== Analytics for Cassandra Web

This document serves as a guide for anybody who wants to re/deploy https://plausible.io/[Plausible].

=== Overview of the current deployment

Plausible is currently deployed on t2.medium instance in AWS EC2, it runs on Debian 10 Buster and
it has Elastic IP 3.124.255.5 (stays after restarts). Plausible runs in Docker Compose and it is
secured by HTTPS. Certificates for HTTPS are obtained from Let's Encrypt and they are
automatically renewed after expiration. There is nginx reverse proxy in front of Plausible as
the other Docker Compose deployment.

Plausible service is hosted under https://www.instaclustr.com[Instaclustr]'s account. Please contact
stefan dot miklosovic at instaclustr dot com for any related questions / issues.

Domain `plausible.cassandra.apache.org` points to 3.124.255.5. Please contact gmcdonald at apache dot org for
any related questions / issues.

For the access to all credentials (PEM to ssh to that box, logins to Plausible etc), please contact
someone on the PMC, as credentials are stored in the private subversion repository.

=== Setup of Plausible from scratch

Please be sure that before proceeding, your box has these services / software installed:

* docker
* docker-compose
* git

The deployment consists of these logical steps:

* setup HTTPS for nginx proxy
* setup & run self-hosted Plausible
* setup & run reverse nginx proxy talking to Plausible

==== Setup of HTTP for nginx proxy

First we generate HTTPS certificates:

----
$ git clone https://github.com/wmnnd/nginx-certbot.git
$ cd nginx-certbox
----

Follow the https://github.com/wmnnd/nginx-certbot[readme] of nginx-certbox repo. You need to change
`domains` in `init-letsencrypt.sh` script to `domains=(plausible.cassandra.apache.org)`.

Next, modify `data/nginx/app.conf`, change `server_name` to `plausible.cassandra.apache.org` for
both 80 and 443 case and change paths to certificates.

`proxy_pass` field of `location /` for server listening to port 443 should be set to `http://plausible:8000/;`.
`plausible` in this address points to name of a service of Docker compose we will setup afterwards.

This will generate HTTPS certificates, please be sure that you have 80 and 443 port open.

----
$ ./init-letsencrypt.sh
$ docker-compose up -d
----

`data/nginx/app.conf` file with described changes is also in the same directory as this document for reference.

==== Setup of Plausible

----
$ git clone https://github.com/plausible/hosting
$ cd hosting
----

The main configuration file is `plausible-conf.env`

----
ADMIN_USER_EMAIL=stefan.miklosovic@instaclustr.com
ADMIN_USER_NAME=admin
ADMIN_USER_PWD=you will log in with this password under ADMIN_USER_EMAIL
BASE_URL=https://plausible.cassandra.apache.org/
SECRET_KEY_BASE=see for the explanation below
PORT=8000
DISABLE_AUTH=false
DISABLE_REGISTRATION=false
GEOLITE2_COUNTRY_DB=/geoip/GeoLite2-Country.mmdb
----

`SECRET_KEY_BASE` is an internal secret key used by Phoenix Framework which Plausible builds on.
Follow the https://hexdocs.pm/phoenix/Mix.Tasks.Phx.Gen.Secret.html#content[instructions] to generate one.
The setup to generate such a secret is very long as it requires you to setup Erlang OTP, Phoenix etc etc and it seems
to me that the value of `SECRET_KEY_BASE` needs to be some random string so https://github.com/plausible/analytics/discussions/824[anything sensible will do].

For GeoIP (seeing a world map with hits from countries), you need to https://dev.maxmind.com/geoip/geoip2/geolite2/[create an account here].
Then provide credentials in `geopip/geoip.conf`.

`DISABLE_REGISTRATION` might be set to `true` after all accounts are created as we will prevent everybody to
try to create accounts and potentially abuse this service.

You do not need to do anything for SMTP to work, all default values are just fine in order to be able to
send transactional emails. Please consult SMTP setup in FAQ section to know more if you happen to run on AWS.

For more details, follow the https://plausible.io/docs/self-hosting-configuration[official docs].

For simplicity, you will find unified Docker compose file with Plausible as well as with GeoLite
integration in `docker-compose.yaml` in this directory.

Do `docker-compose up -d` to start and move it to background.

=== FAQ

==== When I do docker-compose down, does it remove data of Plausible?

No, unless you delete them by `-v` flag. This means that normal down / up will keep data / statistics
from the last run. Some browsers might cache older stats, the rule of thumb is to open analytics dashboard
in a private tab.

==== How to lift limits on SMTP on AWS?

By default, AWS blocks any SMTP traffic outside (e.g. to prevent spamming) so Plausible's mail c
container will not be able to send anything. You need to send an email to AWS support to lift these
restrictions. Submit https://aws.amazon.com/forms/ec2-email-limit-rdns-request[this form] to resolve this.

You may check if your hosting is preventing the box to e.g. contact SMTP servers like this:

----
$ sudo traceroute -n -T -p 25 gmail-smtp-in.l.google.com
----

==== How two verify email in Plausible when e-mails do not work?

----
$ docker exec -ti hosting_plausible_db_1 psql -U postgres plausible_db
----

After getting into Postgres shell:

----
UPDATE users set email_verified=true;
----

==== How much disk space I should specify upon box provisioning?

40G is more you will ever need. 20G is fine too.

==== How do I restart everything?

----
$ cd ~/hosting
$ docker-compose down && docker compose up -d
$ cd ~/ngxin-certbot
$ docker-compose down && docker-compose up -d
----

It might take some time to start Plausible especially when it is for the first time as DB has to be
created etc.