dns-watchdog

#+Title: A Simple DNS-Resolver watchdog
#+Author: Carsten Strotmann
#+Date: <2020-01-20 Mon>

* A DNS-Server process watchdog

This project implements a simple DNS server process watchdog. It sends
DNS queries to a local DNS server projects and writes the results to
syslog.

If the queries are failing, it will terminate with an error code. A
script can then pickup the error code and take an appropriate action
(like restarting the DNS server process).

This watchdog has been created because we've seen BIND 9 processes
getting stuck (not answering any queries, but not terminating) on
(very busy) Ubuntu 18.04 systems.

** Example use with systemd

The project comes with an example systemd unit file for the watchdog.
The watchdog process will query the DNS server/DNS resolver for
"hostname.bind CH TXT". If three (default) queries fail in a row, the
watchdog will exit with error code 128. The Systemd =ExecStartPost=
process will be executed, in this example it will kill a stuck DNS
server process.

#+begin_example
[Unit]
Description=DNS Service Watchdog
Documentation=https://github.com/sys4/dns-watchdog

[Service]
Type=simple
ExecStart=/usr/local/sbin/dns-watchdog
ExecStartPost=/usr/bin/pkill -9 named
Restart=always

PrivateDevices=true
ProtectControlGroups=true
ProtectHome=true
ProtectKernelTunables=true
ProtectSystem=full
RestrictSUIDSGID=true

[Install]
WantedBy=multi-user.target
#+end_example

The DNS server process should be managed by a supervisor (such as
Systemd, or supervised or runit) and will be restarted from the
supervisor process.

* Work in progress

This projects is currently /work in progress/, additional
documentation will be available soon.