{{- $name := printf "%s-%s" .Release.Name .Chart.Name -}}
{{- $replicas := include "freesailarr.replicas" (dict "scaleDown" .Values.scaleDown "replicas" .Values.replicas) -}}
{{- /* The effective (null-safe) startup probe, so the budget below matches the
       container even when the block has been erased (#99). */ -}}
{{- $startup := fromYaml (include "freesailarr.probeFor" (dict "root" $ "app" "gluetun" "kind" "startupProbe" "given" .Values.gluetun.startupProbe)) -}}
{{- $budget := mul (int $startup.periodSeconds) (int $startup.failureThreshold) -}}
{{- $ingressApps := include "freesailarr.ingressApps" . | fromJsonArray -}}
{{- $portApps := include "freesailarr.portApps" . | fromJsonArray -}}
{{- $sharedHost := .Values.ingress.host | default "" -}}
{{- $shared := include "freesailarr.sharedIngressApps" . | fromJsonArray -}}
{{- $sharedNames := list -}}
{{- range $shared }}{{ $sharedNames = append $sharedNames .name }}{{ end -}}
{{- $httpRoute := .Values.httpRoute | default dict -}}
{{- $published := include "freesailarr.published" . -}}
{{- /* NOTHING IN CI READS THIS FILE, so every claim below is unverified by any
       pipeline stage. `helm template` does not emit NOTES.txt at all, and no
       workflow or script under ci/ reads it (measured for Gitea #118: the
       three hits for "NOTES" under ci/ are comments in fixtures, not reads).
       The render tests therefore cannot catch a statement here that has gone
       stale, and the install-test never sees it either — it is prose checked
       by review only. Treat an edit to it with the same care as the templates
       it describes, because nothing else will. */ -}}
freesailarr — release {{ .Release.Name }} in namespace {{ .Release.Namespace }}.
{{ if .Values.mediathekarr.enabled }}Ten{{ else }}Nine{{ end }} containers in ONE pod, sharing gluetun's network namespace.

NOTHING IS READY UNTIL THE TUNNEL IS UP
  gluetun's readiness probe is its health server, which answers 200 only while
  the Wireguard tunnel is actually working. So "pod not Ready" almost always
  means "no tunnel", not "app broken". The startup budget is {{ $budget }}s before
  Kubernetes starts killing the container:

    kubectl -n {{ .Release.Namespace }} logs {{ $name }}-sfs-0 -c gluetun -f

  While gluetun is down its firewall keeps blocking every app container's
  egress. That is the kill switch doing its job — the apps stay up and simply
  reach nothing.

THE 1PASSWORD ITEM AND THE TWO FIELDS IT MUST CONTAIN
  secrets.gluetunSecretRef -> {{ .Values.secrets.gluetunSecretRef | default "(not set)" }}
    wireguard-private-key
    wireguard-addresses

  A missing field does not fail the install: gluetun starts with the variable
  empty and never completes a handshake.

    kubectl -n {{ .Release.Namespace }} get onepassworditem {{ $name }}-gluetun-secret
{{- if not .Values.gluetun.vpn.firewallOutboundSubnets }}

SET gluetun.vpn.firewallOutboundSubnets — IT IS EMPTY RIGHT NOW
  Without it the gluetun firewall blocks cluster-internal egress as well:
  lookups against cluster DNS and seerr reaching an in-cluster Jellyfin both
  fail, in a way that looks like an application bug. Set it to the cluster pod
  and service CIDRs, e.g. "10.42.0.0/16,10.43.0.0/16" on k3s.
{{- end }}
{{- /* The servarr apps that are published through a routing object AND have no
       trusted proxy network set - the one combination in which the empty
       default actually costs something. Gated on freesailarr.published plus
       membership in freesailarr.ingressApps, the same pair the ACCESS section
       below uses, so it holds in Gateway mode too: reading ingress.enabled
       directly would silence this warning in exactly the mode whose routes
       still publish the apps. */}}
{{- $noTrust := list }}
{{- if $published }}
{{- range $a := $ingressApps }}
{{- if and (has $a.name (list "prowlarr" "radarr" "sonarr")) (not (index $.Values $a.name).trustedNetworks) }}
{{- $noTrust = append $noTrust $a.name }}
{{- end }}
{{- end }}
{{- end }}
{{- if $noTrust }}

FORWARDED HEADERS ARE NOT TRUSTED — {{ join ", " $noTrust }}
  These apps are published through a routing object and their trustedNetworks
  value is empty, which is the chart's default. They therefore trust no proxy:
  the client address each one logs and displays is your ingress controller's
  pod IP rather than the real client, and each sees the scheme as http even
  where TLS is terminated in front of it. If an app's authentication is set to
  "Disabled for Local Addresses", a proxied request no longer counts as local
  and is asked to authenticate.

  This is not a regression introduced by your values. Upstream removed the
  hard-coded private networks (prowlarr 2.6.3, radarr 6.4.3; sonarr before
  4.0.19) because trusting X-Forwarded-For from any private network is a
  spoofing vector, and this chart follows that default rather than reinstating
  it behind your back.

  To honour the headers, set the network your ingress controller's pods sit in
  — the chart cannot know it, as it is a property of the cluster:

    prowlarr.trustedNetworks: "10.42.0.0/16"   # e.g. the k3s pod CIDR

  Leaving it empty is a valid choice; it is only not a silent one.
{{- end }}

ACCESS
{{- if and $published $ingressApps }}
{{- if .Values.ingress.enabled }}
  INGRESS MODE (ingress.enabled: true, httpRoute.enabled: false)
{{- else }}
  GATEWAY MODE (httpRoute.enabled: true, ingress.enabled: false)
{{- end }}
{{- range $shared }}
  {{ .name }}: https://{{ $sharedHost }}{{ if ne .path "/" }}{{ .path }}{{ end }}
{{- end }}
{{- range $ingressApps }}
{{- $cfg := (index $.Values .name).ingress }}
{{- if not (has .name $sharedNames) }}
  {{ .name }}: https://{{ $cfg.host }}
{{- end }}
{{- end }}
{{- if $shared }}

  These apps share one host{{ if .Values.ingress.enabled }}, one Ingress and one certificate{{ else }} and one HTTPRoute ({{ $name }}-httproute), one
  rule each{{ end }}.
{{- if has "bazarr" $sharedNames }}

  BAZARR NEEDS ONE MANUAL STEP UNDER A SUB-PATH
    Bazarr reads its base path only from its own configuration - it has no
    environment variable for it, so the chart cannot set it. Open bazarr,
    go to Settings -> General -> Base URL, enter

      {{ include "freesailarr.appPath" (dict "root" $ "app" "bazarr") }}

    and restart the pod. Until then bazarr answers only at the root and the
    sub-path shows a broken UI. radarr, sonarr and prowlarr need nothing -
    the chart sets their UrlBase itself.
{{- end }}
{{- if has "qbittorrent" $sharedNames }}

  QBITTORRENT SITS BEHIND A PREFIX-STRIPPING REWRITE
    It has no base path of its own, so the prefix has to be cut off before the
    request reaches it.
    {{- if .Values.ingress.enabled }} Under an Ingress that costs a SECOND object on the
    same host, because nginx applies rewrite-target per object and the other
    apps must not be rewritten.
    {{- else }} In Gateway mode that second object is gone: a
    URLRewrite filter belongs to a RULE, so qBittorrent is a rule of
    {{ $name }}-httproute with the filter on it and its neighbours are
    untouched by construction — no extra object, no regex path.
    {{- end }}

    OPEN IT WITH THE TRAILING SLASH:

      https://{{ $sharedHost }}{{ include "freesailarr.appPath" (dict "root" $ "app" "qbittorrent") }}/

{{ if .Values.ingress.enabled }}    Without the slash the prefix is stripped to "/", qBittorrent answers 200,
    and every RELATIVE link in that page resolves against the ROOT of this
    host — which is {{ if has "seerr" $sharedNames }}seerr{{ else }}whatever serves "/" here{{ end }}, not qBittorrent. The login form then looks
    unstyled and does nothing while another application answers its requests.
    This chart does NOT render that redirect in Ingress mode: nginx offers only
    a configuration-snippet annotation for it, and snippets are ignored unless
    the controller runs with allow-snippet-annotations=true, which has been off
    by default since ingress-nginx 1.9 and means accepting annotation risk
    level "Critical". The working snippet, and its price, are in values.yaml at
    qbittorrent.ingress (Gitea #118).

{{ else }}    The bare, slashless path is redirected 301 to the slashed form by a rule of
    {{ $name }}-httproute — an Exact path match, which the Gateway API ranks
    above every prefix match, so it wins wherever it sits in the list. Without
    it qBittorrent would answer 200 under a URL whose relative links resolve to
    the root of this host, which is a different application (Gitea #118). This
    is the one thing the Gateway mode does here that the Ingress mode cannot.

{{ end }}    Two further things look exactly like a login failure and send you
    looking in the wrong place: its session cookie is marked "Secure" (over
    plain HTTP the browser never sends it back and every click bounces to the
    login form), and it bans the client IP for a while after a few failed
    attempts. If it does not work at all, set qbittorrent.ingress.host and it
    is back on a host of its own - no chart change needed.
{{- end }}
{{- if not (has "seerr" $sharedNames) }}

  NOTHING SERVES THE ROOT OF {{ $sharedHost }}
    seerr is the app that sits at "/" and it is not on the shared host, so
    https://{{ $sharedHost }}/ answers 404. The sub-paths are unaffected.
{{- end }}
{{- end }}

{{- if .Values.ingress.enabled }}
  Every app listed with a host of its own has its OWN Ingress and its own TLS
  secret. While cert-manager is still issuing, the browser warns:

    kubectl -n {{ .Release.Namespace }} get certificate
{{- else }}
  Every app listed with a host of its own has its OWN HTTPRoute
  ({{ $name }}-<app>-httproute).

  THREE THINGS BELONG TO THE CLUSTER AND NOT TO THIS CHART IN THIS MODE, and if
  any is missing the addresses above do not answer:

    1. The Gateway-API CRDs. Without them this install would have failed.
    2. The Gateway named in httpRoute.parentRefs
{{- range $httpRoute.parentRefs }}
         {{ .name }}{{ if .namespace }} in namespace {{ .namespace }}{{ end }}{{ if .sectionName }}, listener {{ .sectionName }}{{ end }}
{{- end }}
       must exist AND its listener must allow routes from this namespace
       ({{ .Release.Namespace }}) in allowedRoutes. Every route of this release
       attaches to that same parent. A rejected attachment shows up only in a
       route's status:

         kubectl -n {{ .Release.Namespace }} get httproute -o yaml

       Read status.parents[].conditions — "Accepted: True" and
       "ResolvedRefs: True" is a working attachment. An empty status means no
       controller has looked at the route at all.
    3. The TLS certificates for these hostnames, on the Gateway's listener.
       ingress.clusterIssuer is NOT used in this mode and is not required, and
       neither are ingress.className or the per-app overrides of either: the
       chart renders no Certificate and no class, so there are no
       tls-{{ $name }}-* secrets and nothing for cert-manager to do.

  ingress.annotations HAS NO COUNTERPART HERE, per-app overrides included: the
  Gateway API has no annotation channel. The chart needs nothing from it —
  qBittorrent's rewrite is a filter in this mode — but an annotation you added
  yourself is gone without a word. A body-size cap for qBittorrent's
  torrent-file uploads is the likely one; set it at the Gateway or through its
  implementation's own policy object.
{{- end }}
{{- else }}
  Nothing is published{{ if and (not .Values.ingress.enabled) (not $httpRoute.enabled) }} — ingress.enabled and httpRoute.enabled are both false{{ else if not $ingressApps }} — no enabled app carries <app>.ingress.enabled: true{{ end }}. Reach an app through its own service:
{{- range $portApps }}

    kubectl -n {{ $.Release.Namespace }} port-forward svc/{{ $name }}-{{ .name }}-service {{ .port }}:{{ .port }}
{{- end }}
{{- end }}
{{- if .Values.qbittorrent.enabled }}

  qBittorrent's WebUI rejects a Host header whose port differs from its own
  ({{ "8080" }}) with a plain 401. A port-forward on a different local port therefore
  looks like a login failure — forward 8080:8080, or go through the published
  address above.
{{- if .Values.qbittorrent.savePath }}

QBITTORRENT DOWNLOADS GO TO {{ .Values.qbittorrent.savePath }}
  The chart seeds Session\DefaultSavePath into qBittorrent's config, because
  its built-in default is /config/Downloads — the config volume{{ with .Values.qbittorrent.storage.size }} ({{ . }}){{ end }}, which
  fills within minutes and puts EVERY torrent into "errored" while a large,
  empty media volume sits next to it. The failure looks like a network or VPN
  problem; only "df" shows what it is.

  The value is seeded ONLY when the key is absent, never replaced. So if you
  changed the path in the WebUI, it stays yours and this value does nothing —
  check which one is actually in effect:

    kubectl -n {{ .Release.Namespace }} exec {{ $name }}-sfs-0 -c qbittorrent -- \
      grep SavePath /config/qBittorrent/qBittorrent.conf

  If that shows a path under /config, downloads are landing on the config
  volume: fix it under Options -> Downloads -> Default Save Path (the running
  app owns the file; editing it under the app is overwritten on exit).
{{- else }}

QBITTORRENT HAS NO SAVE PATH SET (qbittorrent.savePath is empty)
  It therefore falls back to /config/Downloads — the small config volume. That
  fills within minutes and puts every torrent into "errored". Either set
  qbittorrent.savePath or set Options -> Downloads -> Default Save Path in the
  WebUI before starting any download.
{{- end }}
{{- end }}
{{- if .Values.mediathekarr.enabled }}
{{- $dir := .Values.mediathekarr.downloads.directory }}
{{- $gate := .Values.mediathekarr.startGate }}
{{- $mediaStartup := fromYaml (include "freesailarr.probeFor" (dict "root" $ "app" "mediathekarr" "kind" "startupProbe" "given" .Values.mediathekarr.startupProbe)) }}

MEDIATHEKARR: TWO PORTS, AND THE SABNZBD API IS NOT AT /api
  Every container shares one network namespace, so both consumers reach it over
  127.0.0.1 — no service name, no DNS, no ingress:

    Prowlarr      -> Newznab indexer, http://127.0.0.1:5008
    Sonarr/Radarr -> SABnzbd host 127.0.0.1, port 5007, API path /download/api

  A request to /api on 5007 answers with a Newznab validation error and sends
  you looking at the wrong half of the app — the upstream documentation only
  mentions port 5007 at all.

  Sonarr and Radarr must import from

    /downloads/{{ $dir }}/complete     (one sub-directory per category below it)

  which is exactly the path the chart hands MediathekArr and which it reports
  back as the SABnzbd complete_dir. Any deviation — a remote path mapping, a
  different mount — imports NOTHING, without an error.

  The UI (setup wizard and download queue) is not published{{ if and .Values.mediathekarr.ingress.enabled .Values.mediathekarr.ingress.host }}, other than on the
  host you gave it, {{ .Values.mediathekarr.ingress.host }}{{ end }}:

    kubectl -n {{ .Release.Namespace }} port-forward svc/{{ $name }}-mediathekarr-service 5007:5007
{{- if $gate.enabled }}

  IT DELIBERATELY STARTS LATE, AND THAT IS THE POINT
    The container waits (up to {{ $gate.timeoutSeconds }}s) for gluetun to report healthy before
    it starts MediathekArr{{ if gt (int $gate.settleSeconds) 0 }}, plus {{ $gate.settleSeconds }}s of settling time{{ end }}. Started before the tunnel,
    MediathekArr fails its one ruleset fetch, never retries it, and then answers
    EVERY search with 0 results — container up, Ready, nothing in the log. That
    is why a green pod proves nothing here and why the gate exits instead of
    starting anyway on timeout: a crash-looping container is visible, a silently
    empty indexer is not.

      kubectl -n {{ .Release.Namespace }} logs {{ $name }}-sfs-0 -c mediathekarr | head -3

    The startup probe has to outlast the gate ({{ mul (int $mediaStartup.periodSeconds) (int $mediaStartup.failureThreshold) }}s budget against a {{ $gate.timeoutSeconds }}s gate),
    otherwise the kubelet kills the container while it is still waiting.
{{- else }}

  THE START GATE IS OFF (mediathekarr.startGate.enabled: false)
    MediathekArr may then start before the Wireguard tunnel is up. Its ruleset
    fetch fails once, is never retried, and the indexer answers every search
    with 0 results afterwards — measured over eleven minutes, with the container
    up, Ready and no error anywhere. Nothing reports this state. Check a search
    by hand after every restart, or switch the gate back on.
{{- end }}

A FAILED MEDIATHEKARR DOWNLOAD NEVER SAYS WHY
  MediathekArr does not check the HTTP status code. On a geo-blocked fetch it
  writes the ~476-byte HTML error page out as .mp4 and logs "Download
  completed"; only mkvmerge then refuses the file and the status flips to
  Failed — with fail_message: null. Measured A/B with one identical NZB:
  direct 94,235,525 bytes and Completed, through a foreign exit 0 bytes and
  Failed, fail_message null in BOTH cases.

  So a geo block is indistinguishable from a broken server, an expired URL or a
  full disk, in Sonarr and in the container log alike. And the protection is an
  accident: it rests entirely on mkvmerge rejecting an HTML page. Anything a
  geo endpoint returns that mkvmerge accepts is reported as Completed and walks
  into your library.

WHETHER A SHOW WORKS THROUGH YOUR EXIT: READ THE MARKER, NOT A LIST
  The tunnel currently exits {{ with .Values.gluetun.vpn.serverCities }}in {{ . }}{{ else }}{{ with .Values.gluetun.vpn.serverCountries }}in {{ . }}{{ else }}wherever gluetun picks{{ end }}{{ end }}. ARD and ZDF block parts of their
  catalogue outside Germany, and every container hangs in gluetun's namespace —
  there is no per-container exit.

  A table of series goes stale; the marker in the media URL does not. It
  predicted the outcome in 584 of 584 measured cases:

    /de/, /geo/de/, /deutschlandweit/, geoavailability_D  -> blocked outside DE
    /weltweit/                                            -> open
    /none/          (ZDF)                                 -> open
    /dach/, deChAt  (WDR variant)                         -> open

  The rule of thumb "own production = available worldwide" is MEASURABLY WRONG
  (an ARD own production was blocked, a licensed series was open). Only the
  marker counts.

  The escape route is one value — gluetun.vpn.serverCountries/serverCities set
  to a German exit — and its price is that it moves the WHOLE stack, all ten
  containers, qBittorrent included, into that jurisdiction and onto those
  servers. That is a deployment decision, not a chart setting to flip lightly.

OPEN, EXPLICITLY UNMEASURED: HLS ENTRIES
  Part of the catalogue is served as an HLS playlist instead of a progressive
  MP4. The image carries an ffmpeg-based M3U8 path, but whether such an entry
  completes end to end has not been measured. If it does not, those entries fail
  regardless of any geo block — and, per the paragraph above, without a reason.
{{- end }}

EVERY SERVICE POINTS AT THE SAME POD IP
  One ClusterIP service per app{{ if .Values.recyclarr.enabled }} (recyclarr has none — no port in daemon mode){{ end }}, but all
  containers share gluetun's network namespace, so the services differ only in
  their port. They give you stable per-app DNS names, NOT network separation
  between the apps.

TOGGLING AN APP OFF AND ON LATER NEEDS ONE MANUAL STEP
  The config volumes come from volumeClaimTemplates, and those are immutable.
  Changing any <app>.enabled after the first install makes the upgrade fail
  until the StatefulSet is recreated — the data survives:

    kubectl -n {{ .Release.Namespace }} delete statefulset {{ $name }}-sfs --cascade=orphan

THE MEDIA CLAIMS ARE YOURS, THE CONFIG VOLUMES ARE THE CHART'S
  media.<category>.existingClaim is mounted, never created, and is meant to be
  shared with your Jellyfin. One claim can serve all three categories via
  media.<category>.subPath.

  Measured, so you do not plan around something that does not happen: Radarr
  and Sonarr do NOT hardlink between the categories, not even on one claim.
  Kubernetes makes every volumeMount its own bind mount and link() across two
  mount points fails with EXDEV regardless of subPath, so an import copies.

  HARDLINKS NEED ONE MOUNT, NOT ONE CLAIM
    This is the point where the obvious conclusion is wrong, so it is spelled
    out: putting downloads on the same claim as movies and tv, as a third
    subPath, does NOT produce hardlinks. It buys the layout, nothing else.

    The kernel compares the VFSMOUNT in do_linkat, not the superblock, and two
    bind mounts of the same filesystem are two vfsmounts — so link() across
    them is EXDEV even though it is one volume. Measured on Longhorn/NFS: both
    mount points report st_dev 0:56, the mount table shows two mount IDs on
    that same device, a cross-mount link() fails, a link() inside one mount
    succeeds with link count 2.

    Note the trap in that measurement: "stat" reports the SAME device number
    on both sides. Every intuition says a hardlink must work there. It does
    not, and the device number is exactly what makes this look like a bug in
    the app instead of a property of the mount namespace.

    The layout that WOULD hardlink is one claim with exactly ONE mount and the
    categories as sub-directories inside it — the "single /data mount" the *arr
    guides recommend. This chart cannot express that today: it mounts each
    category separately. Naming the possibility, not giving instructions —
    there are no values for it.

  The containers write as {{ .Values.puid }}:{{ .Values.pgid }}. The claim's root directory has to be
  writable for them, and a subPath directory the kubelet creates inherits that
  root's mode — so fix the ownership before the first start. The pod will not
  do it for you: fsGroupChangePolicy is OnRootMismatch so a large library is
  not re-chowned on every start, and on an RWX claim a driver like Longhorn
  (fsGroupPolicy ReadWriteOnceWithFSType) ignores fsGroup altogether.
{{- if eq $replicas "0" }}

SCALED TO ZERO
  {{ if .Values.scaleDown }}scaleDown: true{{ else }}replicas: 0{{ end }} — the StatefulSet runs with replicas: 0. PVCs, secrets,
  services{{ if .Values.ingress.enabled }} and ingresses{{ else if $httpRoute.enabled }} and HTTPRoutes{{ end }} stay in place; set it back and upgrade to bring the
  stack up again.
{{- else if ne $replicas "1" }}

MORE THAN ONE REPLICA IS NOT MEANINGFUL
  replicas is {{ $replicas }}. Each replica would run its own isolated stack on its own
  config volumes, all of them fighting over the same media claims and the same
  VPN credentials.
{{- end }}
{{- if .Values.networkPolicy.enabled }}

NETWORKPOLICY IS ACTIVE — WHAT IT CAPS, AND WHAT IT DOES NOT
  {{ .Release.Name }}-{{ .Chart.Name }}-networkpolicy selects this workload's pods.
  Inbound: the nine probed ports — gluetun's health server (9999) and the eight
  web interfaces (8080, 9696, 8191, 7878, 8989, 5055, 6767, 5007) — plus
  mediathekarr's newznab port (5008), which is RESTRICTED by source because
  nothing probes it and its only consumer is Prowlarr in this same pod.
  Outbound: DNS, UDP 51820 (the WireGuard tunnel) and TCP 443 (gluetun's
  server-list API){{ with .Values.networkPolicy.egress }}{{ with .extraPorts }}, plus TCP {{ join ", " . }}{{ end }}{{ end }}.

  THE TEN CONTAINERS TALK TO EACH OTHER OVER LOCALHOST, so none of that traffic
  is governed by this object at all — they share a network namespace. And
  because that namespace belongs to gluetun, everything the *arr containers
  send is encapsulated into the tunnel before it leaves the pod: indexer and
  tracker traffic needs no egress rule of its own.

  IF THIS POD WILL NOT BECOME READY, SUSPECT THE TUNNEL RULE. gluetun's health
  server answers 200 only while the tunnel is up, and all three of its probes
  address it, so a blocked UDP 51820 shows up as a pod that never goes Ready
  rather than as a network error. 51820 is WireGuard's and Mullvad's standard
  port; Mullvad also offers 53, 123, 500 and 4500 for obfuscation, and a
  release on one of those has to widen the rule in the template —
  networkPolicy.egress.extraPorts is TCP only.

  AND IF SEERR CANNOT REACH AN IN-CLUSTER SERVICE (Jellyfin, say), two things
  have to be true, not one: its port in networkPolicy.egress.extraPorts, and
  gluetun.vpn.firewallOutboundSubnets set — gluetun's own firewall blocks
  cluster-internal egress before this policy is ever consulted.

  Three limits belong in the same breath, because none of them is visible in
  the object itself:

  - EGRESS IS BY PORT, NEVER BY DESTINATION. A NetworkPolicy matches CIDRs and
    selectors and cannot express a hostname, so every egress rule permits its
    port to ANY address. That is a limit of the object, not a shortcut — this
    cluster's CNI (Canal: Flannel plus Calico) has no FQDN rule either. An
    outbound port the chart does not know about is BLOCKED: add it to
    networkPolicy.egress.extraPorts, and expect to find out through an
    application-level connection error rather than through this object.
{{- if .Values.networkPolicy.probeSources }}
  - THE PROBE PATH IS RESTRICTED in this release, because
    networkPolicy.probeSources is set to
      {{ join ", " .Values.networkPolicy.probeSources }}
    Probes come from the IP of the WORKER NODE running the pod. If none of those
    CIDRs covers the node this pod lands on, the kubelet's probes are dropped
    and THE POD NEVER BECOMES READY — the rollout then times out and the symptom
    looks like a broken application. Compare

      kubectl -n {{ .Release.Namespace }} describe pod -l app={{ .Release.Name }}-{{ .Chart.Name }}-sfs

    against `kubectl get node -o wide`.
{{- else }}
  - THE PROBED PORT IS OPEN INSIDE THE CLUSTER. networkPolicy.probeSources is
    empty, so that rule carries no source selector at all and any pod the CNI
    routes to can reach it. Not from the internet — but this is NOT pod-level
    isolation. Narrowing it means listing the worker node IPs in
    networkPolicy.probeSources, because the kubelet probes from the node's own
    address and the chart cannot know it; a wrong guess there leaves the pod
    permanently un-Ready.
{{- end }}
  - IT IS ONLY A BOUNDARY WHERE THE CNI ENFORCES POLICIES. Where it does not,
    this object is accepted and filters nothing, so everything reads as
    configured while nothing is filtered. `kubectl -n {{ .Release.Namespace }} get networkpolicy`
    shows the object exists; only a connection attempt from an unrelated
    namespace shows that it bites.

  Switch it off with networkPolicy.enabled: false.
{{- end }}
