From f5f020d79cab9480f20907549d2eff332e1aab88 Mon Sep 17 00:00:00 2001
From: Roman Khimov <roman@nspcc.ru>
Date: Sun, 6 Aug 2023 22:32:07 +0300
Subject: [PATCH] ir: retry mainnet RPC connection if the node is outdated

We can have a set of RPC nodes configured and we will switch them in runtime,
but if the first node is outdated then the IR node will _always_ fail to
start until RPC node catches up. We can have a better behaviour with node
switching to other RPCs in this phase.

Can't be done this easily for the sidechain since in most cases we have an
internal node and it needs to go through the chain completely.

Refs. #2426.

Signed-off-by: Roman Khimov <roman@nspcc.ru>
---
 CHANGELOG.md                       | 1 +
 pkg/innerring/innerring.go         | 9 +++++++--
 pkg/morph/subscriber/subscriber.go | 3 ++-
 3 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index f8b6134b9d4..41c174021cd 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -16,6 +16,7 @@ Changelog for NeoFS Node
 - Double voting for validators on IR startup (#2365)
 - Skip unexpected notary events on notary request parsing step (#2315)
 - Session inactivity on object PUT request relay (#2460)
+- Missing connection retries on IR node startup whe the first configured mainner RPC node is not in sync (#XXX)
 
 ### Removed
 - Deprecated `morph.rpc_endpoint` SN and `morph.endpoint.client` IR config sections (#2400)
diff --git a/pkg/innerring/innerring.go b/pkg/innerring/innerring.go
index e4689972598..6195cfd3ce7 100644
--- a/pkg/innerring/innerring.go
+++ b/pkg/innerring/innerring.go
@@ -479,8 +479,13 @@ func New(ctx context.Context, log *zap.Logger, cfg *viper.Viper, errChan chan<-
 			return nil, err
 		}
 
-		// create mainnet listener
-		server.mainnetListener, err = createListener(ctx, server.mainnetClient, mainnetChain)
+		// create mainnet listener, retry with a different node if current one is not up to date
+		for {
+			server.mainnetListener, err = createListener(ctx, server.mainnetClient, mainnetChain)
+			if !errors.Is(err, subscriber.ErrStaleNode) || !server.mainnetClient.SwitchRPC() {
+				break
+			}
+		}
 		if err != nil {
 			return nil, err
 		}
diff --git a/pkg/morph/subscriber/subscriber.go b/pkg/morph/subscriber/subscriber.go
index 21f91d2fb95..5cf81dc307e 100644
--- a/pkg/morph/subscriber/subscriber.go
+++ b/pkg/morph/subscriber/subscriber.go
@@ -68,6 +68,7 @@ func (s *subscriber) NotificationChannels() NotificationChannels {
 }
 
 var (
+	ErrStaleNode = errors.New("RPC node is not yet up to date")
 	errNilParams = errors.New("chain/subscriber: config was not provided to the constructor")
 
 	errNilLogger = errors.New("chain/subscriber: logger was not provided to the constructor")
@@ -333,7 +334,7 @@ func awaitHeight(cli *client.Client, startFrom uint32) error {
 	}
 
 	if height < startFrom+1 {
-		return fmt.Errorf("RPC block counter %d didn't reach expected height %d", height, startFrom)
+		return fmt.Errorf("%w: expected %d height, got %d count", ErrStaleNode, startFrom, height)
 	}
 
 	return nil