* Download the [Composer](https://getcomposer.org/download/) PHP package manager inside the project folder.
* Run `./build.sh 1.0.0`, replacing the version with the version number you are going to build (the version number is only used for the generated dist file).
After that, you will have a `shlink_x.x.x_dist.zip` dist file inside the `build` directory, that you need to decompress in the location fo your choice.
> This is the process used when releasing new shlink versions. After tagging the new version with git, the Github release is automatically created by [travis](https://travis-ci.org/shlinkio/shlink), attaching the generated dist file to it.
* Setup the application by running the `bin/install` script. It is a command line tool that will guide you through the installation process. **Take into account that this tool has to be run directly on the server where you plan to host Shlink. Do not run it before uploading/moving it there.**
Once Shlink is configured, you need to expose it to the web, either by using a traditional web server + fast CGI approach, or by using a [swoole](https://www.swoole.co.uk/) non-blocking server.
For example, assuming your domain is doma.in and shlink is in the `/path/to/shlink` folder, these would be the basic configurations for Nginx and Apache.
Once installed, it's actually pretty easy to get shlink up and running with swoole. Run `./vendor/bin/mezzio-swoole start -d` and you will get shlink running on port 8080.
However, by doing it this way, you are loosing all the access logs, and the service won't be automatically run if the server has to be restarted.
For that reason, you should create a daemon script, in `/etc/init.d/shlink_swoole`, like this one, replacing `/path/to/shlink` by the path to your shlink installation:
if [[ -f "$PIDFILE" ]] && kill -0 $(cat "$PIDFILE"); then
echo 'Shlink with swoole already running' >&2
return 1
fi
echo 'Starting shlink with swoole' >&2
mkdir -p "$LOGDIR"
touch "$LOGFILE"
local CMD="$SCRIPT &> \"$LOGFILE\" & echo \$!"
su -c "$CMD" $RUNAS > "$PIDFILE"
echo 'Shlink started' >&2
}
stop() {
if [[ ! -f "$PIDFILE" ]] || ! kill -0 $(cat "$PIDFILE"); then
echo 'Shlink with swoole not running' >&2
return 1
fi
echo 'Stopping shlink with swoole' >&2
kill -15 $(cat "$PIDFILE") && rm -f "$PIDFILE"
echo 'Shlink stopped' >&2
}
case "$1" in
start)
start
;;
stop)
stop
;;
restart)
stop
start
;;
*)
echo "Usage: $0 {start|stop|restart}"
esac
```
Then run these commands to enable the service and start it:
*`sudo chmod +x /etc/init.d/shlink_swoole`
*`sudo update-rc.d shlink_swoole defaults`
*`sudo update-rc.d shlink_swoole enable`
*`/etc/init.d/shlink_swoole start`
Now again, you can access shlink on port 8080, but this time the service will be automatically run at system start-up, and all access logs will be written in `/var/log/shlink/shlink_swoole.log` (you will probably want to [rotate those logs](https://www.digitalocean.com/community/tutorials/how-to-manage-logfiles-with-logrotate-on-ubuntu-16-04). You can find an example logrotate config file [here](data/infra/examples/shlink-daemon-logrotate.conf)).
Finally access to [https://app.shlink.io](https://app.shlink.io) and configure your server to start creating short URLs.
Geo-locating visits to your short links is a time-consuming task. When serving Shlink with swoole, the geo-location task is automatically run asynchronously just after a visit to a short URL happens.
However, if you are not serving Shlink with swoole, you will have to schedule the geo-location task to be run regularly in the background (for example, using cron jobs):
The command you need to run is `/path/to/shlink/bin/cli visit:locate`, and you can optionally provide the `-q` flag to remove any output and avoid your cron logs to be polluted.
2. Download and extract the new version of Shlink, and set the directory name to that of the old version (ie. `shlink`).
3. Run the `bin/update` script in the new version's directory to migrate your configuration over. You will be asked to provide the path to the old instance (ie. `shlink-old`).
The `bin/update` will use the location from previous shlink version to import the configuration. It will then update the database and generate some assets shlink needs to work.
**Important!** It is recommended that you don't skip any version when using this process. The update tool gets better on every version, but older versions might make assumptions.
* **The REST API**. The complete docs on how to use the API can be found [here](https://shlink.io/api-docs), and a sandbox which also documents every endpoint can be found in the [API Spec](https://api-spec.shlink.io/) portal.
However, you probably don't want to consume the raw API yourself. That's why a nice [web client](https://github.com/shlinkio/shlink-web-client) is provided that can be directly used from [https://app.shlink.io](https://app.shlink.io), or you can host it yourself too.
While in many cases you will just have one short domain and you'll want all your short URLs to be served from it, there are some cases in which you might want to have multiple short domains served from the same Shlink instance.
If that's the case, you need to understand how Shlink will behave when managing your short URLs or any of them is visited.
### Management
When you create a short URL it is possible to optionally pass a `domain` param. If you don't pass it, the short URL will be created for the default domain (the one provided during Shlink's installation or in the `SHORT_DOMAIN_HOST` env var when using the docker image).
However, if you pass it, the short URL will be "linked" to that domain.
> Note that, if the default domain is passed, Shlink will ignore it and will behave as if no `domain` param was provided.
The main benefit of being able to pass the domain is that Shlink will allow the same custom slug to be used in multiple short URLs, as long as the domain is different (like `example.com/my-compaign`, `another.com/my-compaign` and `foo.com/my-compaign`).
Then, each short URL will be tracked separately and you will be able to define specific tags and metadata for each one of them.
However, this has a side effect. When you try to interact with an existing short URL (editing tags, editing meta, resolving it or deleting it), either from the REST API or the CLI tool, you will have to provide the domain appropriately.
Let's imagine this situation. Shlink's default domain is `example.com`, and you have the next short URLs:
*`https://example.com/abc123` -> a regular short URL where no domain was provided.
*`https://example.com/my-campaign` -> a regular short URL where no domain was provided, but it has a custom slug.
*`https://another.com/my-campaign` -> a short URL where the `another.com` domain was provided, and it has a custom slug.
*`https://another.com/def456` -> a short URL where the `another.com` domain was provided.
These are some of the results you will get when trying to interact with them, depending on the params you provide:
* Providing just the `abc123` short code -> the first URL will be matched.
* Providing just the `my-campaign` short code -> the second URL will be matched, since you did not specify a domain, therefor, Shlink looks for the one with the short code/slug `my-campaign` which is also linked to default domain (or not linked to any domain, to be more accurate).
* Providing the `my-campaign` short code and the `another.com` domain -> The third one will be matched.
* Providing just the `def456` short code -> Shlink will fail/not find any short URL, since there's none with the short code `def456` linked to default domain.
* Providing the `def456` short code and the `another.com` domain -> The fourth short URL will be matched.
* Providing any short code and the `foo.com` domain -> Again, no short URL will be found, as there's none linked to `foo.com` domain.
### Visits
Before adding support for multiple domains, you could point as many domains as you wanted to Shlink, and they would have always worked for existing short codes/slugs.
In order to keep backwards compatibility, Shlink's behavior when a short URL is visited is slightly different, getting to fallback in some cases.
Let's continue with previous example, and also consider we have three domains that will resolve to our Shlink instance, which are `example.com`, `another.com` and `foo.com`.
With that in mind, this is how Shlink will behave when the next short URLs are visited:
*`https://another.com/abc123` -> There was no short URL specifically defined for domain `another.com` and short code `abc123`, but it exists for default domain (`example.com`), so it will fall back to it and redirect to where `example.com/abc123` is configured to redirect.
*`https://example.com/def456` -> The fall-back does not happen from default domain to specific ones, only the other way around (like in previous case). Because of that, this one will result in a not-found URL, even though the `def456` short code exists for `another.com` domain.
*`https://foo.com/abc123` -> This will also fall-back to `example.com/abc123`, like in the first case.
*`https://another.com/non-existing` -> The combination of `another.com` domain with the `non-existing` slug does not exist, so Shlink will try to fall-back to the same but for default domain (`example.com`). However, since that combination does not exist either, it will result in a not-found URL.
* Any other short URL visited exactly as it was configured will, of course, resolve as expected.
### Special redirects
It is currently possible to configure some special redirects when the base domain is visited, a URL does not match, or an invalid/disabled short URL is visited.
Those are configured during Shlink's installation or via env vars when using the docker image.
Currently those are all shared for all domains serving the same Shlink instance, but the plan is to update that and allow specific ones for every existing domain.