Merge branch 'release-docs' into 'develop'

Docs on OTP releases

See merge request pleroma/pleroma!1315
This commit is contained in:
lain 2019-06-28 15:26:09 +00:00
commit 2286f62a36
9 changed files with 494 additions and 23 deletions

View File

@ -15,9 +15,12 @@ For clients it supports both the [GNU Social API with Qvitter extensions](https:
If you want to run your own server, feel free to contact us at @lain@pleroma.soykaf.com or in our dev chat at #pleroma on freenode or via matrix at <https://matrix.heldscal.la/#/room/#freenode_#pleroma:matrix.org>.
## Installation
**Note:** The guide below may be outdated and in most cases shouldn't be used. Instead check out our [wiki](https://docs.pleroma.social) for platform-specific installation instructions, most likely [Installing on Linux using OTP releases](https://docs.pleroma.social/otp_en.html) is the guide you need.
### OS/Distro packages
Currently Pleroma is not packaged by any OS/Distros, but feel free to reach out to us at [#pleroma-dev on freenode](https://webchat.freenode.net/?channels=%23pleroma-dev) or via matrix at <https://matrix.heldscal.la/#/room/#freenode_#pleroma-dev:matrix.org> for assistance. If you want to change default options in your Pleroma package, please **discuss it with us first**.
### Docker
While we dont provide docker files, other people have written very good ones. Take a look at <https://github.com/angristan/docker-pleroma> or <https://github.com/sn0w/pleroma-docker>.
### Dependencies

View File

@ -0,0 +1,151 @@
# Switching a from-source install to OTP releases
## What are OTP releases?
OTP releases are as close as you can get to binary releases with Erlang/Elixir. The release is self-contained, and provides everything needed to boot it, it is easily administered via the provided shell script to open up a remote console, start/stop/restart the release, start in the background, send remote commands, and more.
### Can I still run the develop branch if I decide to use them?
Yes, we produce builds for every commit in `develop`. However `develop` is considered unstable, please don't use it in production because of faster access to new features, unless you need them as an app developer.
## Why would one want to switch?
Benefits of OTP releases over from-source installs include:
* **Less space used.** OTP releases come without source code, build tools, have docs and debug symbols stripped from the compiled bytecode and do not cointain tests, docs, revision history.
* **Minimal system dependencies.** Excluding the database and reverse proxy, only `curl`, `unzip` and `ncurses` are needed to download and run the release. Because Erlang runtime and Elixir are shipped with Pleroma, one can use the latest BEAM optimizations and Pleroma features, without having to worry about outdated system repos or a missing `erlang-*` package.
* **Potentially less bugs and better performance.** This extends on the previous point, because we have control over exactly what gets shipped, we can tweak the VM arguments and forget about weird bugs due to Erlang/Elixir version mismatches.
* **Faster and less bug-prone mix tasks.** On a from-source install one has to wait untill a new Pleroma node is started for each mix task and they execute outside of the instance context (for example if a user was deleted via a mix task, the instance will have no knowledge of that and continue to display status count and follows before the cache expires). Mix tasks in OTP releases are executed by calling into a running instance via RPC, which solves both of these problems.
### Sounds great, how do I switch?
Currently we support Linux machines with GNU (e.g. Debian, Ubuntu) or musl (e.g. Alpine) libc and `x86_64`, `aarch64` or `armv7l` CPUs. If you are unsure, check the [Detecting flavour](otp_en.html#detecting-flavour) section in OTP install guide. If your platform is supported, proceed with the guide, if not check the [My platform is not supported](#my-platform-is-not-supported) section.
### I don't think it is worth the effort, can I stay on a from-source install?
Yes, currently there are no plans to deprecate them.
### My platform is not supported
If you think your platform is a popular choice for running Pleroma instances, or has the potential to become one, you can [file an issue on our Gitlab](https://git.pleroma.social/pleroma/pleroma/issues/new). If not, guides on how to build and update releases by yourself will be available soon.
## Pre-requisites
You will be running commands as root. If you aren't root already, please elevate your priviledges by executing `sudo su`/`su`.
The system needs to have `curl` and `unzip` installed for downloading and unpacking release builds.
Debian/Ubuntu:
```sh
apt install curl unzip
```
Alpine:
```
apk add curl unzip
```
## Moving content out of the application directory
When using OTP releases the application directory changes with every version so it would be a bother to keep content there (and also dangerous unless `--no-rm` option is used when updating). Fortunately almost all paths in Pleroma are configurable, so it is possible to move them out of there.
Pleroma should be stopped before proceeding.
### Moving uploads/custom public files directory
```sh
# Create uploads directory and set proper permissions (skip if using a remote uploader)
# Note: It does not have to be `/var/lib/pleroma/uploads`, you can configure it to be something else later
mkdir -p /var/lib/pleroma/uploads
chown -R pleroma /var/lib/pleroma
# Create custom public files directory
# Note: It does not have to be `/var/lib/pleroma/static`, you can configure it to be something else later
mkdir -p /var/lib/pleroma/static
chown -R pleroma /var/lib/pleroma
# If you use the local uploader with default settings your uploads should be located in `~pleroma/uploads`
mv ~pleroma/uploads /var/lib/pleroma/uploads
# If you have created the custom public files directory with default settings it should be located in `~pleroma/instance/static`
mv ~pleroma/instance/static /var/lib/pleroma/static
```
### Moving emoji
Assuming you have all emojis in subdirectories of `priv/static/emoji` moving them can be done with
```sh
mkdir /var/lib/pleroma/static/emoji
ls -d ~pleroma/priv/static/emoji/*/ | xargs -i sh -c 'mv "{}" "/var/lib/pleroma/static/emoji/$(basename {})"'
```
But, if for some reason you have custom emojis in the root directory you should copy the whole directory instead.
```sh
mv ~pleroma/priv/static/emoji /var/lib/pleroma/static/emoji
```
and then copy custom emojis to `/var/lib/pleroma/static/emoji/custom`.
This is needed because storing custom emojis in the root directory is deprecated, but if you just move them to `/var/lib/pleroma/static/emoji/custom` it will break emoji urls on old posts.
Note that globs have been replaced with `pack_extensions`, so if your emojis are not in png/gif you should [modify the default value](config.html#emoji).
### Moving the config
```sh
# Create the config directory
# The default path for Pleroma config is /etc/pleroma/config.exs
# but it can be set via PLEROMA_CONFIG_PATH environment variable
mkdir -p /etc/pleroma
# Move the config file
mv ~pleroma/config/prod.secret.exs /etc/pleroma/config.exs
# Change `use Mix.Config` at the top to `import Config`
$EDITOR /etc/pleroma/config.exs
```
## Installing the release
Before proceeding, get the flavour from [Detecting flavour](otp_en.html#detecting-flavour) section in OTP installation guide.
```sh
# Delete all files in pleroma user's directory
rm -r ~pleroma/*
# Set the flavour environment variable to the string you got in Detecting flavour section.
# For example if the flavour is `arm64-musl` the command will be
export FLAVOUR="arm64-musl"
# Clone the release build into a temporary directory and unpack it
# Replace `master` with `develop` if you want to run the develop branch
su pleroma -s $SHELL -lc "
curl 'https://git.pleroma.social/api/v4/projects/2/jobs/artifacts/master/download?job=$FLAVOUR' -o /tmp/pleroma.zip
unzip /tmp/pleroma.zip -d /tmp/
"
# Move the release to the home directory and delete temporary files
su pleroma -s $SHELL -lc "
mv /tmp/release/* ~pleroma/
rmdir /tmp/release
rm /tmp/pleroma.zip
"
# Start the instance to verify that everything is working as expected
su pleroma -s $SHELL -lc "./bin/pleroma daemon"
# Wait for about 20 seconds and query the instance endpoint, if it shows your uri, name and email correctly, you are configured correctly
sleep 20 && curl http://localhost:4000/api/v1/instance
# Stop the instance
su pleroma -s $SHELL -lc "./bin/pleroma stop"
```
## Setting up a system service
OTP releases have different service files than from-source installs so they need to be copied over again.
Debian/Ubuntu:
```sh
# Copy the service into a proper directory
cp ~pleroma/installation/pleroma.service /etc/systemd/system/pleroma.service
# Reload service files
systemctl reload-daemon
# Reenable pleroma to start on boot
systemctl reenable pleroma
# Start pleroma
systemctl start pleroma
```
Alpine:
```sh
# Copy the service into a proper directory
cp -f ~pleroma/installation/init.d/pleroma /etc/init.d/pleroma
# Start pleroma
rc-service pleroma start
```
## Running mix tasks
Refer to [Running mix tasks](otp_en.html#running-mix-tasks) section from OTP release installation guide.
## Updating
Refer to [Updating](otp_en.html#updating) section from OTP release installation guide.

261
docs/installation/otp_en.md Normal file
View File

@ -0,0 +1,261 @@
# Installing on Linux using OTP releases
## Pre-requisites
* A machine running Linux with GNU (e.g. Debian, Ubuntu) or musl (e.g. Alpine) libc and `x86_64`, `aarch64` or `armv7l` CPU, you have root access to. If you are not sure if it's compatible see [Detecting flavour section](#detecting-flavour) below
* A (sub)domain pointed to the machine
You will be running commands as root. If you aren't root already, please elevate your priviledges by executing `sudo su`/`su`.
While in theory OTP releases are possbile to install on any compatible machine, for the sake of simplicity this guide focuses only on Debian/Ubuntu/Alpine.
### Detecting flavour
Paste the following into the shell:
```sh
arch="$(uname -m)";if [ "$arch" = "x86_64" ];then arch="amd64";elif [ "$arch" = "armv7l" ];then arch="arm";elif [ "$arch" = "aarch64" ];then arch="arm64";else echo "Unsupported arch: $arch">&2;fi;if getconf GNU_LIBC_VERSION>/dev/null;then libc_postfix="";elif [ "$(ldd 2>&1|head -c 9)" = "musl libc" ];then libc_postfix="-musl";elif [ "$(find /lib/libc.musl*|wc -l)" ];then libc_postfix="-musl";else echo "Unsupported libc">&2;fi;echo "$arch$libc_postfix"
```
If your platform is supported the output will contain the flavour string, you will need it later. If not, this just means that we don't build releases for your platform, you can still try installing from source.
### Installing the required packages
Other than things bundled in the OTP release Pleroma depends on:
* curl (to download the release build)
* unzip (needed to unpack release builds)
* ncurses (ERTS won't run without it)
* PostgreSQL (also utilizes extensions in postgresql-contrib)
* nginx (could be swapped with another reverse proxy but this guide covers only it)
* certbot (for Let's Encrypt certificates, could be swapped with another ACME client, but this guide covers only it)
Debian/Ubuntu:
```sh
apt install curl unzip libncurses5 postgresql postgresql-contrib nginx certbot
```
Alpine:
```sh
echo "http://nl.alpinelinux.org/alpine/latest-stable/community" >> /etc/apk/repositories
apk update
apk add curl unzip ncurses postgresql postgresql-contrib nginx certbot
```
## Setup
### Configuring PostgreSQL
#### (Optional) Installing RUM indexes
RUM indexes are an alternative indexing scheme that is not included in PostgreSQL by default. You can read more about them on the [Configuration page](config.html#rum-indexing-for-full-text-search). They are completely optional and most of the time are not worth it, especially if you are running a single user instance (unless you absolutely need ordered search results).
Debian/Ubuntu (available only on Buster/19.04):
```sh
apt install postgresql-11-rum
```
Alpine:
```sh
apk add git build-base postgresql-dev
git clone https://github.com/postgrespro/rum /tmp/rum
cd /tmp/rum
make USE_PGXS=1
make USE_PGXS=1 install
cd
rm -r /tmp/rum
```
#### (Optional) Performance configuration
For optimal performance, you may use [PGTune](https://pgtune.leopard.in.ua), don't forget to restart postgresql after editing the configuration
Debian/Ubuntu:
```sh
systemctl restart postgresql
```
Alpine:
```sh
rc-service postgresql restart
```
### Installing Pleroma
```sh
# Create the Pleroma user
adduser --system --shell /bin/false --home /opt/pleroma pleroma
# Set the flavour environment variable to the string you got in Detecting flavour section.
# For example if the flavour is `arm64-musl` the command will be
export FLAVOUR="arm64-musl"
# Clone the release build into a temporary directory and unpack it
su pleroma -s $SHELL -lc "
curl 'https://git.pleroma.social/api/v4/projects/2/jobs/artifacts/master/download?job=$FLAVOUR' -o /tmp/pleroma.zip
unzip /tmp/pleroma.zip -d /tmp/
"
# Move the release to the home directory and delete temporary files
su pleroma -s $SHELL -lc "
mv /tmp/release/* /opt/pleroma
rmdir /tmp/release
rm /tmp/pleroma.zip
"
# Create uploads directory and set proper permissions (skip if planning to use a remote uploader)
# Note: It does not have to be `/var/lib/pleroma/uploads`, the config generator will ask about the upload directory later
mkdir -p /var/lib/pleroma/uploads
chown -R pleroma /var/lib/pleroma
# Create custom public files directory (custom emojis, frontend bundle overrides, robots.txt, etc.)
# Note: It does not have to be `/var/lib/pleroma/static`, the config generator will ask about the custom public files directory later
mkdir -p /var/lib/pleroma/static
chown -R pleroma /var/lib/pleroma
# Create a config directory
mkdir -p /etc/pleroma
chown -R pleroma /etc/pleroma
# Run the config generator
su pleroma -s $SHELL -lc "./bin/pleroma_ctl instance gen --output /etc/pleroma/config.exs --output-psql /tmp/setup_db.psql"
# Create the postgres database
su postgres -s $SHELL -lc "psql -f /tmp/setup_db.psql"
# Create the database schema
su pleroma -s $SHELL -lc "./bin/pleroma_ctl migrate"
# If you have installed RUM indexes uncommend and run
# su pleroma -s $SHELL -lc "./bin/pleroma_ctl migrate --migrations-path priv/repo/optional_migrations/rum_indexing/"
# Start the instance to verify that everything is working as expected
su pleroma -s $SHELL -lc "./bin/pleroma daemon"
# Wait for about 20 seconds and query the instance endpoint, if it shows your uri, name and email correctly, you are configured correctly
sleep 20 && curl http://localhost:4000/api/v1/instance
# Stop the instance
su pleroma -s $SHELL -lc "./bin/pleroma stop"
```
### Setting up nginx and getting Let's Encrypt SSL certificaties
```sh
# Get a Let's Encrypt certificate
certbot certonly --standalone --preferred-challenges http -d yourinstance.tld
# Copy the Pleroma nginx configuration to the nginx folder
# The location of nginx configs is dependent on the distro
# For Debian/Ubuntu:
cp /opt/pleroma/installation/pleroma.nginx /etc/nginx/sites-available/pleroma.nginx
ln -s /etc/nginx/sites-available/pleroma.nginx /etc/nginx/sites-enabled/pleroma.nginx
# For Alpine:
cp /opt/pleroma/installation/pleroma.nginx /etc/nginx/conf.d/pleroma.conf
# If your distro does not have either of those you can append
# `include /etc/nginx/pleroma.conf` to the end of the http section in /etc/nginx/nginx.conf and
cp /opt/pleroma/installation/pleroma.nginx /etc/nginx/pleroma.conf
# Edit the nginx config replacing example.tld with your (sub)domain
$EDITOR path-to-nginx-config
# Verify that the config is valid
nginx -t
# Start nginx
# For Debian/Ubuntu:
systemctl start nginx
# For Alpine:
rc-service nginx start
```
At this point if you open your (sub)domain in a browser you should see a 502 error, that's because pleroma is not started yet.
### Setting up a system service
Debian/Ubuntu:
```sh
# Copy the service into a proper directory
cp /opt/pleroma/installation/pleroma.service /etc/systemd/system/pleroma.service
# Start pleroma and enable it on boot
systemctl start pleroma
systemctl enable pleroma
```
Alpine:
```sh
# Copy the service into a proper directory
cp /opt/pleroma/installation/init.d/pleroma /etc/init.d/pleroma
# Start pleroma and enable it on boot
rc-service pleroma start
rc-update add pleroma
```
If everything worked, you should see Pleroma-FE when visiting your domain. If that didn't happen, try reviewing the installation steps, starting Pleroma in the foreground and seeing if there are any errrors.
Still doesn't work? Feel free to contact us on [#pleroma on freenode](https://webchat.freenode.net/?channels=%23pleroma) or via matrix at <https://matrix.heldscal.la/#/room/#freenode_#pleroma:matrix.org>, you can also [file an issue on our Gitlab](https://git.pleroma.social/pleroma/pleroma/issues/new)
## Post installation
### Setting up auto-renew Let's Encrypt certificate
```sh
# Create the directory for webroot challenges
mkdir -p /var/lib/letsencrypt
# Uncomment the webroot method
$EDITOR path-to-nginx-config
# Verify that the config is valid
nginx -t
```
Debian/Ubuntu:
```sh
# Restart nginx
systemctl restart nginx
# Ensure the webroot menthod and post hook is working
certbot renew --cert-name yourinstance.tld --webroot -w /var/lib/letsencrypt/ --dry-run --post-hook 'systemctl nginx reload'
# Add it to the daily cron
echo '#!/bin/sh
certbot renew --cert-name yourinstance.tld --webroot -w /var/lib/letsencrypt/ --dry-run --post-hook "systemctl reload nginx"
' > /etc/cron.daily/renew-pleroma-cert
chmod +x /etc/cron.daily/renew-pleroma-cert
# If everything worked the output should contain /etc/cron.daily/renew-pleroma-cert
run-parts --test /etc/cron.daily
```
Alpine:
```sh
# Restart nginx
rc-service nginx restart
# Start the cron daemon and make it start on boot
rc-service crond start
rc-update add crond
# Ensure the webroot menthod and post hook is working
certbot renew --cert-name yourinstance.tld --webroot -w /var/lib/letsencrypt/ --dry-run --post-hook 'rc-service nginx reload'
# Add it to the daily cron
echo '#!/bin/sh
certbot renew --cert-name yourinstance.tld --webroot -w /var/lib/letsencrypt/ --dry-run --post-hook "rc-service nginx reload"
' > /etc/periodic/daily/renew-pleroma-cert
chmod +x /etc/periodic/daily/renew-pleroma-cert
# If everything worked this should output /etc/periodic/daily/renew-pleroma-cert
run-parts --test /etc/periodic/daily
```
### Running mix tasks
Throughout the wiki and guides there is a lot of references to mix tasks. Since `mix` is a build tool, you can't just call `mix pleroma.task`, instead you should call `pleroma_ctl` stripping pleroma/ecto namespace.
So for example, if the task is `mix pleroma.user set admin --admin`, you should run it like this:
```sh
su pleroma -s $SHELL -lc "./bin/pleroma_ctl user set admin --admin"
```
### Updating
Generally, doing the following is enough:
```sh
# Download the new release
su pleroma -s $SHELL -lc "./bin/pleroma_ctl update"
# Migrate the database, you are advised to stop the instance before doing that
su pleroma -s $SHELL -lc "./bin/pleroma_ctl migrate"
```
But you should **always check the release notes/changelog** in case there are config deprecations, special update steps, etc.
## Further reading
* [Configuration](config.html)
* [Pleroma's base config.exs](https://git.pleroma.social/pleroma/pleroma/blob/master/config/config.exs)
* [Hardening your instance](hardening.html)
* [Pleroma Clients](clients.html)
* [Emoji pack manager](Mix.Tasks.Pleroma.Emoji.html)

View File

@ -14,17 +14,19 @@ server {
listen 80;
listen [::]:80;
return 301 https://$server_name$request_uri;
# Uncomment this if you need to use the 'webroot' method with certbot. Make sure
# that the directory exists and that it is accessible by the webserver. If you followed
# the guide, you already ran 'sudo mkdir -p /var/lib/letsencrypt' to create the folder.
# the guide, you already ran 'mkdir -p /var/lib/letsencrypt' to create the folder.
# You may need to load this file with the ssl server block commented out, run certbot
# to get the certificate, and then uncomment it.
#
# location ~ /\.well-known/acme-challenge {
# root /var/lib/letsencrypt/.well-known/acme-challenge;
# root /var/lib/letsencrypt/;
# }
location / {
return 301 https://$server_name$request_uri;
}
}
# Enable SSL session caching for improved performance

View File

@ -114,12 +114,12 @@ def run(["gen" | rest]) do
options,
:db_configurable,
"Do you want to store the configuration in the database (allows controlling it from admin-fe)? (y/n)",
"y"
"n"
) === "y"
dbhost = get_option(options, :dbhost, "What is the hostname of your database?", "localhost")
dbname = get_option(options, :dbname, "What is the name of your database?", "pleroma_dev")
dbname = get_option(options, :dbname, "What is the name of your database?", "pleroma")
dbuser =
get_option(
@ -198,28 +198,16 @@ def run(["gen" | rest]) do
rum_enabled: rum_enabled
)
shell_info(
"Writing config to #{config_path}. You should rename it to config/prod.secret.exs or config/dev.secret.exs."
)
shell_info("Writing config to #{config_path}.")
File.write(config_path, result_config)
shell_info("Writing #{psql_path}.")
shell_info("Writing the postgres script to #{psql_path}.")
File.write(psql_path, result_psql)
write_robots_txt(indexable, template_dir)
shell_info(
"\n" <>
"""
To get started:
1. Verify the contents of the generated files.
2. Run `sudo -u postgres psql -f #{escape_sh_path(psql_path)}`.
""" <>
if config_path in ["config/dev.secret.exs", "config/prod.secret.exs"] do
""
else
"3. Run `mv #{escape_sh_path(config_path)} 'config/prod.secret.exs'`."
end
"\n All files successfully written! Refer to the installation instructions for your platform for next steps"
)
else
shell_error(

11
mix.exs
View File

@ -37,7 +37,7 @@ def project do
pleroma: [
include_executables_for: [:unix],
applications: [ex_syslogger: :load, syslog: :load],
steps: [:assemble, &copy_files/1]
steps: [:assemble, &copy_files/1, &copy_nginx_config/1]
]
]
]
@ -48,6 +48,15 @@ def copy_files(%{path: target_path} = release) do
release
end
def copy_nginx_config(%{path: target_path} = release) do
File.cp!(
"./installation/pleroma.nginx",
Path.join([target_path, "installation", "pleroma.nginx"])
)
release
end
# Configuration for the OTP application.
#
# Type `mix help compile.app` for more information.

View File

@ -2,7 +2,7 @@
# XXX: This should be removed when elixir's releases get custom command support
detect_flavour() {
arch="$(arch)"
arch="$(uname -m)"
if [ "$arch" = "x86_64" ]; then
arch="amd64"
elif [ "$arch" = "armv7l" ]; then

View File

@ -0,0 +1,21 @@
#!/sbin/openrc-run
supervisor=supervise-daemon
# Requires OpenRC >= 0.35
directory=/opt/pleroma
command=/opt/pleroma/bin/pleroma
command_args="start"
command_user=pleroma
command_background=1
# Ask process to terminate within 30 seconds, otherwise kill it
retry="SIGTERM/30/SIGKILL/5"
pidfile="/var/run/pleroma.pid"
depend() {
want nginx
need postgresql
}

View File

@ -0,0 +1,36 @@
[Unit]
Description=Pleroma social network
After=network.target postgresql.service nginx.service
[Service]
KillMode=process
Restart=on-failure
; Name of the user that runs the Pleroma service.
User=pleroma
; Make sure that all paths fit your installation.
; Path to the home directory of the user running the Pleroma service.
Environment="HOME=/opt/pleroma"
; Path to the folder containing the Pleroma installation.
WorkingDirectory=/opt/pleroma
; Path to the Pleroma binary.
ExecStart=/opt/pleroma/bin/pleroma start
ExecStop=/opt/pleroma/bin/pleroma stop
; Some security directives.
; Use private /tmp and /var/tmp folders inside a new file system namespace, which are discarded after the process stops.
PrivateTmp=true
; The /home, /root, and /run/user folders can not be accessed by this service anymore. If your Pleroma user has its home folder in one of the restricted places, or use one of these folders as its working directory, you have to set this to false.
ProtectHome=true
; Mount /usr, /boot, and /etc as read-only for processes invoked by this service.
ProtectSystem=full
; Sets up a new /dev mount for the process and only adds API pseudo devices like /dev/null, /dev/zero or /dev/random but not physical devices. Disabled by default because it may not work on devices like the Raspberry Pi.
PrivateDevices=false
; Ensures that the service process and all its children can never gain new privileges through execve().
NoNewPrivileges=true
; Drops the sysadmin capability from the daemon.
CapabilityBoundingSet=~CAP_SYS_ADMIN
[Install]
WantedBy=multi-user.target