Severity: critical · Fires after: 3m · Rule: central/prometheus/rules/edge.yml
Pi-hole on cloudflare (192.168.0.3) has stopped answering DNS. It is the
LAN's resolver, so name resolution is broken for every host — including
this stack's own scrapes and remote-writes.
probe_success{job="integrations/blackbox/pihole-dns"} == 0
Probed from docker2, deliberately: that means a network path failure or a
host failure trips it too, not just an FTL crash. 3 minutes avoids paging on a
single dropped UDP packet.
Nothing on the LAN works properly without DNS, and the failure presents as everything being slow or intermittently broken rather than as a DNS error. Expect a wave of other alerts to follow if this is not fixed quickly.
dig @192.168.0.3 wikantik.com +short
ssh jakefear@cloudflare 'systemctl status pihole-FTL'
ssh jakefear@cloudflare 'sudo journalctl -u pihole-FTL --since "15 min ago" --no-pager | tail -40'
ssh jakefear@cloudflare 'ss -lnup | grep :53'
ssh jakefear@cloudflare 'sudo systemctl restart pihole-FTL'
If the host itself is unreachable, this is really
IncidentHostNotReporting for cloudflare, and
DNS is one of several things you have lost.
While it is down, resolution can be pointed at a public resolver on the affected host to unblock work:
# temporary, on one host only
echo 'nameserver 1.1.1.1' | sudo tee /etc/resolv.conf
Remember to revert — Pi-hole's filtering is doing real work.
Added 2026-08 (feat(dns): probe Pi-hole DNS and alert when the LAN loses resolution). cloudflare is the LAN's single resolver, so this is a
single point of failure by design; a second resolver would be the structural fix.