Home-Network/home-network-k3s-impl
2
0 Fork 0
Code Issues 3 Pull Requests 4 Actions Packages Projects Releases Wiki Activity

revert(vpn-ingress): withdraw the L2 lease-collision fix — a node CAN reach a VIP it announces
gate / gate (pull_request) Successful in 1m52s
Details
gate / conformance (pull_request) Has been skipped
Details

Browse Source 
PR #69 claimed a node cannot reach a LoadBalancer VIP whose Cilium L2 announcement
lease it holds, and excluded the VPN ingress node from announcing on that basis.
The claim makes a sharp prediction; today's power-cycle put ALL EIGHT
cilium-l2announce-* leases on k3s-prod-1, and the prediction failed.

Measured 2026-10-17 from the wireguard pod's host netns on each node: the holder
reaches 192.168.2.20/.22/.23/.24 normally (curl 200 in 18ms on .20) while its
ip neigh for every one of them sits at FAILED. The asserted cause is present
without the asserted effect, which falsifies it. .21 (unifi) fails from all three
nodes and is a separate pre-existing fault.

It is inert because no path resolves a VIP by ARP: cilium-dbg reports
KubeProxyReplacement on [eno1, wg0] and programs 192.168.2.20:443 =>
127.0.0.1:13606, so VIPs are DNAT'd in BPF before any neighbour lookup, on the
forwarded path as well as the host one.

Both probes behind the original diagnosis were bad. `ip neigh` FAILED is
permanent and harmless, and a ping artefact. `curl --interface wg0` — the control
used to rule out a host-netns artefact — fails on all three nodes, including ones
whose ARP resolves, so it discriminates nothing. The deeper error: a MASQUERADEd
client is FORWARDED while every pod-runnable probe is socket-originated. Sharing
a netns is not sharing a hook.

Reverted the policy nodeSelector, the Ansible node label, both alerts, their
promtool tests and the kube-state-metrics allowlist entry. Kept documentation
only: the measurements, which probes lie, and the fact that the forwarded path
remains unmeasured without a connected client.

Also recorded, for the reported "LAN and cluster work, internet does not":
new-peer.sh has emitted a FULL tunnel (0.0.0.0/0) by default since f92707f, so
there is no full-tunnel option to add and no code change is made. The server side
is capable — ip_forward=1, the 10.90.0.0/24 MASQUERADE present with an empty
CILIUM_POST_nat ahead of it, FORWARD accepts, default route intact, nodes reach
1.1.1.1. A full tunnel egresses the HOUSE WAN, not PIA: the egress policy selects
pods by egress-class=vpn, this pod is labelled internet, and tun0 exists on
k3s-prod-3 only. Root cause NOT found; no client has connected since the
power-cycle (0 forwarded packets, no handshake), so it is not reproducible here.

Co-Authored-By: Claude Opus 5 (1M context) <[email protected]>
This commit is contained in:
jakman85andClaude Opus 5 committed 2026-10-17 23:58:40 -05:00
1 parent 4656b07178
commit e82dab5df5
13 files changed
+338 -871
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

platform/vpn-ingress/README.md
+74 -44
View File
@@ -117,19 +117,8 @@ boundary around this port is WireGuard's peer authentication, full stop.
The WAN forward names **one node address**, by hand, off-repo (same shape as the plex tcp/32400
forward). A single-replica Deployment would float on every reschedule and silently invalidate it.
A DaemonSet instead puts an identical listener on all three nodes — every node answers the same
keys — so losing the forwarded node never costs a re-issued client profile. The two idle listeners
cost ~15 MiB each.
> **This section used to say "losing the forwarded node is a one-line router change, with no cluster
> work". It was wrong, and it cost two days of silently broken remote access on 2026-10-08.** The
> nodes are warm, but they are **not interchangeable**: the node the forward points at must be the
> one labelled `homelab.keithjackson.dev/vpn-ingress=true`, because that is the node
> `platform-config/cilium/l2-announcement-policy.yaml` excludes from announcing LoadBalancer VIPs.
> **A node cannot reach a VIP whose Cilium L2 announcement lease it holds itself** (see "Measured on
> prod, 2026-10-08" below), and since this pod MASQUERADEs clients to the node's own address, a VPN
> client inherits that. Failover is therefore **two** changes: move
> `node_baseline_vpn_ingress_node` and re-run that role, then re-point the router —
> `runbooks/vpn-remote-access.md` step 3.
keys — so **losing the forwarded node is a one-line router change**, with no cluster work and no
re-issued client profiles. The two idle listeners cost ~15 MiB each.
A pinned `LoadBalancer` VIP was the considered alternative (it would make the forward immune to node
replacement). Rejected: it reintroduces the L2-announcement lease, SNAT, and an extra node-to-node
@@ -197,44 +186,85 @@ right way; one was a defect in my own manifest.
| Is the image's **iptables** backend compatible with the host's? | **Yes — both nft.** The image ships `iptables v1.8.13 (nf_tables)` via `xtables-nft-multi`, and Debian 13 is nf_tables. This was the documented worst fear (rules silently landing in a table the kernel does not consult); it is closed. |
| Can the pod set **`net.ipv4.ip_forward`**? | **No, and it must not try.** `/proc/sys` is mounted **read-only** under `hostNetwork` (`proc /proc/sys proc ro,…`), because those are the host's globals; `sysctl -w` fails with EPERM whatever the capabilities. A failed `PostUp` is fatal to `wg-quick`, so this one line took the whole tunnel down. The node already has it set to `1`; the init container now **verifies** rather than writes. |
## Measured on prod, 2026-10-08 — why the forwarded node cannot be any node
## Measured on prod, 2026-10-17 — L2 announcement does NOT affect what a client can reach
The first device enrolled, handshook, resolved names, and could not open a single connection to any
cluster web app. The tunnel was not at fault — every part of it measured healthy:
`net.ipv4.ip_forward=1`, `wg0` up at `10.90.0.1/24` MTU 1420, the peer authorised with real
bidirectional transfer (346 KiB in / 200 KiB out), `MASQUERADE` present and counting 26 flows,
`FORWARD -i/-o wg0 ACCEPT` counting 108 / 171 packets, and PiHole DNS answering from the node.
A change was proposed (PR #69) on the belief that **a node cannot reach a LoadBalancer VIP whose
Cilium L2 announcement lease it holds**, and that the VPN node therefore had to be excluded from
announcing. **That belief is wrong, and the change was withdrawn.** The half of it that is true is
real but inert, and the distinction matters enough to record, because the wrong version is very
easy to re-derive from `ip neigh` alone.
**The fault was that a node cannot reach a LoadBalancer VIP whose Cilium L2 announcement lease it
holds itself.** Cilium's announcer answers ARP for the VIP on behalf of other LAN hosts, but not for
the announcing node's own request, so the neighbour entry settles at `FAILED`:
Measured with all **eight** `cilium-l2announce-*` leases held by `k3s-prod-1` — the extreme case,
after the UPS power-cycle — from the host netns of the `wireguard` pod on each node:
| from | `192.168.2.20` (Gateway) | `192.168.2.22` (crafty) | `192.168.2.24` (plex) | announces |
|---|---|---|---|---|
| `k3s-prod-1` | **FAILED** | — | — | `.20` |
| `k3s-prod-2` | ok | **INCOMPLETE** | ok | `.21`, `.22`, `.23-tcp` |
| `k3s-prod-3` | ok | ok | **INCOMPLETE** | `.24` |
| | `k3s-prod-1` (holds all 8) | `k3s-prod-2` / `-3` (hold none) |
|---|---|---|
| `ip neigh show 192.168.2.20` | `FAILED` (no lladdr) | `lladdr bc:e9:2f:7e:c4:a6 REACHABLE` |
| `curl -k https://192.168.2.20/` | **`200`, 18 ms** | `200`, 13–40 ms |
| `nc -z 192.168.2.20 443` | **OK** | OK |
| plex `192.168.2.24:32400` | **`200`** | `200` |
Symmetric across 3 nodes × 5 VIPs: each node reached every VIP **except** the ones it alone
announced. It is a property of L2 announcement, not of any one Service, and it is harmless for
ordinary pods — they reach Services through the cluster datapath and never ARP for a VIP. This
component is the exception, because `hostNetwork` + `MASQUERADE` put remote clients on the node's
own LAN path.
So the announcer really does **not** answer its own ARP request — that part of the observation is
reproducible. It changes nothing, because **no traffic path resolves a LoadBalancer VIP by ARP**:
The forward pointed at `k3s-prod-1`, which held the `shared` Gateway lease, so **all 16 Gateway
hostnames were dead over the VPN**. Two things hid it:
* `cilium-dbg status --verbose` reports `KubeProxyReplacement: True [eno1 …, wg0 10.90.0.1]` and
`Masquerading: BPF [eno1, wg0]`. Cilium's LB datapath is attached to **`wg0` as well as `eno1`**,
so a remote client's packet arriving on `wg0` for a VIP is DNAT'd to a backend **in BPF, before
any route or neighbour lookup for the VIP happens**.
* `cilium-dbg service list` on `k3s-prod-1` programs `192.168.2.20:443 => 127.0.0.1:13606` — the
shared Gateway's backend is a **node-local** Envoy socket. Every node can serve that VIP without
a packet crossing the LAN at all.
* Host-originated traffic is socket-LB'd (`Socket LB: Enabled`, `bpf-lb-sock-hostns-only: true`),
which is why `curl` succeeds on the lease holder despite `FAILED`. The discriminator: a **real**
service port connects, a **bogus** one (`nc -z 192.168.2.20 9999`) fails, on all three nodes —
that is service-aware DNAT, not wire traffic.
* **PiHole survived by accident.** `192.168.2.23` is announced by *two* Services
(`pihole-dns-udp` on prod-1, `pihole-dns-tcp` on prod-2), so prod-2 answered the ARP prod-1 could
not. Names resolved perfectly and every app hung — the hardest shape to diagnose from a client.
* **Nothing reported it.** No probe in this repo compares the forwarded node against the lease
holders, and from outside WireGuard is silent to unauthenticated traffic anyway.
**Two tests that look like proof and are not.** Both were used to support the withdrawn diagnosis:
Closed by labelling the forwarded node `homelab.keithjackson.dev/vpn-ingress=true` and excluding
that label in `platform-config/cilium/l2-announcement-policy.yaml` (the node then ARP-resolves every
VIP), with `VpnIngressNodeAnnouncesL2VIP` watching the lease holders so a recurrence is loud and
`VpnIngressNodeLabelMissing` watching for the label going absent, which would silently re-arm the
whole failure. Cost: VIP announcement candidacy drops from three nodes to two.
* **`ip neigh` on its own.** `FAILED` on the holder is permanent and harmless. It is also an
artefact of probing: VIPs answer no ICMP, so a `ping` leaves a failed entry behind whether or not
anything is wrong.
* **`curl --interface wg0`.** This fails (`rc=28`, timeout) from **all three** nodes, including the
two whose ARP for the VIP resolves perfectly. It is not correlated with lease holding, so it
discriminates nothing — and it is not the client path either (see below).
**Why neither probe can see the client path.** A MASQUERADEd remote client's packets are
**forwarded** — netfilter `FORWARD`, then routing — whereas everything runnable from the pod is
**socket-originated**. The two traverse different hooks, so a host-netns probe cannot measure the
forwarded path in either direction: it reports success where the forwarded path might fail, and
`--interface wg0` reports failure where it does not. **The forwarded path can only be measured with
a client actually connected**, by watching the `FORWARD -i wg0` and `POSTROUTING … MASQUERADE`
counters move.
## Measured on prod, 2026-10-17 — where a full tunnel's internet traffic actually egresses
`new-peer.sh` has emitted a **full tunnel** (`AllowedIPs = 0.0.0.0/0`) **by default since the
component's first commit**; `--split` is the opt-in for LAN-only. There is no full-tunnel feature
to add. The server side was checked end to end and is **capable and correct** — if a full-tunnel
client cannot reach the internet, nothing in the list below is the reason:
| checked | result on all three nodes |
|---|---|
| `net.ipv4.ip_forward` (read from `/proc/sys`, never written) | `1` |
| `iptables -t nat -S POSTROUTING` | carries `-s 10.90.0.0/24 ! -o wg0 -j MASQUERADE` — NATs to **any** outbound interface, so internet NAT needs no change |
| `CILIUM_POST_nat`, which precedes it in the chain | **empty, 0 packets** — it cannot short-circuit the MASQUERADE |
| `iptables -L FORWARD` | `-i wg0` and `-o wg0` ACCEPT both present |
| default route | `default via 192.168.2.1 dev eno1 onlink`, intact |
| node's own internet reachability | `http://1.1.1.1/` returns `301` |
**A full tunnel egresses the HOUSE WAN, not PIA.** This was the trap worth checking, and it is
closed: `platform/vpn-egress/egress-steering.yaml`'s `CiliumEgressGatewayPolicy` selects by
`podSelector` on `egress-class: vpn`, and this DaemonSet is deliberately labelled
`egress-class: internet`, so it is not selected. Forwarded `wg0` traffic is not pod-identity
traffic at all, and `tun0` exists on **`k3s-prod-3` only** (`10.150.192.3/18`) — the egress-gateway
node. A remote client's apparent public address is therefore the **home WAN IP**, which is the
intended behaviour and the thing to state when enrolling a device.
**Not measurable without a connected client.** At the time of writing `wg show wg0` listed the one
enrolled peer with **no endpoint and no handshake** on every node, and both the `FORWARD -i wg0` and
the `MASQUERADE` counters read **0 packets** — so no client traffic had been forwarded since the
power-cycle, and which node the router forwards `udp/51820` to was **undeterminable**. Only the
forwarded node shows a peer with an `endpoint` and a recent `latest handshake`.
## Prerequisites outside this directory
platform/vpn-ingress/daemonset.yaml
+3 -33
View File
@@ -62,39 +62,9 @@
# as the plex tcp/32400 forward — runbooks/plex-remote-access.md). A Deployment would float to
# whichever node the scheduler picked and silently invalidate that forward on every reschedule. A
# DaemonSet instead puts an identical warm listener on ALL THREE nodes: the forward names one of
# them, and if that node is lost every other node is already warm and answers the same keys, so no
# client profile ever changes. Each node binds udp/51820 in its own netns, so there is no conflict,
# and the two idle listeners cost ~15 MiB each.
#
# BUT THE NODES ARE NOT INTERCHANGEABLE, AND THIS COMMENT USED TO CLAIM THEY WERE.
# It said "FAILOVER IS A ONE-LINE ROUTER CHANGE with no cluster work". That was wrong, and the
# wording cost two days of silently broken remote access on 2026-10-08. MEASURED that day: a node
# cannot reach a LoadBalancer VIP whose Cilium L2 announcement lease it holds ITSELF — the announcer
# does not answer its own ARP request, so the neighbour entry settles at `FAILED` and packets to
# that VIP die before leaving the node:
#
# k3s-prod-1, holding the shared Gateway lease: ip neigh show 192.168.2.20 -> FAILED
# k3s-prod-2 / k3s-prod-3: ip neigh show 192.168.2.20 -> lladdr …(prod-1)
#
# Because this pod MASQUERADEs clients to the node's own address (see PostUp below), a remote client
# inherits exactly that failure. With the forward on the Gateway's announcing node, all 16 Gateway
# hostnames were unreachable over the VPN while PiHole DNS still resolved — a tunnel that
# handshakes, resolves names, and then hangs on every single app.
#
# THE FIX IS DECLARATIVE AND LIVES OUTSIDE THIS FILE: the forwarded node is labelled
# homelab.keithjackson.dev/vpn-ingress=true by infra/ansible/roles/node_baseline (from
# node_baseline_vpn_ingress_node), and platform-config/cilium/l2-announcement-policy.yaml excludes
# that label from announcing ANY VIP, so the node can ARP-resolve all of them. A label rather than a
# cluster-vars knob because the alert that guards this has to read it too, and the alert unit-test
# harness refuses a postBuild placeholder in an expr. Failover is therefore TWO steps — move
# node_baseline_vpn_ingress_node and re-run the role, THEN re-point the router
# (runbooks/vpn-remote-access.md step 3).
# `VpnIngressNodeAnnouncesL2VIP` (platform/monitoring/alerts.yaml) fires if the two collide again.
#
# This pod deliberately does NOT probe that itself. A readinessProbe on VIP reachability would wedge
# a DaemonSet rollout behind an unrelated Gateway outage (maxUnavailable: 1 below), and a liveness
# probe would tear down a tunnel that is carrying LAN traffic perfectly well. The condition is a
# property of the CLUSTER's announcement layout, so it is watched there.
# them, and if that node is lost, FAILOVER IS A ONE-LINE ROUTER CHANGE with no cluster work and no
# new client profiles — every node answers the same keys. Each node binds udp/51820 in its own
# netns, so there is no conflict, and the two idle listeners cost ~15 MiB each.
#
# All three nodes carry no taints (measured 2026-10-05), so no tolerations are needed for the
# DaemonSet to cover the cluster; add them if a node is ever tainted.