Deploying Dogra to a Remote Server 🚀

Hi, my name is Sabiha and I'm a software engineer at Dogra.

Dogra is an open source voice agents platform which is very quick to get started with this tutorial is specifically to deploy Dogra at a remote machine without a custom domain.

So the relevant documentation for this can be found@docs.dogra.com you need to go to the developer tab and then select Docker here.

the option two is for the remote server deployment.

So you need to just follow these steps.

The prerequisite for this is that you should have the public IP address of the server where we are trying to deploy Dogra and you should have Docker installed on that server.

So I have already done that.

I have Docker deployed on the server where I'm trying to make the deployment and I'm connected to that server over this terminal right now.

So you can see Docker is available.

So let's go to the tutorial.

So we're done with the prerequisites.

I have also made sure that the TCP and UDP ports that are specified here are available.

I'll just check the next step which is to copy this command and run it at the terminal.

a small section says here that the script is going to prompt us for a few pieces of information and let's see what happens.

I'll just paste the copied command here and I'll run it.

So what it asks me is the server's public IP address.

I had copied it in some text file.

I have the IP address with me.

The next thing is it asks is shared secret for the turn server which is optional.

So I just hit enter and it's doing a couple of things here.

the summary is that the setup is complete and I just want to see what all got downloaded here.

So you see there's a script and there is a directory.

Let's go to the directory.

Okay, so the directory is empty and I'm sorry, so the directory does have some files.

So before we jump into the next steps, I just want to quickly go through the contents of the script that we had just downloaded with this, with this command here.

So okay, I need to come out of it.

Yeah.

So what the script contains is like this is the mammoth which is taking care of the entire deployment.

it is setting up all the all the configuration that we need for this deployment.

So I'm going to cover this at a very high level because I basically wanted to touch upon how the How there are multiple steps which are a part of this script and every step has to succeed for this deployment to work.

so the steps are basically the downloading of Docker Compose which we just saw in the Dogra directory.

I'll open that again.

The next thing is that we are creating an nginx, configuration.

The reason for that is most of the modern browsers today allow microphone permissions to be managed over a website which is served with HTTPs.

So that's what we need to do and which is why we're using nginx.

So nginx it will basically redirect all the HTTP URLs to the HTTPs ones and it will take care of the SSL termination and also make use of some of the certificates that we are generating in the later steps.

we are also manipulating some minio audio file URLs that the application uses so that we use the public IP instead of the local host which you see here.

So the next steps are So there are some helper scripts to generate the SSL certificates.

These are self signed certificates which are tied to this specific server to this IP address.

after that we are creating a turn server configuration which is for WebRTC, nat traversal so so which is basically the relaying of audio data over a WebRTC connection across the browser that we'll be using and the DOG server that we are setting up.

So it also makes use of the IP that we had previously entered and then again we have some environment variables that the Docker compose injects during runtime.

what includes is the API endpoint which has our IP address, the public IP address and we also have the turn server credentials and a JWT secret.

So this is used to sign the token that will be used in JWT authentication which is a primary mode of authentication for this OSS deployment that we are doing.

So in case you are having any challenges with the deployment you can reach out to us but it will be really helpful for both you and us to troubleshoot if we know, know what if what step is exactly is failing and if a certain step is actually resulting into an expected outcome.

So all of that you can see over here in on the terminal.

So what we see here right now is that the setup is complete and some files were created and these are those files.

So and there are also some instructions that in order to start Dogra you need to run this command.

Okay.

And we will run this command where we have the docker compose file so which is inside this directory.

So if you take a look we have all the configurations that we saw previously in the script and we also have a directory called certs and there should be the surf signed certificates that got generated.

Yeah so yeah so I'll run the docker compose command here.

And once all the containers are up on the server I'll be launching the front end application on my browser on the local machine using the IP address.

Alright so some of the containers are up.

So we'll wait for both ap, AI and UI to be up and running.

So yeah a point to note that for storage we use by default we're using minio.

there is you can enable AWS S3 if that's your choice of storage.

okay we can ignore this.

So the application startup for API is complete and we have the local our front end application ready as well.

So I'll just launch this the website in an incognito tab and because these are self signed certificates so we will see this this warning Okay we'll sign up first.

I will use my email.

Here.

We have the voice agent builder before us so it's a it, it needs very minimal information to spin up a voice agent as needed.

So I'll I'm going to use a very easy example so screens interview candidates but yeah you should be using you should be making the description as detailed as possible covering everything that you would want your voice agent to do.

So so the voice agent has been built by the system.

The next step is to test it out so we should be able to grant the audio permission.

Okay and we'll start the call.

Hi sam, this is sabiha.

Yes I do have a bachelor's degree in computer science.

So that was the first call that we made for this voice agent.

Everything is all right, all the features are working as expected.

we can see the summary of the call on this page.

It includes the live transcript and options to download or preview the transcript and recording and the context that was collected throughout the call is summarized here so you can check out the documentation to know all the features of the platform and everything is in great detail over here.

There are different ways to deploy the application.

what we covered just now was how do we deploy Dogra on a remote server without a custom domain.

So this is our GitHub, repo, the name is Dogra HQ Dogra.

the code base is entirely open sourced so that everyone can access it and you can customize the implementation, build stuff on the top of this.

we have a very active Slack community so if you are facing any challenges, if you need any kind of support, you can join that and we'll try our best.

If you face any issues with the deployment, please do tag me.

reach out at Slack and if you find the project meaningful in some way or another just do see if you can star us that helps us out.

the cloud hosted version of the platform is@app.dogra.com and so yeah that's about it.

Thank you.