fix(tests): make required-stack.bats portable across nodes with different app rosters

Found live during the .5 multinode-pass run: this suite was hardcoded to
.116's exact app bundle (including the mempool stack), so any node missing
an app hard-failed instead of skipping — and a missing local fail() helper
(present in 3 sibling bats files, absent here) masked the real error as
"command not found" (exit 127). Add the same skip-if-absent idiom already
used in mempool.bats per-app, and define fail() locally like the others.
Verified: skips cleanly on .116 (no bitcoin-knots here), still exercises
real checks for apps that are installed.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
archipelago 2026-07-01 13:56:24 -04:00
parent 6b7af884ab
commit f1055164d2

View File

@ -1,7 +1,14 @@
#!/usr/bin/env bats
# tests/lifecycle/bats/required-stack.bats
#
# Read-only release-gate checks for the required Bitcoin stack on .116.
# Read-only release-gate checks for the Bitcoin/electrum/lnd/mempool stack.
# Originally written against .116's fixed app roster; the "present"/"running"
# checks below now only require containers actually installed on THIS node
# (podman_all_names — present in `podman ps -a` even if stopped), so a node
# with a different app subset (e.g. no mempool stack) doesn't hard-fail on
# apps it was never meant to have. Per-app checks further down (mempool,
# filebrowser, ...) skip individually if that app isn't installed, matching
# the mempool_skip_if_absent idiom in mempool.bats.
#
# This suite is intentionally non-destructive and does not use RPC auth;
# it can run anytime as a health gate during long sync/reindex windows.
@ -19,15 +26,38 @@ required_containers=(
"archy-electrs-ui"
)
fail() { echo "$@" >&2; return 1; }
podman_names() {
podman ps --format '{{.Names}}'
}
podman_all_names() {
podman ps -a --format '{{.Names}}'
}
container_running() {
local name="$1"
podman inspect --format '{{.State.Running}}' "$name" 2>/dev/null
}
container_installed() {
local name="$1"
podman_all_names | grep -Fx "$name" >/dev/null
}
skip_if_not_installed() {
container_installed "$1" || skip "$1 not installed on this node"
}
# The subset of required_containers actually installed on this node.
installed_required_containers() {
local c
for c in "${required_containers[@]}"; do
container_installed "$c" && echo "$c"
done
}
bitcoin_rpc() {
curl -fsS --max-time 60 \
--user "archipelago:$(cat /var/lib/archipelago/secrets/bitcoin-rpc-password)" \
@ -42,13 +72,16 @@ bitcoin_json() {
@test "required containers are present" {
# Under sustained 5× churn an app may still be mid-restart when this runs;
# wait for the whole required set rather than single-shot.
# wait for the whole required set rather than single-shot. Only checks
# containers actually installed on this node (see installed_required_containers).
local targets; targets="$(installed_required_containers)"
[[ -n "$targets" ]] || skip "none of required_containers installed on this node"
local deadline=$(( $(date +%s) + 180 )) names missing
while (( $(date +%s) < deadline )); do
names="$(podman_names)"; missing=""
for c in "${required_containers[@]}"; do
while IFS= read -r c; do
echo "$names" | grep -Fx "$c" >/dev/null || missing="$missing $c"
done
done <<< "$targets"
[[ -z "$missing" ]] && return 0
sleep 3
done
@ -56,12 +89,14 @@ bitcoin_json() {
}
@test "required containers are running" {
local targets; targets="$(installed_required_containers)"
[[ -n "$targets" ]] || skip "none of required_containers installed on this node"
local deadline=$(( $(date +%s) + 180 )) notrunning
while (( $(date +%s) < deadline )); do
notrunning=""
for c in "${required_containers[@]}"; do
while IFS= read -r c; do
[[ "$(container_running "$c" 2>/dev/null)" == "true" ]] || notrunning="$notrunning $c"
done
done <<< "$targets"
[[ -z "$notrunning" ]] && return 0
sleep 3
done
@ -69,12 +104,14 @@ bitcoin_json() {
}
@test "bitcoin-knots RPC responds" {
skip_if_not_installed bitcoin-knots
run bitcoin_rpc
[ "$status" -eq 0 ]
echo "$output" | python3 -c 'import json,sys; r=json.load(sys.stdin)["result"]; assert r["chain"] == "main" and r["blocks"] >= 0'
}
@test "bitcoin backend is synced archival for electrumx/lnd gate" {
skip_if_not_installed bitcoin-knots
run bitcoin_rpc
[ "$status" -eq 0 ]
@ -95,6 +132,7 @@ bitcoin_json() {
}
@test "electrumx TCP port accepts connections" {
skip_if_not_installed electrumx
run python3 - <<'PY'
import socket
s = socket.create_connection(("127.0.0.1", 50001), 3)
@ -105,6 +143,7 @@ PY
}
@test "lnd CLI getinfo succeeds" {
skip_if_not_installed lnd
# lnd RPC readiness lags the container "running" state (wallet auto-unlock on
# start), so retry until ready rather than single-shot. See lnd.bats note.
run sh -lc 'for i in $(seq 1 30); do
@ -115,6 +154,7 @@ PY
}
@test "lnd REST port accepts connections" {
skip_if_not_installed lnd
run python3 - <<'PY'
import socket
s = socket.create_connection(("127.0.0.1", 18080), 3)
@ -125,17 +165,20 @@ PY
}
@test "mempool api endpoint responds" {
skip_if_not_installed mempool-api
# mempool-api reconnects to electrumx after a stack restart — retry ~180s.
run sh -lc 'for i in $(seq 1 60); do curl -fsS -m 5 -o /dev/null "http://127.0.0.1:8999/api/v1/backend-info" && exit 0; sleep 3; done; exit 1'
[ "$status" -eq 0 ]
}
@test "mempool frontend responds" {
skip_if_not_installed mempool
run sh -lc 'for i in $(seq 1 60); do curl -fsS -m 5 -o /dev/null "http://127.0.0.1:4080/" && exit 0; sleep 3; done; exit 1'
[ "$status" -eq 0 ]
}
@test "bitcoin ui responds" {
skip_if_not_installed archy-bitcoin-ui
# The companion (archy-bitcoin-ui) may have just been recreated by an earlier
# companion-survives test; its nginx takes a moment to serve. Retry ~120s
# rather than single-shot.
@ -144,11 +187,13 @@ PY
}
@test "lnd ui responds" {
skip_if_not_installed archy-lnd-ui
run curl -fsS "http://127.0.0.1:18083/"
[ "$status" -eq 0 ]
}
@test "filebrowser responds" {
skip_if_not_installed filebrowser
run curl -fsS "http://127.0.0.1:8083/"
[ "$status" -eq 0 ]
}