test: gate that LND wallet is unlocked after restart (catches fleet-wide lock)

A wrong/locked LND wallet password leaves the wallet LOCKED after every
restart/OTA, breaking all Bitcoin-receive + Lightning ops fleet-wide — and the
harness was blind to it: live-lnd-address-type treats 'wallet locked' as PASS,
os-audit treated lnd-unreachable as WARN, and the archipelago lnd.getinfo RPC
masks a locked wallet (returns all-zero success).

- tests/release/run.sh: new 'live-lnd-unlocked' stage polls LND's unauth
  /v1/state and FAILs if still LOCKED after a 60s grace window.
- tests/lifecycle/os-audit.sh: probe lnd.newaddress (the real receive path,
  which surfaces LND_WALLET_LOCKED) instead of lnd.getinfo; locked = hard FAIL,
  not-installed = WARN.

Proven on .116 (genuinely locked): os-audit now reports
'[FAIL] lnd wallet unlocked (lnd.newaddress) wallet LOCKED'.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
archipelago 2026-06-14 10:36:12 -04:00
parent 9d3347463a
commit 8c8e4d7a29
2 changed files with 42 additions and 2 deletions

View File

@ -133,10 +133,23 @@ section_a() {
else
record WARN "bitcoin RPC reachable" "bitcoin.getinfo/relay-status did not answer (not installed?)"
fi
# LND wallet must be UNLOCKED. NB: lnd.getinfo masks a locked wallet (it
# returns an all-zero success, error:null), so it can't detect the lock. Probe
# the actual receive path (lnd.newaddress) instead: a LOCKED wallet returns the
# LND_WALLET_LOCKED reason code — the exact fleet-wide receive breakage. A
# locked wallet is a hard FAIL; "not installed" is a WARN. (newaddress derives
# a fresh address — harmless; LND tolerates address gaps.)
if rpc_ok lnd.getinfo; then
record PASS "lnd RPC reachable" ""
local na; na=$(rpc lnd.newaddress)
if grep -qE "LND_WALLET_LOCKED|wallet is locked|WALLET_LOCKED" <<<"$na"; then
record FAIL "lnd wallet unlocked (lnd.newaddress)" "wallet LOCKED — auto-unlock failed (Bitcoin-receive broken)"
elif [[ "$(jq -r '(has("result") and (.result!=null))' <<<"$na" 2>/dev/null)" == "true" ]]; then
record PASS "lnd wallet unlocked (lnd.newaddress)" ""
else
record WARN "lnd RPC reachable" "lnd.getinfo did not answer (not installed / wallet locked?)"
record WARN "lnd wallet unlocked (lnd.newaddress)" "newaddress: $(jq -rc '.error.message // "no address"' <<<"$na" 2>/dev/null | head -c 60)"
fi
else
record WARN "lnd RPC reachable" "lnd.getinfo did not answer (not installed?)"
fi
if rpc_ok system.stats || rpc_ok system.get-metrics; then
record PASS "system metrics reachable" ""

View File

@ -128,6 +128,33 @@ if [[ $LIVE -eq 1 ]]; then
done
echo "SKIP: LND REST not reachable on 18080/8080 — cannot validate address type live"; exit 0
'
# Wallet-unlock guard. After a restart/OTA, LND comes up LOCKED and the backend
# must auto-unlock it; if the unlock password is wrong (e.g. a fleet-wide
# constant vs a per-wallet password) the wallet stays LOCKED forever and ALL
# Bitcoin-receive / Lightning ops fail — fleet-wide, silently. Nothing else in
# this harness catches that: live-lnd-address-type explicitly treats "wallet
# locked" as a PASS, and os-audit treats lnd-unreachable as a WARN. This stage
# polls LND's unauthenticated /v1/state and FAILS if it is still LOCKED after a
# grace window. RPC_ACTIVE = unlocked (pass); NON_EXISTING/WAITING = no wallet
# yet (not a regression); unreachable = skip.
stage "live-lnd-unlocked" bash -c '
deadline=$(( $(date +%s) + 60 ))
while :; do
seen=""
for port in 18080 8080; do
st=$(curl -sk --max-time 6 "https://127.0.0.1:$port/v1/state" 2>/dev/null)
[ -z "$st" ] && continue
seen=1
echo "LND($port) state: $st"
echo "$st" | grep -q "RPC_ACTIVE" && { echo "OK: LND wallet is unlocked"; exit 0; }
echo "$st" | grep -qE "NON_EXISTING|WAITING_TO_START" && { echo "OK: LND wallet not initialized yet — not a lock regression"; exit 0; }
done
[ -z "$seen" ] && { echo "SKIP: LND /v1/state not reachable on 18080/8080"; exit 0; }
[ "$(date +%s)" -ge "$deadline" ] && { echo "FAIL: LND wallet still LOCKED after 60s — auto-unlock failed; Bitcoin-receive/Lightning are broken"; exit 1; }
sleep 5
done
'
fi
summary 0