faild monitors nodes on the same network for availability and in case of errors tries to claim their ip addresses. This can be useful when there is no load balancer available or the setup of some cluster resource manager like pacemaker would be an overkill. The daemon is written in bash, for monitoring it uses arping provided by iputils, for assigning the ip addresses ip provided by iproute2.

• bash >= 4
• arping (provided by iputils)
• ip

copy faild to /usr/sbin/ and make it executable

sudo cp faild /usr/sbin; chmod a+x /usr/sbin/faild

copy faild.conf to /etc/

sudo cp faild.conf /etc

edit /etc/faild.conf

# HOSTS=(
#     '<IP> <NETMASK> <MAC> <DEVICE> <ALIAS>'
#     '1.2.3.4 24 00:00:00:00:00:00 eth0 1'
# )
HOSTS=(
    '10.10.10.1 24 01:23:45:67:89:AB eth0 1'
    '10.10.10.2 24 01:23:45:67:89:CD eth0 2'
)

# timeout in seconds after which a host is declared dead
PING_TIMEOUT=2

# how often should a host be checked before being declared dead
PING_COUNT=2

# how many seconds to wait between checks
CHECK_INTERVAL=0

If you use systemd, just copy faild.service to /lib/systemd/system/ and activate it

sudo cp faild.service /lib/systemd/system
sudo systemctl enable faild.service
sudo systemctl daemon-reload
sudo systemctl restart faild.service

If you use something else, make sure you launch faild on startup and log stderr to some log file.

Checking the logs

journalctl -u faild.service

Reload the configuration using systemd

systemctl reload faild.service

Reload the configuration manually

kill -1 `cat /var/run/faild.pid`

Log current status

kill -USR1 `cat /var/run/faild.pid`

• Thomas Kaltenbrunner - initial work - comexos

This project is licensed under the MIT License - see the LICENSE file for details.