# 🏗 Install on your server

## Use Twake in SaaS

You can test or use Twake in our SaaS : [chat.twake.app](https://chat.twake.app)

## Run Twake in your server

1. First you'll need to [install docker and docker-compose](https://docs.docker.com/compose/install/).

2. Then you can install Twake on your server with this command

```
git clone https://github.com/TwakeApp/Twake.git
cd Twake/twake
docker-compose -f docker-compose.onpremise.mongo.yml up -d
```

Twake will be running on port 3000

### What's next ?

If you kept the default configuration, you can simply follow the signup steps, no email verification is required by default so you will get into Twake right after the signup steps.

## Ship Twake in production

See how to [manage configuration](./configuration/index.md)). And then how to [update security keys](./configuration/security.md), and finally how to use your [custom domain](./configuration/custom-domain-and-https/README.md)).

### Update Twake

```
docker-compose stop
docker-compose rm #Remove images (not volumes so your data is safe)
docker-compose pull #Get new images
docker-compose up -d
docker-compose exec nginx yarn build #If you have custom frontend configuration
```

## Requirements and scalability

Currently you'll need at least a **2 cpu + 4 gb of ram** machine for **20-50 users** depending on their usage and with ElasticSearch disabled.

If you enable ElasticSearch, use two machines, or limit the cpu/ram dedicated to it and use a larger machine (at least 2gb of ram and 1 cpu dedicated to ES for 20-50 users).

If you need to deploy Twake for more users, you can use only one big machine up to 500 users (Will need something like **12 cpu and 32go of ram**), then you'll need to use multiple nodes.

---

# 🎡 Scale with Twake

> We deployed Twake on production for companies of 10 to 50 users in a single node. We also deployed Twake in a scalable mode and we support currently thousands of concurrent users.
> If you deploy Twake in your own company we would love to have your feedback here [https://github.com/TwakeApp/Twake/issues/289](https://github.com/TwakeApp/Twake/issues/289) to improve our requirements documentation.
> Now if you want to scale with Twake and support thousand of users, continue reading.

Scaling with Twake is possible if you install Twake with **RabbitMQ, Redis, ElasticSearch and ScyllaDB**.

```
git clone https://github.com/TwakeApp/Twake.git
cd Twake/twake

cp -n docker-compose.yml.dist.onpremise docker-compose.yml
cp -nR default-configuration/ configuration/

docker-compose pull

docker-compose up -d scylladb
sleep 5m #Wait scylladb to startup
docker-compose up -d php rabbitmq
sleep 10m #Wait php to create tables in scylladb

docker-compose up -d
```

> To run ElasticSearch (optional, but enabled by default in the Twake docker-compose) you must increase the max_map_count of your system: [https://www.elastic.co/guide/en/elasticsearch/reference/current/docker.html#\_set_vm_max_map_count_to_at_least_262144](https://www.elastic.co/guide/en/elasticsearch/reference/current/docker.html#_set_vm_max_map_count_to_at_least_262144)
>
> To fix an other bug with ElasticSearch container, you must also run this command: `chmod 777 ./docker-data/es_twake` (create the folder if it doesn't exists in your docker-compose.yml folder)

Twake will be running on port 8000 🎉