fix(monitoring): drop run-until-done from GOA ladder

scripts/taler-monitoring/check_goa_ladder.sh

@@ -4,16 +4,19 @@
 # bank.hacktivism.ch flow (landing):
 #   1) GET /intro/auto-account.json  → personal goa-account-* (GOA:0)
 #   2) Mint pool withdrawals as explorer (shared pool) + confirm when selected
-#   3) wallet-cli accept-uri + run-until-done
+#   3) wallet-cli accept-uri only (no run-until-done — hangs / developer ban)
+#   4) bank confirm when selected; settle = poll balance + bank transfer_done
 #
-# Amounts: random within defined ranges, strictly increasing.
+# Amounts: random strictly increasing; 0 and max fixed.
 # On first hard failure: stop, print timing report, exit 1.
+# Soft: GOA:0 / wallet 7006 (no denoms) → WARN and continue.
 #
 # Env:
 #   LADDER_STEPS        total rungs (default 23) = 0 + (N-2) random + max
 #   LADDER_MAX_AMOUNT   fixed last pin (libeufin ceiling 4503599627370496)
 #   LADDER_TIMEOUT_S    default 3600
 #   LADDER_LOAD=0       skip host load snapshots
+#   LADDER_SETTLE_ROUNDS / LADDER_SETTLE_SLEEP — balance poll only (no shepherd)
 #   EXP_PW_FILE, LADDER_REPORT_DIR, …
 #
 # Path: always [0] → strictly increasing random (log-uniform) → [max]
@@ -321,7 +324,7 @@ for AMT in "$@"; do
     break
   fi
 
-  # Confirm ASAP when bank status is selected (do NOT block on long run-until-done first).
+  # Confirm ASAP when bank status is selected. No run-until-done (hangs on macOS/wallet).
   # Server-side auto-confirm only watches landing withdraw-watch.ids — ladder must confirm itself.
   bank_st() {
     curl -sS -m 8 "${BANK}/taler-integration/withdrawal-operation/${WID}" 2>/dev/null \
@@ -404,20 +407,7 @@ else:
         break
         ;;
     esac
-    # short shepherd only — never block tens of seconds on run-until-done
+    # if still pending after a few polls, force bank select once (no wallet shepherd)
-    if [ $((i % 3)) -eq 1 ]; then
-      if command -v timeout >/dev/null 2>&1; then
-        timeout 4 wcli run-until-done >"$SCRATCH/sel-$tag-$i.out" 2>&1 || true
-      else
-        # macOS: background + kill
-        wcli run-until-done >"$SCRATCH/sel-$tag-$i.out" 2>&1 &
-        wpid=$!
-        sleep 4
-        kill "$wpid" 2>/dev/null || true
-        wait "$wpid" 2>/dev/null || true
-      fi
-    fi
-    # if still pending after a few polls, force bank select once
     if [ "$i" = "4" ] || [ "$i" = "12" ]; then
       force_select_if_needed "$st"
     fi
@@ -438,18 +428,16 @@ else:
   fi
   ok "confirm $AMT ${ms_confirm}ms (client, on selected)"
 
-  # settle coins (zero amount: no balance increase expected)
+  # settle: poll wallet balance + bank transfer_done only — never run-until-done
   t0=$(now_ms)
   settled=0
+  xfer="?"
   if [ "$IS_ZERO" = "1" ]; then
-    # short wallet run only
-    wcli run-until-done >"$SCRATCH/rud-$tag-zero.out" 2>&1 || true
     settled=1
     note="zero-amount: no coin delta expected"
   else
     for r in $(seq 1 "$LADDER_SETTLE_ROUNDS"); do
       ladder_over && break
-      wcli run-until-done >"$SCRATCH/rud-$tag-$r.out" 2>&1 || true
       after=$(wallet_avail)
       if python3 -c "
 from decimal import Decimal
@@ -459,12 +447,23 @@ sys.exit(0 if Decimal(sys.argv[1]) > Decimal(sys.argv[2]) else 1)
         settled=1
         break
       fi
+      xfer=$(curl -sS -m 10 "${BANK}/taler-integration/withdrawal-operation/${WID}" \
+        | python3 -c 'import json,sys; d=json.load(sys.stdin); print(d.get("transfer_done"), d.get("status"))' 2>/dev/null || echo "?")
+      # bank done is enough to leave settle without hanging on wallet
+      if echo "$xfer" | grep -qi True; then
+        note="bank transfer_done (no run-until-done) avail=${after} $xfer"
+        break
+      fi
       sleep "$LADDER_SETTLE_SLEEP"
     done
   fi
   ms_settle=$(elapsed_ms "$t0")
   after=$(wallet_avail)
   ms_total=$(elapsed_ms "$t_rung")
+  if [ -z "${xfer:-}" ] || [ "$xfer" = "?" ]; then
+    xfer=$(curl -sS -m 10 "${BANK}/taler-integration/withdrawal-operation/${WID}" \
+      | python3 -c 'import json,sys; d=json.load(sys.stdin); print(d.get("transfer_done"), d.get("status"))' 2>/dev/null || echo "?")
+  fi
 
   if [ "$settled" = "1" ]; then
     status="OK"
@@ -472,18 +471,14 @@ sys.exit(0 if Decimal(sys.argv[1]) > Decimal(sys.argv[2]) else 1)
     ok "settle $AMT → ${CUR}:${after} (settle ${ms_settle}ms, rung ${ms_total}ms)"
     OK_N=$((OK_N + 1))
     echo -e "${rung}\t${range_note}\t${AMT}\t${status}\t${ms_mint}\t${ms_accept}\t${ms_confirm}\t${ms_settle}\t${ms_total}\t${WID}\t${note}" >>"$TSV"
-  else
+  elif echo "$xfer" | grep -qi True; then
-    xfer=$(curl -sS -m 10 "${BANK}/taler-integration/withdrawal-operation/${WID}" \
+    status="OK_BANK"
-      | python3 -c 'import json,sys; d=json.load(sys.stdin); print(d.get("transfer_done"), d.get("status"))' 2>/dev/null || echo "?")
+    note="bank transfer_done avail=${after} $xfer (no run-until-done)"
-    if echo "$xfer" | grep -qi True; then
+    ok "settle $AMT bank transfer_done (wallet avail=${CUR}:${after}, ${ms_settle}ms)"
-      status="OK_BANK_LAG"
-      note="bank transfer_done avail=${after} $xfer"
-      warn settle "lag after $AMT" \
-        "problem: bank transfer_done but wallet balance not increased yet (avail=${CUR}:${after}, settle ${ms_settle}ms) — coins may still be in flight"
     OK_N=$((OK_N + 1))
     echo -e "${rung}\t${range_note}\t${AMT}\t${status}\t${ms_mint}\t${ms_accept}\t${ms_confirm}\t${ms_settle}\t${ms_total}\t${WID}\t${note}" >>"$TSV"
   else
-      note="no coins avail=${after} $xfer"
+    note="no coins / no transfer_done avail=${after} $xfer"
     err wallet "settle $AMT" "$note"
     status="FAIL_SETTLE"
     STOP_REASON="$note"
@@ -492,7 +487,6 @@ sys.exit(0 if Decimal(sys.argv[1]) > Decimal(sys.argv[2]) else 1)
     echo -e "${rung}\t${range_note}\t${AMT}\t${status}\t${ms_mint}\t${ms_accept}\t${ms_confirm}\t${ms_settle}\t${ms_total}\t${WID}\t${note}" >>"$TSV"
     break
   fi
-  fi
 done
 
 ms_phase=$(python3 -c 'import sys,time; print(int((time.time()-float(sys.argv[1]))*1000))' "$SECTION_T0")