IPFS and Docker

Let’s take the example of Dropbox. I could share with you a file with some content as http://somedropboxlink.com/myfile.pdf. After sharing this file with you, I am free to modify the content. The link will remain valid but what you get when downloading the file will depend on WHEN you download it. If you consider some sort of contractual agreement, this is not a nice property.

Using IPFS, things are different. When sharing a file over IPFS, the content you get from a given link is guaranteed to be always exactly the same. I no longer can sneak changes into the contract after sharing it with you :)

You do not need to install anything to view and retrieve files from the IPFS network. This is possible thanks to the IPFS Gateway hosted at https://ipfs.io.

Opening this link in your Browser will show exactly the following content:

CONGRATULATIONS! You just used IPFS!

When I say exactly, this time it will NOT depend on when you open the URL. In 10 years from now, when opening this IPFS link, the content will be either no longer available, or if it is available, exactly the same.

If someone would modify the content, even a very small change such as adding a space, it would modify the hash you see at the end of the URL. So the original content remains available at the original URL (which is mainly the IPFS hash) and the new content become available at a NEW url (= new IPFS Hash).

You may wonder why the data would then no longer be available?

IPFS is a peer to peer network. When you request a document, your peer (or the IPFS gateway when using the link above) is asking peers around whether someone has the content. If yes, your peer will fetch it, keep it for a while, and returned it. if you need this content again, your peer will already have it (for a while, depending on the configuration of your node).

If no peer has the content matching the hash, no data can be returned. It is up to the peers to decide what is 'interesting content' and how long to keep it.

There is a way to tell your IPFS node to PIN data. It means that we tell our node that some piece of data is important and we want to keep it forever so that it remains available through the network.

The number of Docker related articles on my blog is rapidly growing because Docker makes it very straight forward to run any service, especially if you want to test things out.

Installing IPFS on a new server does require a few steps as you can see here: https://ipfs.io/ipfs/QmR7tiySn6vFHcEjBeZNtYGAFh735PJHfEMdVEycj9jAPy/docs/install/

In short, you will need to install some GOLANG stuff, tweak a bit your system, tell the IPFS daemon to start as service, take care of the access rights, etc…​ I am not sure about you but I seldom have the time or the patience to deal with all of that…​

Installing Docker on a server also requires a few steps but once this is done, Docker really makes it simpler.

Here is how you can run an IPFS node on your server and access the ipfs command running inside your Docker container.

First, you will need to connect to your server using ssh.

We will now create 2 folders. I know you hate copy/pasting so here it is as a single line:

If you have the tree command on your system, it will show:

We are now ready for the docker part. While the line is a little longer, it is also rather straight forward and you should have problem with that as long as you installed Docker properly:

That’s it! We are done :) IPFS is now running.

But how can we use it? I could give you a few docker commands to do that here but that is rather longer and boring. Let’s make an alias that will make things much easier. I am naming my alias dipfs (for docker-ipfs) since I have an old ipfs command on my system. If you don’t, you may replace dipfs by ipfs in the commands below:

In order for this alias to be persistent, you need to add it to your .bashrc. Here is another one-liner:

We now can use our new command. Let’s check out the help command first:

Now let’s fetch some content:

Now let’s add one of our files to IPFS. For the example, we will generate a file. Since IPFS is effectively de-duplicating content, adding a "Hellow World" would be rather useless since someone likely did it already. So you would not add NEW content to IPFS. Instead, we will use the current date as content, that should be fair enough:

You will likely add files and content using the IPFS Desktop utility we saw at the beginning of this article. Now lets see how we can pin the content we added to our new server so it keeps on providing those files.

Let’s take the IPFS hash QmYwAPJzv5CZsnA625s3Xf2nemtYgPpHdWEz79ojWnPbdG as example. It may be data from a friend or colleague and you want to help making this content available.

You may use the IPFS Desktop utility to see the content: