Keepalived failover script for Hetzner API
Copy keepalived-hetzner.py to /usr/local/sbin/keepalived-hetzner.py.
Install pip packages from requirements.txt: pip install -r requirements.txt
usage: keepalived-hetzner.py [-h] [--config CONFIG] [--exit] [--debug]
optional arguments:
-h, --help show this help message and exit
--config CONFIG, -c CONFIG
config file (default: /etc/keepalived/hetzner.conf)
--exit, -e exit on error immediately (default: False)
--debug, -d output messages to stdout (default: False)
By default the script writes to syslog (/var/log/syslog). Use --debug to print message to stdout.
Configuration file path is /etc/keepalived/hetzner.conf. Use --config to set different path.
This software tries really hard to make failover. If it fails for whatever reason, it tries every minute again and again. It exits only if failover is confirmed or number of retries is hit (1000). Use --exit to exit immediately after error. Especially useful for debugging.
If you spot any errors or uncaught exceptions please open an issue on github.
Add following line to your keepalived.conf:
notify_master "/usr/local/sbin/keepalived-hetzner.py"
See keepalived-sample.conf file for full example.
Alternatively you can use keepalived-cron.py script to run keepalived-hetzner.py script from crontab every minute.
In some edge cases with keepalived (e.g. flapping scenario) notify_master excecution cannot be trusted leading to inconsistent routing configuration.
See sample file hetzner-sample.conf.
You can specify many failover addresses separated by comma in failover_address config option. The script will fork as many times as addresses you put there. Notice that you may exceed process limit or Hetzner's API requests limit.