious, like, how it all

43:38

works and when's the

43:40

best. Use case, right, where

43:42

when it's not good thing

43:45

to use as an

43:47

accessory, like what is an

43:49

accessory, right? Do you want

43:52

to just give us

43:54

like a high level like

43:56

what the Kamala accessories are

43:59

and like one example of

44:01

like the most common

44:03

one that you would add

44:06

to in that new app?

44:08

Yeah, so basically accessories.

44:10

are come away of running

44:13

like some kind of permanent

44:15

services, you know, that are

44:18

not part of that deploy

44:20

life cycle, they simply run.

44:23

Next to it, whether that's

44:25

relational database is key value

44:28

stores, search, maybe headless browsers,

44:30

you know, anything else that

44:33

you just run as

44:35

a service, basically, as your

44:37

application is using. Most people

44:40

will simply use a

44:42

database as an accessory or

44:44

first accessory. And then you

44:47

can run accessories on the

44:50

same host or different holes.

44:52

Basically, that's what I really

44:55

like about, come out, you

44:57

simply put the IP there,

45:00

IP address, and it will

45:02

deploy on that host, you

45:05

know, whether that was

45:07

an application role or accessory,

45:09

it just puts it there

45:12

where you want it.

45:14

If you want to collocate

45:16

it with the roles, you

45:19

can also just say, just

45:21

put it where the role.

45:24

That is... Ryan

45:29

Reynolds here for Mint Mobile.

45:31

I don't know if you

45:33

knew this, but anyone can

45:35

get the same premium wireless

45:37

for $15 a month planned

45:39

that I've been enjoying. It's

45:42

not just for celebrities, so

45:44

do like I did, and

45:46

have one of your assistance

45:48

assistance to switch you to

45:50

Mint Mobile today. I'm told

45:52

it's super easy to do

45:54

at MintMobile.com/switch. Up front payment

45:56

of $45 for three-month plan

45:58

equivalent to $15 per month

46:00

required. Intro rate for three

46:02

months only. Then full price

46:05

plan options available. Taxes and

46:07

fees extra. See full terms

46:09

at mintMobile.com. it will automatically

46:11

put the application next to

46:13

your, I mean, your database

46:15

next to your application on

46:17

that server. So you don't

46:19

even have to specify the

46:21

IP again, for example. But

46:23

in principle, you can really

46:25

spin it up anywhere, anywhere

46:28

you want. That's neat. So

46:30

like, let's say a more,

46:32

maybe more advanced one would

46:34

be something like sidekick or

46:36

like a background processor. How

46:38

do the employees then work

46:40

like as you deploy a

46:42

new version and it needs

46:44

to like do like a

46:46

rolling restart to wait for

46:49

you know whatever jobs to

46:51

finish before it it cycles

46:53

over to the new version

46:55

for new jobs? What's that

46:57

kind of process? you would

46:59

call it like jobs for

47:01

example, right? And as I

47:03

said before, it's not a

47:05

primary role. So the health

47:07

check there is not silly

47:09

done. It waits for the

47:12

web container to boot. And

47:14

once at least one of

47:16

them boots, so it doesn't,

47:18

it can be on different

47:20

hosts, you know, like once

47:22

it successfully boots, come on,

47:24

okay, I know the application.

47:26

I probably built well with

47:28

the assumption, then it will

47:30

simply start the good process

47:32

for also the other roles.

47:35

So for roles that are

47:37

not web-based, there's this assumption

47:39

that you can simply decommission

47:41

them and start them again.

47:43

If you have more web

47:45

roles, you can. you can

47:47

also say that, okay, run

47:49

the proxy, come up proxy

47:51

with this role, and then

47:53

the request will... be still

47:55

trained. So come up, we

47:58

will still bait for the

48:00

request to finish before it

48:02

will be switching, you know,

48:04

but that's really just the

48:06

roles that are what we

48:08

could call proxy roles, roles

48:10

that use proxy. But the

48:12

other ones, they are not

48:14

really a big assumption of

48:16

like, Kamel is the tool

48:18

that doesn't know that you

48:21

are deploying sidekick or that

48:23

you are deploying post-ge sequel

48:25

accessory, it doesn't know, you

48:27

know. That's why it is

48:29

also not like some kind

48:31

of plug-in to run post-ge

48:33

sequel or my sequel. It's

48:35

the configuration you have to

48:37

do. That's the simplicity, but

48:39

maybe also annoyance for some

48:41

people that you just does

48:44

what you tell it to

48:46

do. Gotcha. So you... It's

48:48

just the interface, you have

48:50

to like kind of figure

48:52

out all the implementation details

48:54

of it, how you're using

48:56

it. Yeah, in the simplest

48:58

way, it's really just trans,

49:00

ducar containers with ducaram, and

49:02

sometimes they are health check

49:05

involved, you know, and it's

49:07

important to learn how the

49:09

life cycle goes like that,

49:11

okay, accessories are booked first,

49:13

then then the servers are.

49:15

who did first, primary role

49:17

than the others, and then

49:19

you realize that then you

49:21

can think about how to

49:23

write a configuration you even

49:25

need. So what are the

49:28

latest releases in version two

49:30

that has gotten you pretty

49:32

excited about? Yeah, so I

49:34

think Kamao is now probably

49:36

interesting even to people that

49:38

very interesting in common one,

49:40

you know. For one, deploying

49:42

to a single server is

49:44

so much easier and straightforward.

49:46

There were ways how to

49:48

do it in common one,

49:51

but you have to configure

49:53

traffic and it was a

49:55

bit annoying. And now it's

49:57

all seamless in the config,

49:59

you just write which holes,

50:01

as is all true. And

50:03

there you have it. It's

50:05

like number one thing for

50:07

Kamal 2, that it out

50:09

of provision to your as

50:11

a cell certificate with lots

50:14

and encrypt. You don't have

50:16

to do anything. It's just

50:18

in the configuration. But there

50:20

is one pivot I would

50:22

just say with that. it

50:24

does not apply for multi-host

50:26

deployment because again Kamal is

50:28

so simple there is no

50:30

centralized really way of sharing

50:32

the the issued certificates for

50:34

example you know the different

50:37

host doesn't know about each

50:39

other so if you would

50:41

be issuing certificates again on

50:43

every host then maybe you

50:45

run out of the attempts

50:47

at that's encrypt or something

50:49

like that. But for single

50:51

server, this big leap forward,

50:53

I would say. But in

50:55

general, the depots get faster,

50:57

you know, because of the

51:00

process I described, so there's

51:02

maybe a little bit less

51:04

health checks done, because there's

51:06

just like one for the

51:08

primary roles, and then we

51:10

are rolling, we are going

51:12

for it. But that makes

51:14

it really faster. Also, Traffic

51:16

has one nice side effect

51:18

that the server CPU utilization

51:21

went down because that was

51:23

one thing that was constantly

51:25

pulling for like. Hello, it

51:27

is Ryan, and I was

51:29

on a flight the other

51:31

day playing one of my

51:33

favorite social spin slot games

51:35

on Chumba Casino.com. I looked

51:37

over the person sitting next

51:39

to me, and you know

51:41

what they were doing? They

51:44

were also playing Chumba Casino.

51:46

Everybody's loving, having fun with

51:48

it. Chumba Casino is home

51:50

to hundreds of casino-style games

51:52

that you can play for

51:54

free anytime, anywhere. So sign

51:56

up now at Chumba Casino.com

51:58

to live the Chumba Casino.

52:00

sponsored by Shumba Casino. No

52:02

purchase necessary VGW Group void

52:04

were prohibited by law 21

52:07

plus terms and conditions apply.

52:09

State of Docker containers, right?

52:11

Because the configuration doesn't live

52:13

on the traffic side, but

52:15

on the container side, so

52:17

it's constantly polling. So if you

52:20

would look, you would have some

52:22

CPU utilization all the time and

52:24

that that's not the case with

52:26

the new new common proxy, you

52:28

know. The proxy is... pretty like

52:30

simply doesn't do too much

52:32

the roads to host, you

52:35

can register service, it takes

52:37

care of draining the request when

52:39

it's switching over, you know,

52:41

and there are not many

52:43

features of like what proxy

52:45

could have like keddy or

52:48

engine X, you know, either

52:50

some basic request response buffering

52:52

or handles like maximum

52:55

request sizes and proxy

52:57

headers, but It doesn't even

53:00

do anything with, let's

53:02

say, your static assets, right?

53:04

For that we have truster,

53:06

so they didn't really

53:08

copy, for example, this

53:10

functionality to come out

53:13

proxy. It's still simple.

53:15

I think simple is

53:18

good. Some people might find

53:20

it limiting. Let's see what

53:22

we'll see in the future, because

53:24

I would say the development of.

53:27

Kamal, since I started

53:30

to look into it, and

53:32

then even since I started

53:34

to rank Kamalhan book,

53:36

was really, really fast.

53:38

Right. I did already like

53:41

two updates, like basically two

53:43

editions, the first one, with

53:45

a big update for

53:47

Kamal 1-7. So yeah, it's

53:50

faster, it's much easier to use,

53:52

I would say, and things are

53:54

things are going well. Yeah,

53:58

you did. You'd

54:00

bring up something that I

54:02

was trying to work through,

54:04

how you were doing Kamal.

54:06

I guess, like, let's say,

54:08

like, you know, you want

54:10

to, like, configure a tune,

54:12

some, like, engine X caching

54:14

or proxies specific in engine

54:16

X or some other system

54:18

level configuration. Is that all

54:20

handled, like, in whatever server

54:22

that you would configure with

54:25

Kamal? Like, it's... it's a

54:27

step away from that aspect?

54:29

And maybe how does it

54:31

like, can it work with

54:33

it? To auto generate that

54:35

kind of stuff? So, come

54:37

I'll work with some assumption.

54:39

As I said, it doesn't

54:41

provision load bonds in front

54:43

of your servers, right? And

54:45

similarly, it doesn't really handle

54:47

some... serving of static assets.

54:49

The assumption is that you

54:51

either use something like cluster

54:53

inside your image, for example,

54:55

so you have one more

54:57

little proxy there. You could

54:59

also have, for example, your

55:01

own engines load balancer, that

55:03

also does the sketching, but

55:05

you will have to deploy

55:07

it yourself. That's the thing.

55:10

Kamal won't do it for

55:12

you. Or there's this assumption

55:14

that most people maybe will

55:16

just proxy with cloud player.

55:18

and do the cashing there,

55:20

you know. So there's some

55:22

limitation to it. For most

55:24

people, I think there's a

55:26

way how to do it.

55:28

It's just like, it's not

55:30

handled directly and come out

55:32

right now. Gotcha. Yeah, that

55:34

makes sense. I mean, still

55:36

early days. Yeah, yeah, I

55:38

think so. Well, cool. So

55:40

is there any like, dashboard

55:42

or monitoring or something like

55:44

that, that's available to like

55:46

kind of see things. I

55:48

see there's like a, you

55:50

know, come all details. where

55:52

you can kind of peek

55:54

at the apps and services

55:57

running. Is there any like

55:59

UX stuff out there that

56:01

people are using to visualize

56:03

or gather metrics or things

56:05

like this? Well, I believe

56:07

that for the sounds that

56:09

they use something like from

56:11

elastics, such tag like a

56:13

phenomena or something like that.

56:15

thing server depots for example

56:17

I take advantage of digital

56:19

ocean agent because you can

56:21

just click you want it

56:23

and then you have very

56:25

beautiful dashboard that you get

56:27

for free basically you can

56:29

even set up alerts so

56:31

for very small hosting for

56:33

example what I do I

56:35

think that works very nicely

56:37

I know that somebody is

56:39

making something but it was

56:42

not announced so maybe there

56:44

will be some come out

56:46

specific tooling but right now

56:48

we don't have it so

56:50

you have to do the

56:52

usual way if you want

56:54

host monitoring install that data

56:56

dot agents you know or

56:58

being Grafana is definitely something

57:00

that's not really solved in

57:02

the I would say easy

57:04

way to just to just

57:06

like enable and run with

57:08

but I think that would

57:10

be some nice Nice project

57:12

for somebody for sure. Hey,

57:14

you definitely got me wanting

57:16

to deploy a couple of

57:18

apps just for fun. So

57:20

what are maybe some things

57:22

that you're looking forward to

57:24

or maybe hopeful that they

57:27

get added? I mean, outside

57:29

of like the two items

57:31

that we just talked about

57:33

with the, you know. maybe

57:35

making proxying a little easier

57:37

or the metrics aspects like.

57:39

Is there anything missing from

57:41

the process that you like

57:43

hope gets at it? For

57:45

me, I was kind of

57:47

already happy with Kamal 1.

57:49

I mean, now that I

57:51

see Kamal 2, I'm even

57:53

more happy. But I kind

57:55

of already made it work,

57:57

you know, it did the

57:59

job. And for example, when

58:01

you, when we were comparing

58:03

with Kapistrano in the beginning,

58:05

you know, it doesn't also

58:07

have some kind of dashboard

58:09

for you, you know. because

58:11

the tool is limited. And

58:14

in some way, I think

58:16

it's good that it has

58:18

some kind of limits or

58:20

some kind of philosophy of

58:22

what it wants to be.

58:24

And then there can be

58:26

a sister project that provides

58:28

the extra features. So I

58:30

will talk about, of course,

58:32

like this kind of dashboard

58:34

or maybe even nicer backups,

58:36

all these things would be

58:38

nice or maybe provisioning. Actually,

58:40

it was my idea that

58:42

I could develop. this project

58:44

that it would be like

58:46

zero config provisioning for Kamo

58:48

with the headline, you know.

58:50

I don't know if I

58:52

will get to it, but

58:54

it's an idea I have.

58:56

So personally, I would see

58:59

it as sister projects. I

59:01

also think that David suggested

59:03

that yeah, Kamo will probably

59:05

stay focused and I think

59:07

I think it's a good

59:09

thing because when I look

59:11

at Docu for example, it

59:13

was like the single server

59:15

pass and it was very

59:17

well for that. But now

59:19

I checked the docs recently

59:21

and I see that there

59:23

is also a way how

59:25

to spin up a K3S

59:27

scheduler. You know, and then

59:29

you are like kind of

59:31

deleting the tool to be

59:33

everything for everybody. And that's

59:35

also in the end negative.

59:37

for me. So I'm happy

59:39

if it stays simple, you

59:41

know, improves in little ways.

59:44

Yeah, I hear you. It's

59:46

going to be hard to

59:48

shake, like, the existing, like,

59:50

larger infrastructure stuff, focusing on,

59:52

like, the very specific, simple

59:54

utility of Docker maintenance and

59:56

services. Definitely is appealing. Like,

59:58

going off of that, then,

1:00:00

is there anything like you

1:00:02

would like to see removed

1:00:04

from the overall, like, architecture

1:00:06

process that maybe is just

1:00:08

causing more headache and confusion

1:00:10

than is adding value? Not

1:00:12

really. I have to say

1:00:14

that the change with secrets

1:00:16

for me was a little

1:00:18

bit maybe a downside in

1:00:20

my specific way because what

1:00:22

I was doing, since I

1:00:24

knew how Kama really worked,

1:00:26

I kept on CI maybe

1:00:28

just like two amps and

1:00:31

I could like just do

1:00:33

deployed and not until it

1:00:35

put my secrets to to

1:00:37

get up, you know, which

1:00:39

I was very happy about.

1:00:41

And now you either have

1:00:43

to put it there or

1:00:45

use a secret manager, which

1:00:47

of course, if you are

1:00:49

a company, you should be

1:00:51

using secret management. But for

1:00:53

me, I was like, I

1:00:55

was just little projects and

1:00:57

I was happy that I

1:00:59

could do it this way

1:01:01

and now I can't. But

1:01:03

in other, I still think

1:01:05

it's probably better. It's just

1:01:07

like, in my specific case,

1:01:09

I was happy to deploy

1:01:11

without the without the change.

1:01:13

two environments because before it

1:01:16

was two steps, come on

1:01:18

with five, this is push

1:01:20

the amps and then come

1:01:22

on deploy. But now it's

1:01:24

all together. So you do

1:01:26

need all amps because if

1:01:28

you don't have them, then

1:01:30

it will fail. So there's

1:01:32

maybe like one thing in

1:01:34

my specific case, but I

1:01:36

think generally it's still a

1:01:38

step forward for for common.

1:01:40

And you know, I wish

1:01:42

people would Maybe replace some

1:01:44

Kubernetes clusters, but to be

1:01:46

honest, it's probably not what

1:01:48

will necessarily have. I think

1:01:50

the most value from come

1:01:52

out can get people that

1:01:54

are now deploying with just

1:01:56

look at compose. Because look

1:01:58

at compose, it's originally a

1:02:01

deaf tool, you have to

1:02:03

put all the containers down

1:02:05

and up. And I know

1:02:07

there are some third-party proxies

1:02:09

that tries to solve this

1:02:11

get-less deploy, but Kamel is

1:02:13

so much nicer in my

1:02:15

opinion, you know, and whatever

1:02:17

criticism we would have for

1:02:19

Kamel. It would probably apply

1:02:21

there as well. So I

1:02:23

think it's just like a

1:02:25

plus, just a better tool.

1:02:27

It is Ryan here and

1:02:29

I have a question for

1:02:31

you. What do you do

1:02:33

when you win? Like are

1:02:35

you a fist-pumper? A woo-hooer!

1:02:37

A hand clap or a

1:02:39

high-fiver? If you want to

1:02:41

hone in on those winning

1:02:43

moves, check out Chumba Casino.

1:02:45

Choose from hundreds of social

1:02:48

casino-style games for your chance

1:02:50

to redeem serious cash prizes.

1:02:52

There are new game releases

1:02:54

weekly plus free daily bonuses

1:02:56

so don't wait. Start having

1:02:58

the most fun ever at

1:03:00

Chumba Casino Casino. But

1:03:06

yeah, we'll we'll see.

1:03:09

Yeah, I think we've

1:03:11

we've covered quite a

1:03:14

lot of topics here.

1:03:16

We're kind of coming

1:03:18

to, you know, time

1:03:21

here, but is there

1:03:23

any other things you

1:03:25

wanted to cover before

1:03:28

we move into our

1:03:30

pick section of the

1:03:33

show? Not not

1:03:35

really cover but if I

1:03:37

can I would just like

1:03:39

want to thank you David

1:03:41

for even starting come out

1:03:43

and Also think don't I

1:03:45

mean who? Done a lot

1:03:47

of work on come out

1:03:49

too and also wrote a

1:03:51

forward so thank you Donal

1:03:54

and also to Kevin McConnell

1:03:56

because he was the main

1:03:58

guy behind Kamal proxy. She's

1:04:00

a very important aspect of

1:04:02

Kamal too. And then also

1:04:04

a final shout out to

1:04:06

Rosa because she kind of

1:04:08

sent Don't all the information

1:04:10

that I am writing the

1:04:12

book and then he wrote

1:04:14

me in the email and

1:04:17

that's how it all happened.

1:04:19

So yeah, thank you. Thank

1:04:21

you guys. And everybody else

1:04:23

that you know works on

1:04:25

Kamal, I would like to

1:04:27

think. Yeah,

1:04:30

seriously. Thank you. All this

1:04:32

stuff is definitely going to

1:04:34

make it easier to dock

1:04:36

or at least. Should be

1:04:39

should help with deploys in

1:04:41

general. It's definitely helpful to

1:04:43

have a definitive rails process

1:04:45

for deployments, which was missing

1:04:48

for quite a long time.

1:04:50

So it's exciting to see.

1:04:52

Well, great. We're going to

1:04:54

move into PICS. It's a

1:04:57

segment of their show where

1:04:59

we just... Find some things

1:05:01

that we're excited about in

1:05:03

life or code or whatever

1:05:06

it may be that pique

1:05:08

your interest. And I can

1:05:10

kick us off. I have

1:05:12

this particle, I guess, infatuation.

1:05:15

It's a little hardware devices,

1:05:17

little microcontrollers. They have this

1:05:19

little development board called the

1:05:21

Muon that lets you do

1:05:24

things with cellular and mesh

1:05:26

it with like You know

1:05:28

Wi-Fi so you can get

1:05:30

like kind of like a

1:05:33

blended solution on a microcontroller.

1:05:35

It's a lot of fun.

1:05:37

I'm messing with Open A-I's

1:05:39

new pipeline for fine-tuning stuff

1:05:42

They call it distillation. It's

1:05:44

a lot of fun where

1:05:46

you could just send all

1:05:48

of your... AI prompt and

1:05:51

responses to open AI and

1:05:53

they'll automatically fine tune a

1:05:55

new model for you based

1:05:57

on the outputs and inputs

1:06:00

that you give it, which

1:06:02

is just wild. So I'm

1:06:04

having a lot of fun

1:06:06

just messing around and seeing

1:06:09

what I can get it

1:06:11

to do. Yeah, I just

1:06:13

wanted to say like, also

1:06:15

like this, just little improvements

1:06:18

in life that you can

1:06:20

do, for example, on my

1:06:22

travels this year, I switched

1:06:24

to finally using e-Sem, so

1:06:27

I didn't have to like,

1:06:29

you know, by physical Sim,

1:06:31

change on the airport when

1:06:33

I was like, 30 and

1:06:36

tired and everything, because I

1:06:38

just like clicked on my

1:06:40

phone and I had service.

1:06:42

And it's like this little

1:06:45

things that like actually are

1:06:47

really amazing that I love.

1:06:49

But since my book is

1:06:52

about also like kind of

1:06:54

doing something independently, you know,

1:06:56

like in the in the

1:06:58

hacking spirit, I want to

1:07:01

do a recommendation for one

1:07:03

show. It's called the Indie

1:07:05

Life. It's by Dagobert. He

1:07:07

was doing. a startup, he

1:07:10

kind of failed, he also

1:07:12

got divorced, and now he's

1:07:14

like starting this show where

1:07:16

you, where he discusses these

1:07:19

people, also the not so

1:07:21

glamorous side of this in

1:07:23

the acting entrepreneurship and stuff.

1:07:25

It's a nice show, so

1:07:28

I would recommend this to

1:07:30

check out. Cool. Is there

1:07:32

any easem provider that you'd

1:07:34

recommend because I'm traveling to

1:07:37

New Zealand in a couple

1:07:39

of an hour? about three

1:07:41

weeks and I was looking

1:07:43

at reasons as well. It

1:07:46

was just my first time

1:07:48

I picked one that... some

1:07:50

of the countries I go

1:07:52

and that's it, but it

1:07:55

worked, it worked nice. It's

1:07:57

probably a bit more expensive,

1:07:59

but I mean, I really

1:08:01

appreciate the leveling up in

1:08:04

the comfort of travel, so

1:08:06

for me. Yeah, yeah, yeah,

1:08:08

it's definitely worth the price.

1:08:10

Right, my picks, I've got,

1:08:13

I've got a movie in

1:08:15

a TV show, so. I

1:08:17

watched a movie last night

1:08:19

called A Night on Earth

1:08:22

by Jim Jarmush. I think

1:08:24

it's from 1991. Absolutely brilliant

1:08:26

if you like character driven

1:08:28

movies that don't have a

1:08:31

lot of plot but have

1:08:33

very colorful characters. This is

1:08:35

the kind of movie for

1:08:37

you I think. Really enjoyed

1:08:40

a guy. I met at

1:08:42

the Haggis Ruby Conference last

1:08:44

week. I recommended this movie.

1:08:46

That's why I watched it.

1:08:49

and yeah I'd recommend it

1:08:51

to everyone else. And season

1:08:53

two of a show called

1:08:55

Shrinking has just started on

1:08:58

Apple TV. I loved season

1:09:00

one and season two started

1:09:02

off really strongly as well.

1:09:04

I think four episodes are

1:09:07

at. It's starring Harrison Ford

1:09:09

and Jason Siegel from any

1:09:11

of my who's familiar with

1:09:13

how I met your mother

1:09:16

would know who he is

1:09:18

know who he is who

1:09:20

he is know who he

1:09:22

is And Ted Lasso fans

1:09:25

will also be happy to

1:09:27

hear that Brett Goldstein has

1:09:29

now has joined the cast

1:09:31

for season two. So if

1:09:34

any of the names ring

1:09:36

a bell, I'd be sure

1:09:38

checking out because I'm a

1:09:40

huge fan. Yeah, those are

1:09:43

my two picks. Awesome. Yeah,

1:09:45

I always love a good

1:09:47

TV movie show combo. Well,

1:09:50

you also it was it was

1:09:53

really great having you on today

1:09:55

Thank you for sharing all

1:09:57

of your, you know deep diving

1:09:59

on Kamal and Those that haven't

1:10:02

checked it out yet, check out

1:10:04

all of Joseph's great books that

1:10:07

he's putting together. The small

1:10:09

handbook for sure. I'm gonna be

1:10:11

diving into that myself. So I

1:10:13

appreciate you taking the time today.

1:10:16

And honestly, you're spending all that

1:10:18

time putting this content together

1:10:20

because it's much needed. Thank you.

1:10:23

And thank you for having me

1:10:25

on the show. I enjoyed. Awesome.

1:10:27

Well, we'll have to have you

1:10:30

come on again when Come

1:10:32

All Three comes out and we

1:10:34

can get you to tear apart

1:10:37

all the great new features that

1:10:39

are to come, I'm sure. Perfect.

1:10:41

Well, that's it for day

1:10:43

folks. I'm Valentino. Thank you, I

1:10:46

wish. Thank you, Joseph. Until next

1:10:48

time. Hey,

1:10:53

it's Ryan Seacrest. Life comes at you

1:10:55

fast, which is why it's important to

1:10:57

find some time to relax. A little

1:11:00

you time. Enter Chumba Casino with no

1:11:02

download required. You can jump on any

1:11:04

time, anywhere for the chance to redeem

1:11:06

some serious prizes. So treat yourself with

1:11:09

Chumba Casino and play over a hundred

1:11:11

online casino style games. All for free.

1:11:13

Go to Chumba Casino.com to collect your

1:11:15

free welcome bonus. Sponsored by Chumba Casino.