cloudflared-watchdog.sh

#!/bin/sh

# Watch for cloudflared failures and restart the service.
#
# This is necessary due to a known issue between cloudflared and Raspberry Pis:
# https://github.com/cloudflare/cloudflared/issues/23
#
# Script based on reddit thread:
# https://www.reddit.com/r/pihole/comments/b94dcr/cloudflare_dns_over_https_doesnt_work_after_a/err2rbj/
#
# Steps:
#
# Install procmail (for lockfile)
# sudo apt-get install procmail
#
# Place this script somewhere like ~/scripts/cloudflared-watchdog.sh 
#
# Add a cron:
# * * * * * root /bin/sh /absolute/path/to/cloudflared-watchdog.sh
#

TEMP_STATUS="/tmp/cloudflared.status"
TEMP_ERR="/tmp/cloudflared.error"
TEMP_LOCK="/tmp/cloudflared-watchdog.lock"

lockfile -r 0 $TEMP_LOCK || exit 1

sudo systemctl status cloudflared > $TEMP_STATUS

cat $TEMP_STATUS | grep "failed to connect to an HTTPS backend" > $TEMP_ERR
cat $TEMP_STATUS | grep "failed to perform an HTTPS request" >> $TEMP_ERR

if [ -s $TEMP_ERR ]; then
	sudo systemctl restart cloudflared
fi

rm -f $TEMP_STATUS $TEMP_ERR $TEMP_LOCK