How to deploy your Jesse project into the production server

In this tutorial, I will show you step by step, how to deploy your Jesse project from your local into a VPS for production.

I highly recommend using a VPS for production instead of running the bot on your local machine. The reason being that your local machine is not as reliable as a server. It has a better Internet connection, and there’s less chance of accidents such as power outages.

Installation and configuration of a Linux VPS seem scary to some users. But don’t worry, because I have made a script that makes the installation super easy.

Recommended cloud provider

Here are a list of providers that I recommend based on my experience:

I’ll be using an Ubuntu 18.04 VPS which is offered by almost all cloud providers.

I’ve been using Hetzner for past years because not only it has the cheapest price, but also it has the highest quality among providers that I tried. Its starting plan which comes with 1 core CPU and 2GBs of RAM is what I’ve been using for running Jesse and it costs only €2.96 per month. You can use my referral link which gives you €20 free credits at the time of sign up. That is months of the server for free!

Creating a cloud VPS

After logging into your Hetzner account, choose the default project, or create a new one. Then click on the red “Add server” button. You can choose the location of the server as whatever you want it to be.

As for the “Image”, I’ll be choosing Ubuntu 18.04:

Leave the rest be. The beginning €2.96 is more than enough.

Next, you need to add/choose an SSH key and give your server a name. The name could be anything you want it to be. As for the SSH key, if you have one, just copy it from your local system and add it. If you haven’t created one already, go you Gitlab’s guide and follow the instruction to create one. It’ll only take a minute. And then, add it into the SSH key section:

Now go ahead and click on “CREATE & BUY NOW”.

Installing the server

Connect to the server we just created on Hezner, by opening a terminal and typing:

# "servers_ip" should be the IP address of your server
ssh root@servers_ip

Now execute below command which is from my “ stack installer repository:

source <(curl -fsSL https://raw.githubusercontent.com/jesse-ai/stack-installer/master/ubuntu-18.04.sh)

It’ll take a few minutes for the installation to go through. After that, your server has all that it needs to run Jesse.

Creating a Jesse project

Let’s create a new Jesse project on our local machine. I’m on a macOS machine. After going into your desired location, run:

# bot1 is the name of my project. It could be anything you want jesse make-project bot1

And next I cd into it:

cd bot1

Creating a strategy

Let’s copy a strategy called “SMACrossover” from the “[example strategies](https://github.com/jesse-ai/example-strategies" repository which contains a few strategies submitted by the community members.

Next, I create a new strategy and name it SMACrossover by executing:

jesse make-strategy SMACrossover

Next, I paste the contents of SMACrossover into strategies/SMACrossover/__init__.py in my project.

Getting the live trade plugin

If you are reading this, I assume you already have a valid license for the live trade plugin. If you don’t, please head over to Jesse’s website, register, and purchase a license.

Next, go to the “ releases” page and download the latest version for (in this case) Linux - x86_64:

Now move the download package into your project files.

Deploying everything to the server

I’ll be using Gitlab to host my project privately, and then of course to retrieve it from my server.

If you’re not familiar with Git, I highly recommend checking out this YouTube video to get started with it. Git is a very useful version manager and services such as Gitlab or Github are real-life savers for development.

First, I create a new private project on Gitlab and give it a name. In my case, I entered bot1. Make sure that the "Visibility Level" is set to Private.

Now I need to push my project files into the repository. I do that by opening the terminal, and going into the project, and running:

cd bot1
git init
git remote add origin git@gitlab.com:saleh-mir/bot1.git
git add .
git commit -m "Initial commit"
git push -u origin master

Now I need to retrieve the files from the server. The first thing I need to do is to give my VPS access to this private repository. For that, I need to create an SSH key, and add it to my Gitlab account’s SSH:

ssh-keygen

It’ll ask for the location to store my keys, and I just choose the default one by pressing Enter. Next, it'll ask for a passphrase, which I skip by pressing Enter. And of course, one more time for confirmation.

After that, my keys are generated and I can see it by executing:

cat /root/.ssh/id_rsa.pub

I now copy my SSH key and paste it into my Gitlab accounts SSH keys page:

And finally, click on “Add key”.

Now I open the terminal tab that has the VPS open in it, and clone this repository by typing:

git clone [email protected]:saleh-mir/bot1.git

Type yes for confirmation. Now open the project by typing cd bot1. Notice that our project is missing the config.py and routes.py files. That's because they are added into the .gitignore files of projects created by Jesse. This is to make sure that your sensitive API keys and such, are not accidentally pushed to the repository.

So I create them back by just copying the content of each from my local, and paste it in the server:

# after pasting, press CTRL + C, and then Y to confirm and save the file. 
nano config.py

Now I do the same for routes.py:

# after pasting, press CTRL + C, and then Y to confirm and save the file. 
nano routes.py

Here is the content of routes.py that I copied. You can modify it as you need:

# trading routes
routes = [
    # notice that the exchange should be either "Binance Futures" or "Testnet Binance Futures" and NOT "Binance"
    ('Testnet Binance Futures', 'BTC-USDT', '1m', 'SMACrossover'),
]

# in case your strategy requires extra candles, timeframes, ...
extra_candles = [
    # ('Binance', 'BTC-USDT', anchor_timeframe('4h')),
]

Creating a database and user

Of course, we also need a database. PostgreSQL has already been installed using the script we just used. So let’s create one by below commands:

# switch to postgres user
sudo su - postgres
# open PostgreSQL CLI
psql
# create database
CREATE DATABASE jesse_db;
# create new user
CREATE USER jesse_user WITH PASSWORD 'password';
# set privileges of the created user
GRANT ALL PRIVILEGES ON DATABASE jesse_db to jesse_user;
# exit PostgreSQL CLI
\q
# exit postgres user (back to root)
exit

Installing the live trade plugin

Since I am already inside my project which has the live trade plugin’s package in it, I can easily install it with pip:

pip install jesse_live-0.0.2-cp38-cp38-manylinux2010_x86_64.whl

Now I should see the successful installation message:

Successfully built starkbank-ecdsa
Installing collected packages: starkbank-ecdsa, jesse-live
Successfully installed jesse-live-0.0.2 starkbank-ecdsa-1.1.0
WARNING: Running pip as root will break packages and permissions. You should install packages reliably by using venv: https://pip.pypa.io/warnings/venv

Now I need to log in by typing:

jesse login

This will ask for my email and password which I entered at the time of registration on Jesse’s website. If it goes well you should see:

Authenticated successfully.

Now let’s confirm that the live trade plugin is installed correctly by typing:

The output should have the live and paper modes in it:

➜  bot1 git:(master) ✗ jesse
Usage: jesse [OPTIONS] COMMAND [ARGS]...

Options:
  --version  Show the version and exit.
  --help     Show this message and exit.

Commands:
  backtest        backtest mode.
  import-candles  imports historical candles from exchange
  live            trades in real-time on exchange with REAL money
  login           (Initially) Logins to the website.
  make-project    generates a new strategy folder from...
  make-strategy   generates a new strategy folder from...
  optimize        tunes the hyper-parameters of your strategy
  paper           trades in real-time on exchange with PAPER money
  routes          lists all routes

Exchange keys, notifications, and dashboard settings

In order for Jesse to be able to connect to the server, I have to enter my API keys, and for it to be able to send me notifications I have to enter my keys for either Telegram or Discord (or both!).

I can do that by modifying the live-config.py file which contains this info:

nano live-config.py

Starting and monitoring the running bot

Now everything is in place. I can start a live session by typing:

Or a paper trading session by:

jesse paper

It is often useful to have the debug mode enabled. For that, enter the same commands but this time add the debug flag at the end.

Keeping the session alive after closing the terminal

But there’s a problem here. If I close my terminal, the live session will stop. But I want the bot to continue trading for days even if I shut down my computer. That was the whole point of running it on a VPS.

For that, I use a tool called “screen”. Screen creates a virtual terminal screen that I can attach and detach from whenever I need to. screen gets installed by that installer script that I just used. So I can go ahead and create my first screen:

screen -S bot

Now inside this screen I can run the live session again, but this time it won’t be terminated if I close my terminal. Next time that I connect back to the server, I can reattach to that screen by:

screen -r bot

Originally published at https://jesse.trade.