nico
ac0ea074d5
* update nodename method to not use regex anymore * update readme with 2 new columns for the exit code table We used quite the broad regex to parse the nodename from the status string. But we relied on a single quote to mark the starting point. It seems some distributions don't have this starting quote, therefore breaking the api. |
||
---|---|---|
contrib/init/linux-systemd | ||
.gitignore | ||
.gitlab-ci.yml | ||
.pre-commit-config.yaml | ||
README.md | ||
api.py | ||
calls.py | ||
cleanup.py | ||
config.py | ||
ejabberd-metrics.yml.default | ||
influx.py | ||
metrics.py | ||
prometheus.py | ||
pyproject.toml | ||
requirements.txt | ||
setup.cfg |
README.md
ejabberd-tools
Repository containing various ejabberd tools.
tools
metrics
The metrics class is a tool to gather and aggregate metrics data through the ejabberd rest/ xmlrpc interface. In addition to that a Prometheus and InfluxDB client utilizing the metrics class is provided.
requirements
The easiest way to setup a clean Python project environment is to use a virtual environment inside the cloned
repository directory. The following bash lines install the python-virtualenv
module, create the virtual environment
using Python3 and finally install all dependencies listed in the requirements file.
# Debian
apt install python-virtualenv
# Arch
pacman -S python-virtualenv
# create a venv folder inside the cloned repository
mkdir venv
virtualenv -p python3 venv/
source ./venv/bin/activate
pip install -r requirements.txt
ejabberd
To operate the api it is required to provide a user account eligible to issue api commands. We strongly recommend to use a dedicated user to operate the api and to secure this account with a specifically strong password.
configuration
ejabberd: api permissions
These configurations options first add the mod_http_api
listener. Additionally the user api_user@magicbroccoli.de
is
added to the acl
group, thus enabling him to access the mod_http_api
configured in the api_permissions
block.
listen:
-
port: 5280
ip: "127.0.0.1"
tls: true # optional
module: ejabberd_http
request_handlers:
/api: mod_http_api
acl:
api:
user:
- "api_user@magicbroccoli.de"
loopback:
ip:
- 127.0.0.0/8
- ::1/128
- ::FFFF:127.0.0.1/128
api_permissions:
"api access":
from:
- mod_http_api
who:
access:
allow:
acl: api
acl: loopback # optional but recommended to restrict the api access to the local network
what:
- "*"
- "!stop"
- "!start"
ejabberd-metrics.yml
The ejabberd-metrics.yml
file is the central configuration file which is used by all current and future tools inside
this repository.
We provide an ejabberd-metrics.yml.default
file, listing all possible and potentially necessary parameters. The final
configuration file should be located at /etc/ejabberd-metrics.yml
.
systemd service
To properly operate the metrics exporter tools, we created some systemd service templates, to simplify the whole
process. If the ejabberd-metrics.yml
file is not accessible for the usernobody:nogroup
, it is required to update the
User
and Group
definitions inside the service file.
If a virtualenv is used, it is required to update the Environment=PATH
to include the venv/bin
directory created
earlier.
[Unit]
Description=ejabberd influxdb exporter
[Service]
Type=simple
# this strongly depends on the ejabberd-metrics.yml permissions
User=nobody
Group=nogroup
ExecStart=/opt/ejabberd-metrics/influx.py
Restart=always
RestartSec=5s
# if the virtualenv is used PATH needs to customized
Environment=PATH=/opt/ejabberd-metrics/venv/bin:/usr/bin:/usr/local/bin
[Install]
WantedBy=multi-user.target
Another possible solution would be to edit the ExecStart
parameter to include the virtualenv Python intepreter.
error codes
Lookup table for all custom error codes. The potential reasons are sorted by probability of being the root cause.
| code | raised by | cause by | potential reason | | :---: | :---| : --- | :--- | | 17 | calls | api call returned empty | login credential mismatch, potential api permission problem |
pre-commit framework
This project utilizes the pre-commit framework to automate various small hick-ups that tend to happen prior to committing.
To use the framework it is necessary to follow these steps:
# install the pre-commit hook
pre-commit install
# to test your staged files manually run
pre-commit run