From a971fa19897d5861b1a28bb9d7bdc480c1b02da7 Mon Sep 17 00:00:00 2001
From: NightCrawler <nightcrawler@lumera.ai>
Date: Tue, 10 Mar 2026 13:22:44 +0000
Subject: [PATCH] fix: re-enable legacy metrics reporter for audit bootstrap
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

The audit module's epoch-end recovery requires peer observations from
active probers. When the module was first activated on testnet, all
supernodes running v2.4.5-testnet had already been POSTPONED by the
legacy staleness handler (they stopped submitting MsgReportSupernodeMetrics
~500 blocks after upgrading, before the chain upgrade).

This created a deadlock:
- Recovery needs peer observations from active probers
- No active probers exist (empty active_supernode_accounts in every anchor)
- POSTPONED SNs submit epoch reports but cannot recover
- The 3 SNs on old releases bounce ACTIVE↔POSTPONED via legacy metrics
  but are always POSTPONED at epoch start (anchor freeze time)

Fix: run the legacy metrics reporter alongside the audit host_reporter.
Legacy MsgReportSupernodeMetrics recovers POSTPONED SNs to ACTIVE
mid-epoch. Since they also submit audit epoch reports, the audit
EndBlocker won't re-postpone them (report exists, host minimums are
disabled, no peer-port streak). They survive the epoch end and appear
ACTIVE in the next epoch anchor, bootstrapping the peer-observation
cycle for all remaining POSTPONED SNs.

Once the active set stabilizes, the legacy reporter can be removed
in a future release.
---
 supernode/cmd/start.go | 34 ++++++++++++++++++----------------
 1 file changed, 18 insertions(+), 16 deletions(-)

diff --git a/supernode/cmd/start.go b/supernode/cmd/start.go
index be1c1c15..2ed3a04c 100644
--- a/supernode/cmd/start.go
+++ b/supernode/cmd/start.go
@@ -26,9 +26,7 @@ import (
 	hostReporterService "github.com/LumeraProtocol/supernode/v2/supernode/host_reporter"
 	statusService "github.com/LumeraProtocol/supernode/v2/supernode/status"
 	storageChallengeService "github.com/LumeraProtocol/supernode/v2/supernode/storage_challenge"
-	// Legacy supernode metrics reporter (MsgReportSupernodeMetrics) has been superseded by
-	// epoch-scoped audit reporting in `x/audit`.
-	// supernodeMetrics "github.com/LumeraProtocol/supernode/v2/supernode/supernode_metrics"
+	supernodeMetrics "github.com/LumeraProtocol/supernode/v2/supernode/supernode_metrics"
 	"github.com/LumeraProtocol/supernode/v2/supernode/transport/gateway"
 	cascadeRPC "github.com/LumeraProtocol/supernode/v2/supernode/transport/grpc/cascade"
 	server "github.com/LumeraProtocol/supernode/v2/supernode/transport/grpc/status"
@@ -173,18 +171,22 @@ The supernode will connect to the Lumera network and begin participating in the
 			logtrace.Fatal(ctx, "Failed to initialize host reporter", logtrace.Fields{"error": err.Error()})
 		}
 
-		// Legacy on-chain supernode metrics reporting has been superseded by `x/audit`.
-		// metricsCollector := supernodeMetrics.NewCollector(
-		// 	statusSvc,
-		// 	lumeraClient,
-		// 	appConfig.SupernodeConfig.Identity,
-		// 	Version,
-		// 	kr,
-		// 	appConfig.SupernodeConfig.Port,
-		// 	appConfig.P2PConfig.Port,
-		// 	appConfig.SupernodeConfig.GatewayPort,
-		// )
-		// logtrace.Info(ctx, "Metrics collection enabled", logtrace.Fields{})
+		// Legacy on-chain supernode metrics reporting (MsgReportSupernodeMetrics)
+		// runs alongside the audit epoch reporter. It is needed to recover
+		// POSTPONED supernodes via the supernode module's instant-recovery
+		// path so they appear ACTIVE in the next epoch anchor — which
+		// bootstraps the audit peer-observation cycle.
+		metricsCollector := supernodeMetrics.NewCollector(
+			statusSvc,
+			lumeraClient,
+			appConfig.SupernodeConfig.Identity,
+			Version,
+			kr,
+			appConfig.SupernodeConfig.Port,
+			appConfig.P2PConfig.Port,
+			appConfig.SupernodeConfig.GatewayPort,
+		)
+		logtrace.Info(ctx, "Legacy metrics collection enabled (audit bootstrap)", logtrace.Fields{})
 
 		// Storage challenge history DB (shared by the gRPC handler and runner).
 		historyStore, err := queries.OpenHistoryDB()
@@ -253,7 +255,7 @@ The supernode will connect to the Lumera network and begin participating in the
 		// Start the services using the standard runner and capture exit
 		servicesErr := make(chan error, 1)
 		go func() {
-			services := []service{grpcServer, cService, p2pService, gatewayServer, hostReporter}
+			services := []service{grpcServer, cService, p2pService, gatewayServer, hostReporter, metricsCollector}
 			if storageChallengeRunner != nil {
 				services = append(services, storageChallengeRunner)
 			}