Correct two firewall claims the runtime test disproved

FIREWALL_INPUT_PORTS does not open the app ports inside a pod network. gluetun
detects the pod's own subnet and inserts an INPUT accept for it ahead of the
per-port rules, and every packet arriving at the pod IP matches that rule by
destination. Measured with the packet counters: 50 fresh requests moved the
subnet rule from 306 to 356 packets while every --dport rule stayed at 0 since
container start. A port that listens but is not in the list (gluetun's control
server on 8000) is reachable from another pod all the same.

The variable is kept - it is the explicit statement of which ports are meant to
be reachable, and the subnet detection is gluetun's behaviour rather than a
guarantee of this chart - but the comments no longer claim the traffic would be
dropped without it.

Second correction, same run: the /tmp/gluetun emptyDir fixed the "open ...
permission denied" failure, but gluetun also chowns the file, which needs
CAP_CHOWN. With all capabilities but NET_ADMIN dropped that logs "chown
/tmp/gluetun/ip: operation not permitted" on every VPN restart. The file is
written and the public IP is persisted, so this is a log line rather than a
failure, and granting CHOWN to silence it would widen the container's
capabilities for nothing. Noted where the mount is declared.
This commit is contained in:
jklein committed 2026-08-24 14:38:03 +02:00
1 parent bc6577f321
commit e4f70bc856
2 files changed
+26 -5

No files matched your search

+12 -2
View File
@@ -197,8 +197,18 @@ ingress block with enabled: true. Used by the ingress template.
{{/*
Comma-separated FIREWALL_INPUT_PORTS for gluetun: the ports of the enabled app
containers (ingress traffic and the kubelet probes both arrive on the pod IP)
plus 9999 for gluetun's own health server. Without these the gluetun firewall
drops the traffic.
plus 9999 for gluetun's own health server.
Measured caveat: inside a pod network these rules never actually match. gluetun
detects the pod's own subnet and inserts
"-A INPUT -d <pod-subnet> -i eth0 -j ACCEPT" ahead of the per-port rules, and
every packet arriving at the pod IP matches it by destination - the packet
counters on the --dport rules stayed at 0 while the subnet rule absorbed the
traffic. The declaration is kept as an explicit statement of which ports are
meant to be reachable, and because that subnet detection is gluetun's, not a
guarantee of the chart's: a CNI that hands out a /32 or a future gluetun that
drops the rule would make these the only thing keeping probes and ingress
alive. It is not what opens the ports today.
*/}}
{{- define "freesailarr.firewallInputPorts" -}}
{{- $ports := list -}}
+14 -3
View File
@@ -133,9 +133,12 @@ spec:
- name: FIREWALL_OUTBOUND_SUBNETS
value: {{ . | quote }}
{{- end }}
# Open the firewall for the UI/probe ports of the enabled apps and
# the health server - kubelet probes and ingress traffic arrive on
# the pod IP and would be dropped otherwise.
# Declare the UI/probe ports of the enabled apps and the health
# server as reachable. Note that inside a pod network gluetun's own
# pod-subnet rule already accepts this traffic and these per-port
# rules never match (measured - see the helper's comment); they are
# the explicit statement of intent and the fallback if that subnet
# detection ever stops covering the pod IP.
- name: FIREWALL_INPUT_PORTS
value: {{ include "freesailarr.firewallInputPorts" . | quote }}
# Bind the health server to all interfaces so the kubelet HTTP
@@ -169,6 +172,14 @@ spec:
# and every write fails with "clearing public IP data: open
# /tmp/gluetun/ip: permission denied". An emptyDir provides the
# path with usable permissions.
# Remaining, cosmetic: gluetun also chowns the file, which needs
# CAP_CHOWN. With all capabilities but NET_ADMIN dropped that call
# is denied and logs "clearing public IP data: chown
# /tmp/gluetun/ip: operation not permitted" on every VPN restart.
# The file itself is written and the public IP is persisted
# (verified at runtime), so this is a log line, not a failure -
# granting CHOWN just to silence it would widen the container's
# capabilities for nothing.
- mountPath: /tmp/gluetun
name: gluetun-tmp
{{- if .Values.qbittorrent.enabled }}