Bootstrap fixes

settings.ini

@@ -1,40 +1,42 @@
 ; settings.ini
 
 [Paths]
-BLOCK_PATH             = "./blocks"
-TORRENT_PATH           = "./torrents"
-DB_PATH                = "./state"
-BALANCE_SHEET          = "./balance_sheet"
-LOG_PATH               = "./logs"
-WALLET_PATH            = "./wallets"
-WALLET_NAME            = "contractless.wallet"
+BLOCK_PATH                  = "./blocks"
+TORRENT_PATH                = "./torrents"
+DB_PATH                     = "./state"
+BALANCE_SHEET               = "./balance_sheet"
+LOG_PATH                    = "./logs"
+WALLET_PATH                 = "./wallets"
+WALLET_NAME                 = "contractless.wallet"
 
 
 [Settings]
-LOG_LEVEL              = "info"
-PUBLIC_IP              = "your_public_ip_address"
-LISTEN_IP              = "0.0.0.0"
-RPC_PORT               = "50055"
-TESTNET_RPC_PORT       = "50050"
-INCOMING_CONNECTIONS   = "100"
-OUTGOING_CONNECTIONS   = "10"
-VALIDATOR              = "false"
-; THREADS must 1, 2, or a multple of 4
-THREADS                = "8"
+LOG_LEVEL                   = "info"
+PUBLIC_IP                   = "your_public_ip_address"
+LISTEN_IP                   = "0.0.0.0"
+RPC_PORT                    = "50055"
+TESTNET_RPC_PORT            = "50050"
+INCOMING_CONNECTIONS        = "100"
+OUTGOING_CONNECTIONS        = "10"
+VALIDATOR                   = "false"
+; THREADS must 1, 2, 
+; or a multple of 4
+THREADS                     = "8"
+STORAGE_LOOKUP_FEE_PER_BYTE = "0.0001"
 
 [Piggyback]
-PIGGYBACK_1            = "contractless.dev:50050"
+PIGGYBACK_1                 = "contractless.dev:50050"
 
 [Postgres-Testnet]
-host                   = 127.0.0.1
-port                   = 5432
-user                   = contractless2
-password               = your_paddword
-dbname                 = contractless_db2
+host                        = 127.0.0.1
+port                        = 5432
+user                        = contractless2
+password                    = your_paddword
+dbname                      = contractless_db2
 
 [Postgres]
-host                   = 127.0.0.1
-port                   = 5432
-user                   = contractless
-password               = your_password
-dbname                 = contractless_db
+host                        = 127.0.0.1
+port                        = 5432
+user                        = contractless
+password                    = your_password
+dbname                      = contractless_db

src/bin/create_burn_tx.rs

@@ -1,6 +1,8 @@
 use contractless::blocks::burn::{BurnTransaction, UnsignedBurnTransaction};
 use contractless::common::asset_names::padded_normalized_asset_input;
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::network_paths_and_settings::block_extension_and_paths;
 
 use contractless::common::types::BURN_FEE;

src/bin/create_collateral_claim_tx.rs

@@ -1,7 +1,9 @@
 use contractless::blocks::collateral::{
     CollateralClaimTransaction, UnsignedCollateralClaimTransaction,
 };
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::types::COLLATERAL_FEE;
 
 use contractless::json;

src/bin/create_issue_token_tx.rs

@@ -1,6 +1,8 @@
 use contractless::blocks::issue_token::{IssueTokenTransaction, UnsignedIssueTokenTransaction};
 use contractless::common::asset_names::padded_normalized_asset_input;
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::types::ISSUE_TOKEN_FEE;
 
 use contractless::json;

src/bin/create_loan_payment_tx.rs

@@ -1,7 +1,9 @@
 use contractless::blocks::loan_payment::{
     ContractPaymentTransaction, UnsignedContractPaymentTransaction,
 };
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::types::BORROWER_FEE;
 
 use contractless::json;

src/bin/create_loan_tx.rs

@@ -2,7 +2,9 @@ use contractless::blocks::loans::UnsignedLoanContractTransaction;
 use contractless::common::asset_names::{
     padded_loan_collateral_input_or_base, padded_normalized_asset_input,
 };
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::network_paths_and_settings::block_extension_and_paths;
 use contractless::common::types::LENDER_FEE;
 

src/bin/create_marketing_tx.rs

@@ -1,5 +1,7 @@
 use contractless::blocks::marketing::{MarketingTransaction, UnsignedMarketingTransaction};
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::types::MARKETING_FEE;
 
 use contractless::json;

src/bin/create_nft_tx.rs

@@ -1,6 +1,8 @@
 use contractless::blocks::nft::{CreateNftTransaction, UnsignedCreateNftTransaction};
 use contractless::common::asset_names::padded_normalized_asset_input;
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::types::CREATE_NFT_FEE;
 
 use contractless::json;

src/bin/create_swap_tx.rs

@@ -1,6 +1,8 @@
 use contractless::blocks::swap::UnsignedSwapTransaction;
 use contractless::common::asset_names::padded_normalized_asset_input;
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::network_paths_and_settings::block_extension_and_paths;
 use contractless::common::types::SWAP_FEE;
 

src/bin/create_tokens_tx.rs

@@ -1,6 +1,8 @@
 use contractless::blocks::token::{CreateTokenTransaction, UnsignedCreateTokenTransaction};
 use contractless::common::asset_names::padded_normalized_asset_input;
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::types::CREATE_TOKEN_FEE;
 
 use contractless::json;

src/bin/create_transfer_tx.rs

@@ -1,6 +1,8 @@
 use contractless::blocks::transfer::{TransferTransaction, UnsignedTransferTransaction};
 use contractless::common::asset_names::padded_normalized_asset_input;
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::network_paths_and_settings::block_extension_and_paths;
 
 use contractless::common::types::{NON_BASE_TRANSFER_MIN_FEE, TRANSFER_FEE};

src/bin/create_vanity_tx.rs

@@ -1,5 +1,7 @@
 use contractless::blocks::vanity::{UnsignedVanityAddressTransaction, VanityAddressTransaction};
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::types::{VANITY_ADDRESS_FEE, VANITY_ADDRESS_TYPE};
 
 use contractless::env;

src/bin/validate_torrent_and_block_headers.rs

@@ -1,3 +1,4 @@
+use colored::*;
 use contractless::common::binary_conversions::hex_to_u64;
 use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_wallet_path};
 use contractless::common::network_paths_and_settings::block_extension_and_paths;
@@ -9,7 +10,6 @@ use contractless::standalone_tools::wallet_registry_lookup::lookup_pubkey_from_l
 use contractless::torrent::structs::Torrent;
 use contractless::wallets::structures::Wallet;
 use contractless::{AsyncReadExt, File};
-use colored::*;
 use std::process;
 
 #[tokio::main]

src/bin/verify_message.rs

@@ -1,4 +1,6 @@
-use contractless::common::cli_prompts::{prompt_hidden_nonempty, prompt_visible, prompt_wallet_path};
+use contractless::common::cli_prompts::{
+    prompt_hidden_nonempty, prompt_visible, prompt_wallet_path,
+};
 use contractless::common::skein::skein_256_hash_data;
 use contractless::env;
 use contractless::standalone_tools::wallet_registry_lookup::lookup_pubkey_from_live_registry;

src/miner/flag.rs

@@ -3,7 +3,7 @@ use crate::Duration;
 use crate::{AtomicBool, AtomicOrdering};
 use std::sync::atomic::AtomicU8;
 
-static REORG_GATE: AtomicBool = AtomicBool::new(false);
+static CHAIN_OPERATION_GATE: AtomicBool = AtomicBool::new(false);
 static MINING_STOP_REQUESTED: AtomicBool = AtomicBool::new(false);
 static NODE_MODE: AtomicU8 = AtomicU8::new(NodeMode::Startup as u8);
 static MINING_STATE: AtomicU8 = AtomicU8::new(MiningState::Idle as u8);
@@ -108,35 +108,114 @@ pub async fn wait_for_mining_idle() {
     }
 }
 
-fn try_acquire_reorg_gate() -> bool {
-    // Only one reorg path may own the gate at a time.
-    REORG_GATE
+fn try_acquire_chain_operation_gate() -> bool {
+    CHAIN_OPERATION_GATE
         .compare_exchange(false, true, AtomicOrdering::SeqCst, AtomicOrdering::SeqCst)
         .is_ok()
 }
 
-pub async fn begin_reorg_lock() {
-    // Reorg starts by asking mining to stop and waiting for the
-    // current worker round to leave the save path.
-    request_mining_stop();
-    wait_for_mining_idle().await;
-    if is_reorganizing_mode() {
-        return;
-    }
-    // Wait for any competing reorg attempt to release the gate.
-    while !try_acquire_reorg_gate() {
+async fn acquire_chain_operation_gate() {
+    while !try_acquire_chain_operation_gate() {
         sleep(Duration::from_millis(5)).await;
     }
-    // Reorganizing mode blocks new mining rounds until the lock ends.
-    set_node_mode(NodeMode::Reorganizing);
 }
 
-pub fn end_reorg_lock() {
-    // Restore normal mode only if this task still owns the reorg state.
-    if is_reorganizing_mode() {
-        set_node_mode(NodeMode::Normal);
-    }
-    // Clear the stop flag last so mining cannot resume before mode is restored.
+fn release_chain_operation(mode: NodeMode) {
+    set_node_mode(mode);
     clear_mining_stop_request();
-    REORG_GATE.store(false, AtomicOrdering::SeqCst);
+    CHAIN_OPERATION_GATE.store(false, AtomicOrdering::SeqCst);
+}
+
+// Owns this process's canonical chain mutation slot for one synchronization
+// operation. Torrent metadata and pieces may still download concurrently;
+// this guard only prevents overlapping save/replay and mode transitions.
+pub struct ChainOperationGuard {
+    active: bool,
+}
+
+impl ChainOperationGuard {
+    pub fn finish(mut self) {
+        if self.active {
+            release_chain_operation(NodeMode::Normal);
+            self.active = false;
+        }
+    }
+}
+
+impl Drop for ChainOperationGuard {
+    fn drop(&mut self) {
+        if self.active {
+            release_chain_operation(NodeMode::Normal);
+            self.active = false;
+        }
+    }
+}
+
+pub async fn begin_chain_sync() -> ChainOperationGuard {
+    acquire_chain_operation_gate().await;
+    request_mining_stop();
+    wait_for_mining_idle().await;
+    set_node_mode(NodeMode::Syncing);
+    ChainOperationGuard { active: true }
+}
+
+pub async fn begin_reorg_lock() -> ChainOperationGuard {
+    // Orphan correction shares the canonical chain gate with synchronization,
+    // so rollback/replay cannot overlap a startup or live catch-up save.
+    acquire_chain_operation_gate().await;
+    request_mining_stop();
+    wait_for_mining_idle().await;
+    set_node_mode(NodeMode::Reorganizing);
+    ChainOperationGuard { active: true }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use tokio::sync::oneshot;
+
+    #[tokio::test]
+    async fn canonical_chain_operations_wait_and_release_on_drop() {
+        CHAIN_OPERATION_GATE.store(false, AtomicOrdering::SeqCst);
+        set_mining_state(MiningState::Idle);
+        clear_mining_stop_request();
+        set_node_mode(NodeMode::Normal);
+
+        let first = begin_chain_sync().await;
+        assert!(is_syncing_mode());
+
+        let (started_tx, mut started_rx) = oneshot::channel();
+        let waiter = tokio::spawn(async move {
+            let second = begin_reorg_lock().await;
+            let _ = started_tx.send(());
+            second.finish();
+        });
+
+        sleep(Duration::from_millis(20)).await;
+        assert!(matches!(
+            started_rx.try_recv(),
+            Err(oneshot::error::TryRecvError::Empty)
+        ));
+
+        drop(first);
+        crate::timeout(Duration::from_secs(1), &mut started_rx)
+            .await
+            .expect("waiting chain operation did not acquire the released gate")
+            .expect("waiting chain operation exited before acquiring the gate");
+        waiter.await.expect("waiting chain operation task failed");
+
+        assert!(is_normal_mode());
+        assert!(!is_mining_stop_requested());
+        assert!(!CHAIN_OPERATION_GATE.load(AtomicOrdering::SeqCst));
+
+        async fn guarded_error() -> Result<(), &'static str> {
+            let _guard = begin_chain_sync().await;
+            Err("expected test error")
+        }
+
+        assert_eq!(guarded_error().await, Err("expected test error"));
+        assert!(is_normal_mode());
+        assert!(!is_mining_stop_requested());
+        assert!(!CHAIN_OPERATION_GATE.load(AtomicOrdering::SeqCst));
+    }
 }

src/orphans/deep_sync_rollback.rs

@@ -1,5 +1,4 @@
 use crate::log::warn;
-use crate::miner::flag::begin_reorg_lock;
 use crate::orphans::structs::{CheckUp, UndoTransactions};
 use crate::orphans::undo_block_transactions::undo_transactions;
 use crate::torrent::unpack_local_torrent::load_torrent;
@@ -35,9 +34,6 @@ pub async fn deep_sync_rollback(mut params: CheckUp, wallet: Arc<Wallet>) {
                                     break;
                                 }
 
-                                if !params.node_syncing {
-                                    begin_reorg_lock().await;
-                                }
                                 // Undo one height, then repeat the comparison
                                 // against the next lower local height.
                                 let undo_transactions_params = UndoTransactions {

src/orphans/orphan_checkup.rs

@@ -1,6 +1,5 @@
 use crate::common::skein::skein_128_hash_bytes;
 use crate::log::{info, warn};
-use crate::miner::flag::begin_reorg_lock;
 use crate::orphans::replay_errors::staged_candidate_status_for_error;
 use crate::orphans::structs::{OrphanCheckup, UndoTransactions};
 use crate::orphans::undo_block_transactions::undo_transactions;
@@ -223,9 +222,6 @@ pub async fn checkup(params: OrphanCheckup, wallet: Arc<Wallet>) -> Result<(), S
                     };
 
                     set_torrent_status(height, &competing_info_hash, TorrentStatus::Valid).await;
-                    if !params.node_syncing {
-                        begin_reorg_lock().await;
-                    }
                     info!("[orphan] adopting proven staged chain from height {height}");
                     undo_transactions(undo_transactions_params, wallet.clone()).await?;
                     return Ok(());

src/orphans/snapshot_check.rs

@@ -1,6 +1,6 @@
 use crate::common::binary_conversions::binary_to_string;
 use crate::log::{error, info, warn};
-use crate::miner::flag::{begin_reorg_lock, is_syncing_mode};
+use crate::miner::flag::is_syncing_mode;
 use crate::orphans::structs::UndoTransactions;
 use crate::orphans::undo_block_transactions::undo_transactions;
 use crate::records::memory::connections::live_miner_peer_streams;
@@ -183,9 +183,6 @@ pub async fn snapshot_verified(params: UndoTransactions, wallet: Arc<Wallet>) ->
                     if local_hash != snap_hash {
                         // Local state no longer matches the trusted checkpoint,
                         // so rollback starts at the snapshot height.
-                        if !params.node_syncing {
-                            begin_reorg_lock().await;
-                        }
                         let undo_transactions_params = UndoTransactions {
                             start_height: snap_height,
                             replay_to_height: params.replay_to_height,

src/orphans/sync_check.rs

@@ -1,6 +1,6 @@
 use crate::common::check_genesis::genesis_checkup;
 use crate::log::{error, info, warn};
-use crate::miner::flag::end_reorg_lock;
+use crate::miner::flag::begin_reorg_lock;
 use crate::orphans::add_genesis::create_genesis_block;
 use crate::orphans::checkup_state::take_orphan_recheck_height;
 use crate::orphans::deep_sync_rollback::deep_sync_rollback;
@@ -280,6 +280,12 @@ async fn sync_checkup_pass(params: &OrphanCheckup2, wallet: Arc<Wallet>) -> Resu
 }
 
 pub async fn sync_checkup(mut params: OrphanCheckup2, wallet: Arc<Wallet>) -> Result<(), String> {
+    let reorg_guard = if params.node_syncing {
+        None
+    } else {
+        Some(begin_reorg_lock().await)
+    };
+
     let result = loop {
         match sync_checkup_pass(&params, wallet.clone()).await {
             Ok(()) => {}
@@ -315,8 +321,8 @@ pub async fn sync_checkup(mut params: OrphanCheckup2, wallet: Arc<Wallet>) -> Re
         );
     };
 
-    if !params.node_syncing {
-        end_reorg_lock();
+    if let Some(guard) = reorg_guard {
+        guard.finish();
     }
 
     if result.is_ok() {

src/records/memory/connections.rs

@@ -1,7 +1,6 @@
 use crate::common::binary_conversions::{binary_to_ip, ip_to_binary};
 use crate::lazy_static;
 use crate::log::{info, warn};
-use crate::miner::flag::is_normal_mode;
 use crate::records::memory::enums::{ClientType, ConnectionType};
 use crate::records::memory::network_mapping::monitor::{MONITOR_ACTION_ADD, MONITOR_ACTION_REMOVE};
 use crate::records::memory::network_mapping::structs::{MonitorAddressParams, SignedMonitorEdit};
@@ -9,11 +8,11 @@ use crate::records::memory::network_mapping::NodeInfo;
 use crate::records::memory::response_channels::{
     delete_entry, reserve_entry_with_context, Command,
 };
-use crate::records::memory::structs::{Connection, StoreConnectionParams};
+use crate::records::memory::structs::{Connection, ConnectionHealth, StoreConnectionParams};
 use crate::rpc::client::handshake::connect_and_handshake;
 use crate::rpc::client::handshake_processing::{bootstrap_peer_discovery, BootstrapParams};
 use crate::rpc::client::structs::Connect;
-use crate::rpc::command_maps::RPC_BLOCK_HEIGHT;
+use crate::rpc::command_maps::{RPC_BLOCK_HEIGHT, RPC_SETUP_COMPLETE};
 use crate::rpc::responses::RpcResponse;
 use crate::sled::Db;
 use crate::sleep;
@@ -22,13 +21,13 @@ use crate::timeout;
 use crate::wallets::structures::Wallet;
 use crate::Arc;
 use crate::AsyncWriteExt;
-use crate::AtomicBool;
-use crate::AtomicOrdering;
 use crate::Duration;
 use crate::IteratorRandom;
 use crate::Mutex;
 use crate::RwLock;
 use crate::TcpStream;
+use std::collections::HashSet;
+use std::sync::{Mutex as StdMutex, RwLock as StdRwLock};
 
 fn split_ip_port_key(value: &str) -> Option<(String, u16)> {
     // Connection keys are stored as ip:port strings; IPv6 addresses may arrive
@@ -46,47 +45,70 @@ fn split_ip_port_key(value: &str) -> Option<(String, u16)> {
 use crate::records::memory::structs::{ConnectionInfo, ConnectionKey};
 
 #[derive(Clone)]
-struct ReconnectContext {
+struct NodeRuntimeContext {
+    // These dependencies belong to the running node, not to any individual
+    // outgoing handshake or reconnect attempt.
     db: Db,
     wallet: Arc<Wallet>,
     map: Arc<Mutex<Command>>,
 }
 
 lazy_static! {
-    static ref RECONNECT_CONTEXT: Mutex<Option<ReconnectContext>> = Mutex::new(None);
-    static ref RECONNECT_IN_PROGRESS: AtomicBool = AtomicBool::new(false);
+    static ref RECOVERY_IN_PROGRESS: StdMutex<HashSet<String>> = StdMutex::new(HashSet::new());
+    static ref NODE_RUNTIME_CONTEXT: StdRwLock<Option<NodeRuntimeContext>> = StdRwLock::new(None);
 }
 
-fn try_start_reconnect() -> bool {
-    // Only one reconnect path should run at a time, whether it came from
-    // liveness failure or bootstrap recovery.
-    RECONNECT_IN_PROGRESS
-        .compare_exchange(false, true, AtomicOrdering::SeqCst, AtomicOrdering::SeqCst)
-        .is_ok()
+struct RecoveryGuard {
+    key: String,
 }
 
-fn finish_reconnect() {
-    // Release the reconnect gate after the async reconnect attempt finishes.
-    RECONNECT_IN_PROGRESS.store(false, AtomicOrdering::SeqCst);
+impl Drop for RecoveryGuard {
+    fn drop(&mut self) {
+        if let Ok(mut active) = RECOVERY_IN_PROGRESS.lock() {
+            active.remove(&self.key);
+        }
+    }
 }
 
-pub async fn set_reconnect_context(db: Db, wallet: Arc<Wallet>, map: Arc<Mutex<Command>>) {
-    let mut context = RECONNECT_CONTEXT.lock().await;
-    // Store enough state for later liveness checks to reconnect without
-    // needing the original startup stack.
-    *context = Some(ReconnectContext { db, wallet, map });
+fn try_start_recovery(key: String) -> Option<RecoveryGuard> {
+    // Recovery is deduplicated per peer. One dead endpoint can no longer
+    // suppress retries for every other endpoint for the duration of its backoff.
+    let mut active = RECOVERY_IN_PROGRESS.lock().ok()?;
+    if !active.insert(key.clone()) {
+        return None;
+    }
+    Some(RecoveryGuard { key })
+}
+
+pub fn initialize_node_runtime_context(
+    db: Db,
+    wallet: Arc<Wallet>,
+    map: Arc<Mutex<Command>>,
+) -> Result<(), String> {
+    // Replace the context when the Windows service relaunches an unlocked node
+    // task without restarting the service process.
+    let mut context = NODE_RUNTIME_CONTEXT
+        .write()
+        .map_err(|_| "node runtime context lock was poisoned".to_string())?;
+    *context = Some(NodeRuntimeContext { db, wallet, map });
+    Ok(())
+}
+
+fn node_runtime_context() -> Option<NodeRuntimeContext> {
+    NODE_RUNTIME_CONTEXT
+        .read()
+        .ok()
+        .and_then(|context| context.clone())
 }
 
 async fn reconnect_replacement_inner(excluded_ip: &str) {
     // When an outgoing peer disappears, try to replace it with another
     // active node that is not already connected and is not the failed IP.
-    let context = {
-        let guard = RECONNECT_CONTEXT.lock().await;
-        guard.clone()
+    let Some(_topology_guard) = try_start_recovery("topology".to_string()) else {
+        return;
     };
-
-    let Some(context) = context else {
-        warn!("[reconnect] no reconnect context configured");
+    let Some(context) = node_runtime_context() else {
+        warn!("[reconnect] node runtime context is not initialized");
         return;
     };
 
@@ -132,77 +154,75 @@ async fn reconnect_replacement_inner(excluded_ip: &str) {
 }
 
 async fn retry_dropped_outgoing(ip: String, port: u16) {
-    if !try_start_reconnect() {
-        warn!("[reconnect] reconnect attempt already in progress, skipping duplicate request");
+    let recovery_key = format!("peer:{ip}");
+    let Some(_recovery_guard) = try_start_recovery(recovery_key) else {
+        warn!("[reconnect] reconnect attempt already in progress for peer {ip}");
         return;
-    }
+    };
 
-    async {
-        let context = {
-            let guard = RECONNECT_CONTEXT.lock().await;
-            guard.clone()
-        };
+    let Some(context) = node_runtime_context() else {
+        warn!("[reconnect] node runtime context is not initialized");
+        return;
+    };
 
-        let Some(context) = context else {
-            warn!("[reconnect] no reconnect context configured");
-            return;
-        };
-
-        let addr_string = format!("{ip}:{port}");
-        for attempt in 1..=3 {
+    let addr_string = format!("{ip}:{port}");
+    for attempt in 1..=3 {
+        // A connection-manager drop is definitive, so the first reconnect is
+        // immediate. Backoff applies only after that direct attempt fails.
+        if attempt > 1 {
             sleep(Duration::from_secs(30)).await;
-
-            if Connection::get_stream_from_memory(&addr_string).await.is_some() {
-                info!(
-                    "[reconnect] dropped peer {addr_string} is already connected; stopping reconnect attempts"
-                );
-                return;
-            }
-
-            let socket_addr = match addr_string.parse() {
-                Ok(addr) => addr,
-                Err(err) => {
-                    warn!("[reconnect] invalid dropped peer address {addr_string}: {err}");
-                    break;
-                }
-            };
-
-            let connect = Connect {
-                addr: socket_addr,
-                node_ip: addr_string.clone(),
-                wallet: context.wallet.clone(),
-                db: context.db.clone(),
-                map: context.map.clone(),
-                first: peer_connection_count().await == 0,
-            };
-
-            match connect_and_handshake(connect).await {
-                Ok(()) => {
-                    info!("[reconnect] reconnected dropped peer {addr_string} on attempt {attempt}");
-                    return;
-                }
-                Err(err) => {
-                    let err_string = err.to_string();
-                    if err_string.contains(
-                        "The connection is already in the connection manager Please wait 10 minutes and try again",
-                    ) {
-                        warn!(
-                            "[reconnect] dropped peer {addr_string} is still recorded by remote connection manager on attempt {attempt}/3; retrying"
-                        );
-                        continue;
-                    }
-                    warn!(
-                        "[reconnect] failed to reconnect dropped peer {addr_string} on attempt {attempt}/3: {err_string}"
-                    );
-                }
-            }
         }
 
-        reconnect_replacement_inner(&ip).await;
-    }
-    .await;
+        if Connection::get_stream_from_memory(&addr_string)
+            .await
+            .is_some()
+        {
+            info!(
+                    "[reconnect] dropped peer {addr_string} is already connected; stopping reconnect attempts"
+                );
+            return;
+        }
 
-    finish_reconnect();
+        let socket_addr = match addr_string.parse() {
+            Ok(addr) => addr,
+            Err(err) => {
+                warn!("[reconnect] invalid dropped peer address {addr_string}: {err}");
+                break;
+            }
+        };
+
+        let connect = Connect {
+            addr: socket_addr,
+            node_ip: addr_string.clone(),
+            wallet: context.wallet.clone(),
+            db: context.db.clone(),
+            map: context.map.clone(),
+            first: peer_connection_count().await == 0,
+        };
+
+        match connect_and_handshake(connect).await {
+            Ok(()) => {
+                info!("[reconnect] reconnected dropped peer {addr_string} on attempt {attempt}");
+                return;
+            }
+            Err(err) => {
+                let err_string = err.to_string();
+                if err_string.contains(
+                        "The connection is already in the connection manager Please wait 10 minutes and try again",
+                    ) {
+                    warn!(
+                            "[reconnect] dropped peer {addr_string} is still recorded by remote connection manager on attempt {attempt}/3; retrying"
+                        );
+                    continue;
+                }
+                warn!(
+                        "[reconnect] failed to reconnect dropped peer {addr_string} on attempt {attempt}/3: {err_string}"
+                    );
+            }
+        }
+    }
+
+    reconnect_replacement_inner(&ip).await;
 }
 
 pub fn spawn_retry_dropped_outgoing(ip: String, port: u16) {
@@ -212,21 +232,26 @@ pub fn spawn_retry_dropped_outgoing(ip: String, port: u16) {
 }
 
 pub fn spawn_reconnect_bootstrap(params: BootstrapParams) {
-    if !try_start_reconnect() {
+    let recovery_key = format!("bootstrap:{}", params.connections_key);
+    let Some(recovery_guard) = try_start_recovery(recovery_key) else {
         warn!("[reconnect] bootstrap recovery already in progress, skipping duplicate request");
         return;
-    }
+    };
 
     // Bootstrap discovery can perform network requests, so it runs detached
     // from the caller that noticed the connection problem.
     tokio::spawn(async move {
+        let _recovery_guard = recovery_guard;
         if let Err(err) = bootstrap_peer_discovery(params).await {
             warn!("[reconnect] bootstrap recovery failed: {err}");
         }
-        finish_reconnect();
     });
 }
 
+pub async fn refill_outgoing_connections_once() {
+    reconnect_replacement_inner("").await;
+}
+
 impl Connection {
     // Initialize the in-memory connection manager state.
     pub fn new() -> Self {
@@ -312,7 +337,6 @@ impl Connection {
         if let Some(connection_info) = removed.as_ref() {
             if ClientType::from_bytes(&connection_info.client_type) == Some(ClientType::Miner)
                 && connection_info.ready
-                && is_normal_mode()
             {
                 spawn_monitor_update(
                     ip.clone(),
@@ -388,76 +412,82 @@ impl Connection {
                 message.push(message_type);
                 message.extend_from_slice(&checkup_key);
 
-                RpcResponse::send_raw(&stream, None, &message).await;
-
-                let response_result = {
+                let sent =
+                    RpcResponse::send_raw(&stream, Some(&format!("{ip}:{port}")), &message).await;
+                let reply_received = if sent {
                     let mut checkup_rx = checkup_rx_mutex.lock().await;
-                    timeout(Duration::from_secs(30), checkup_rx.recv()).await
+                    matches!(
+                        timeout(Duration::from_secs(30), checkup_rx.recv()).await,
+                        Ok(Some(_))
+                    )
+                } else {
+                    false
                 };
 
-                match response_result {
-                    Ok(Some(_reply)) => {
-                        consecutive_failures = 0;
-                        info!(
-                            "[connection_manager] liveness check ok: type={} peer={}:{}",
-                            connection_type.as_str(),
-                            ip,
-                            port
-                        );
-                    }
-                    _ => {
-                        let still_monitoring_same_stream = {
-                            let guard = CONNECTIONS.read().await;
-                            guard
-                                .as_ref()
-                                .and_then(|conn| conn.connection_map.get(&connection_key))
-                                .map(|connection_info| {
-                                    Arc::ptr_eq(&connection_info.stream, &stream)
-                                })
-                                .unwrap_or(false)
-                        };
-
-                        if !still_monitoring_same_stream {
-                            delete_entry(command_map.clone(), checkup_key).await;
-                            break;
-                        }
-
-                        consecutive_failures = consecutive_failures.saturating_add(1);
-                        warn!(
-                            "[connection_manager] liveness check failed: type={} peer={}:{} attempt={}/3",
-                            connection_type.as_str(),
-                            ip,
-                            port,
-                            consecutive_failures
-                        );
+                if reply_received {
+                    set_stream_health(&stream, ConnectionHealth::Connected).await;
+                    consecutive_failures = 0;
+                    info!(
+                        "[connection_manager] liveness check ok: type={} peer={}:{}",
+                        connection_type.as_str(),
+                        ip,
+                        port
+                    );
+                } else {
+                    // A busy stream is still owned by active work. It is not
+                    // equivalent to a peer that accepted a ping but never replied.
+                    if stream_health(&stream).await == Some(ConnectionHealth::Busy) {
                         delete_entry(command_map.clone(), checkup_key).await;
+                        continue;
+                    }
+                    set_stream_health(&stream, ConnectionHealth::Unresponsive).await;
+                    let still_monitoring_same_stream = {
+                        let guard = CONNECTIONS.read().await;
+                        guard
+                            .as_ref()
+                            .and_then(|conn| conn.connection_map.get(&connection_key))
+                            .map(|connection_info| Arc::ptr_eq(&connection_info.stream, &stream))
+                            .unwrap_or(false)
+                    };
 
-                        if consecutive_failures < 3 {
-                            continue;
-                        }
-
-                        // Three consecutive timed-out or missing replies drop
-                        // the connection, and outgoing peers trigger
-                        // replacement discovery.
-                        let mut guard = CONNECTIONS.write().await;
-                        if let Some(conn) = guard.as_mut() {
-                            let should_drop = conn
-                                .connection_map
-                                .get(&connection_key)
-                                .map(|connection_info| {
-                                    Arc::ptr_eq(&connection_info.stream, &stream)
-                                })
-                                .unwrap_or(false);
-                            if should_drop {
-                                conn.drop_connection(connection_type, ip.clone(), port);
-                            }
-                        }
-                        drop(guard);
-                        if connection_type == ConnectionType::Outgoing {
-                            spawn_retry_dropped_outgoing(ip.clone(), port);
-                        }
+                    if !still_monitoring_same_stream {
+                        delete_entry(command_map.clone(), checkup_key).await;
                         break;
                     }
+
+                    consecutive_failures = consecutive_failures.saturating_add(1);
+                    warn!(
+                        "[connection_manager] liveness check failed: type={} peer={}:{} attempt={}/3",
+                        connection_type.as_str(),
+                        ip,
+                        port,
+                        consecutive_failures
+                    );
+                    delete_entry(command_map.clone(), checkup_key).await;
+
+                    if consecutive_failures < 3 {
+                        continue;
+                    }
+
+                    // Three consecutive timed-out or missing replies drop
+                    // the connection, and outgoing peers trigger
+                    // replacement discovery.
+                    let mut guard = CONNECTIONS.write().await;
+                    if let Some(conn) = guard.as_mut() {
+                        let should_drop = conn
+                            .connection_map
+                            .get(&connection_key)
+                            .map(|connection_info| Arc::ptr_eq(&connection_info.stream, &stream))
+                            .unwrap_or(false);
+                        if should_drop {
+                            conn.drop_connection(connection_type, ip.clone(), port);
+                        }
+                    }
+                    drop(guard);
+                    if connection_type == ConnectionType::Outgoing {
+                        spawn_retry_dropped_outgoing(ip.clone(), port);
+                    }
+                    break;
                 }
             }
         });
@@ -483,6 +513,17 @@ impl Connection {
             .count()
     }
 
+    pub fn count_ready_outgoing_connections(&self) -> usize {
+        self.connection_map
+            .values()
+            .filter(|info| {
+                ConnectionType::from_bytes(&info.connection_type) == Some(ConnectionType::Outgoing)
+                    && ClientType::from_bytes(&info.client_type) == Some(ClientType::Miner)
+                    && info.ready
+            })
+            .count()
+    }
+
     pub fn mark_wallet_registry_synced(&mut self, key: &str) -> bool {
         let Some((ip, port)) = split_ip_port_key(key) else {
             return false;
@@ -511,38 +552,98 @@ impl Connection {
         false
     }
 
-    pub fn mark_operational(&mut self, key: &str, command_map: Arc<Mutex<Command>>) -> bool {
+    fn finalize_setup_if_complete(
+        info: &mut ConnectionInfo,
+        ip: String,
+        port: u16,
+        command_map: Arc<Mutex<Command>>,
+    ) -> bool {
+        if info.ready {
+            return true;
+        }
+        if ClientType::from_bytes(&info.client_type) != Some(ClientType::Miner)
+            || !info.wallet_registry_synced
+            || !info.network_map_synced
+            || !info.local_setup_complete
+            || !info.remote_setup_complete
+            || !info.local_setup_acknowledged
+        {
+            return false;
+        }
+
+        info.ready = true;
+        spawn_monitor_update(
+            ip.clone(),
+            MONITOR_ACTION_ADD,
+            info.wallet_short_address.clone(),
+            port,
+        );
+        Connection::client_checkup(
+            Arc::clone(&info.stream),
+            ConnectionType::from_bytes(&info.connection_type).unwrap_or(ConnectionType::Incoming),
+            ip,
+            port,
+            command_map,
+        );
+        true
+    }
+
+    pub fn mark_local_setup_complete(
+        &mut self,
+        key: &str,
+        command_map: Arc<Mutex<Command>>,
+    ) -> Option<(Arc<Mutex<TcpStream>>, bool)> {
+        let Some((ip, port)) = split_ip_port_key(key) else {
+            return None;
+        };
+        let ip_bytes = ip_to_binary(&ip);
+        for (connection_key, info) in self.connection_map.iter_mut() {
+            if connection_key.ip == ip_bytes && connection_key.port == port {
+                if ClientType::from_bytes(&info.client_type) != Some(ClientType::Miner)
+                    || !info.wallet_registry_synced
+                    || !info.network_map_synced
+                {
+                    return None;
+                }
+                info.local_setup_complete = true;
+                Self::finalize_setup_if_complete(info, ip, port, command_map);
+                return Some((Arc::clone(&info.stream), info.local_setup_acknowledged));
+            }
+        }
+        None
+    }
+
+    pub fn mark_remote_setup_complete(
+        &mut self,
+        key: &str,
+        command_map: Arc<Mutex<Command>>,
+    ) -> bool {
         let Some((ip, port)) = split_ip_port_key(key) else {
             return false;
         };
         let ip_bytes = ip_to_binary(&ip);
         for (connection_key, info) in self.connection_map.iter_mut() {
             if connection_key.ip == ip_bytes && connection_key.port == port {
-                if info.ready {
-                    return true;
-                }
-                if ClientType::from_bytes(&info.client_type) != Some(ClientType::Miner)
-                    || !info.wallet_registry_synced
-                    || !info.network_map_synced
-                {
-                    return false;
-                }
-                info.ready = true;
-                spawn_monitor_update(
-                    ip.clone(),
-                    MONITOR_ACTION_ADD,
-                    info.wallet_short_address.clone(),
-                    port,
-                );
-                Connection::client_checkup(
-                    Arc::clone(&info.stream),
-                    ConnectionType::from_bytes(&info.connection_type)
-                        .unwrap_or(ConnectionType::Incoming),
-                    ip,
-                    port,
-                    command_map,
-                );
-                return true;
+                info.remote_setup_complete = true;
+                return Self::finalize_setup_if_complete(info, ip, port, command_map);
+            }
+        }
+        false
+    }
+
+    pub fn mark_local_setup_acknowledged(
+        &mut self,
+        key: &str,
+        command_map: Arc<Mutex<Command>>,
+    ) -> bool {
+        let Some((ip, port)) = split_ip_port_key(key) else {
+            return false;
+        };
+        let ip_bytes = ip_to_binary(&ip);
+        for (connection_key, info) in self.connection_map.iter_mut() {
+            if connection_key.ip == ip_bytes && connection_key.port == port {
+                info.local_setup_acknowledged = true;
+                return Self::finalize_setup_if_complete(info, ip, port, command_map);
             }
         }
         false
@@ -797,11 +898,8 @@ impl Connection {
 
 fn spawn_monitor_update(ip: String, action: u8, monitored_address: String, port: u16) {
     tokio::spawn(async move {
-        let context = {
-            let guard = RECONNECT_CONTEXT.lock().await;
-            guard.clone()
-        };
-        let Some(context) = context else {
+        let Some(context) = node_runtime_context() else {
+            warn!("[network_map] node runtime context is not initialized");
             return;
         };
         if !Wallet::short_address_validation(&monitored_address) {
@@ -812,12 +910,16 @@ fn spawn_monitor_update(ip: String, action: u8, monitored_address: String, port:
             return;
         }
         let timestamp = crate::Utc::now().timestamp_millis() as u64;
+        let modified_block =
+            crate::records::block_height::get_block_height::get_height(&context.db)
+                .saturating_add(1);
         let signature = NodeInfo::monitor_signature(
             action,
             &monitored_address,
             &monitoring_address,
             &ip,
             timestamp,
+            modified_block,
             &context.wallet,
         )
         .await;
@@ -827,6 +929,7 @@ fn spawn_monitor_update(ip: String, action: u8, monitored_address: String, port:
             monitoring_address,
             target_ip: ip.clone(),
             modified_timestamp: timestamp,
+            modified_block,
             modified_signature: signature,
         };
         let params = MonitorAddressParams {
@@ -868,6 +971,15 @@ pub async fn outgoing_connection_count() -> usize {
         .unwrap_or(0)
 }
 
+pub async fn ready_outgoing_connection_count() -> usize {
+    CONNECTIONS
+        .read()
+        .await
+        .as_ref()
+        .map(|connection| connection.count_ready_outgoing_connections())
+        .unwrap_or(0)
+}
+
 pub async fn peer_connection_count() -> usize {
     // Mining needs at least one fully initialized miner peer. A raw socket
     // is not enough because block validation depends on wallet registry and
@@ -910,14 +1022,80 @@ pub async fn mark_peer_network_map_synced(key: &str) -> bool {
 }
 
 pub async fn mark_peer_operational(key: &str, map: Arc<Mutex<Command>>) -> bool {
+    let setup = CONNECTIONS
+        .write()
+        .await
+        .as_mut()
+        .and_then(|connection| connection.mark_local_setup_complete(key, map.clone()));
+    let Some((stream, already_acknowledged)) = setup else {
+        return false;
+    };
+    if already_acknowledged {
+        return peer_is_operational(key).await;
+    }
+
+    for attempt in 1..=3 {
+        let (uid, _tx, rx) = reserve_entry_with_context(
+            map.clone(),
+            Some(RPC_SETUP_COMPLETE),
+            Some(key.to_string()),
+        )
+        .await;
+        let mut message = Vec::with_capacity(4);
+        message.push(RPC_SETUP_COMPLETE);
+        message.extend_from_slice(&uid);
+
+        if !RpcResponse::send_raw(&stream, Some(key), &message).await {
+            delete_entry(map.clone(), uid).await;
+            continue;
+        }
+
+        let response = {
+            let mut rx = rx.lock().await;
+            timeout(Duration::from_secs(10), rx.recv()).await
+        };
+        delete_entry(map.clone(), uid).await;
+        if matches!(response, Ok(Some(bytes)) if bytes == b"setup_complete_ack") {
+            return CONNECTIONS
+                .write()
+                .await
+                .as_mut()
+                .map(|connection| connection.mark_local_setup_acknowledged(key, map.clone()))
+                .unwrap_or(false);
+        }
+        warn!(
+            "[handshake_state] setup-complete acknowledgement failed: peer={key} attempt={attempt}/3"
+        );
+    }
+
+    false
+}
+
+pub async fn mark_peer_remote_setup_complete(key: &str, map: Arc<Mutex<Command>>) -> bool {
     CONNECTIONS
         .write()
         .await
         .as_mut()
-        .map(|connection| connection.mark_operational(key, map))
+        .map(|connection| connection.mark_remote_setup_complete(key, map))
         .unwrap_or(false)
 }
 
+pub fn spawn_peer_setup_retry(key: String, map: Arc<Mutex<Command>>) {
+    tokio::spawn(async move {
+        loop {
+            sleep(Duration::from_secs(10)).await;
+            if Connection::get_stream_from_memory(&key).await.is_none()
+                || peer_is_operational(&key).await
+            {
+                break;
+            }
+            if mark_peer_operational(&key, map.clone()).await {
+                break;
+            }
+        }
+    });
+}
+
 pub async fn peer_is_operational(key: &str) -> bool {
     let Some((ip, port)) = split_ip_port_key(key) else {
         return false;
@@ -974,6 +1152,30 @@ pub async fn peer_accepts_live_relay(key: &str) -> bool {
         .unwrap_or(false)
 }
 
+pub async fn set_stream_health(stream: &Arc<Mutex<TcpStream>>, health: ConnectionHealth) {
+    let mut guard = CONNECTIONS.write().await;
+    let Some(connection) = guard.as_mut() else {
+        return;
+    };
+    if let Some(info) = connection
+        .connection_map
+        .values_mut()
+        .find(|info| Arc::ptr_eq(&info.stream, stream))
+    {
+        info.health = health;
+    }
+}
+
+pub async fn stream_health(stream: &Arc<Mutex<TcpStream>>) -> Option<ConnectionHealth> {
+    CONNECTIONS.read().await.as_ref().and_then(|connection| {
+        connection
+            .connection_map
+            .values()
+            .find(|info| Arc::ptr_eq(&info.stream, stream))
+            .map(|info| info.health)
+    })
+}
+
 pub async fn live_miner_peer_streams() -> Vec<(String, Arc<Mutex<TcpStream>>)> {
     // Snapshot consensus and recovery checks vote only across currently
     // connected miner peers, regardless of incoming/outgoing direction.
@@ -1026,3 +1228,43 @@ pub async fn get_client_type_from_memory(key: &str) -> Option<ClientType> {
 
     None
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+    use crate::wallets::structures::SavedWallet;
+
+    #[test]
+    fn runtime_context_can_be_initialized_before_any_handshake() {
+        let db = sled::Config::new().temporary(true).open().unwrap();
+        let wallet = Arc::new(Wallet {
+            saved: SavedWallet {
+                short_address: "ab13318c26250b048db92920a80a86127c933b0c.cltc".to_string(),
+                vanity_address: None,
+                public_key: String::new(),
+                private_key: String::new(),
+            },
+            encryption_key: String::new(),
+        });
+        let map = Arc::new(Mutex::new(Command::new()));
+
+        initialize_node_runtime_context(db.clone(), wallet.clone(), map.clone()).unwrap();
+        let context = node_runtime_context().expect("runtime context should be available");
+
+        assert!(Arc::ptr_eq(&context.wallet, &wallet));
+        assert!(Arc::ptr_eq(&context.map, &map));
+    }
+
+    #[test]
+    fn recovery_is_deduplicated_per_peer_instead_of_globally() {
+        let first = try_start_recovery("test-peer:first".to_string())
+            .expect("first peer recovery should start");
+        assert!(try_start_recovery("test-peer:first".to_string()).is_none());
+
+        let second = try_start_recovery("test-peer:second".to_string())
+            .expect("another peer must recover independently");
+        drop(first);
+        assert!(try_start_recovery("test-peer:first".to_string()).is_some());
+        drop(second);
+    }
+}

src/records/memory/mempool/mod.rs

@@ -232,8 +232,7 @@ pub use lookups::{
 pub use processing::{
     delete_by_signatures, delete_unprocessed_by_address, mark_processed_by_signatures,
     mark_selected_transactions_processed, restore_processed_by_signatures,
-    restore_selected_transactions_processed,
-    spawn_processed_cleanup,
+    restore_selected_transactions_processed, spawn_processed_cleanup,
 };
 pub use schema::{
     clear_mempool, db_client, ensure_db_connection, init_db, pg_execute, pg_query, setup_mempool,

src/records/memory/network_mapping/add.rs

@@ -14,22 +14,10 @@ fn signature_is_empty(signature: &str) -> bool {
             .unwrap_or(false)
 }
 
-fn merge_monitors(existing: &mut Vec<String>, monitors: Vec<String>) -> bool {
-    let mut changed = false;
-    for monitor in monitors {
-        if !existing.iter().any(|known| known == &monitor) {
-            existing.push(monitor);
-            changed = true;
-        }
-    }
-    changed
-}
-
 impl NodeInfo {
     pub async fn import_signed_mapping_address(
         db: &Db,
         edit: SignedNodeEdit,
-        monitors: Vec<String>,
         _blocks_mined: u8,
     ) -> Result<(), String> {
         if !is_public_network_address(&edit.ip) {
@@ -66,11 +54,8 @@ impl NodeInfo {
                 existing_node.added_by = edit.modified_by;
                 existing_node.added_timestamp = edit.modified_timestamp;
                 existing_node.added_signature = edit.modified_signature;
-                existing_node.monitoring = monitors;
-                existing_node.deleted_timestamp = 0_u64;
-                existing_node.deleted_block = 0_u32;
                 drop(address_map);
-                Self::persist_recovery_snapshot("import revive").await;
+                Self::persist_recovery_snapshot("import deleted membership").await;
                 return Ok(());
             } else {
                 if existing_node.ip != edit.ip {
@@ -88,7 +73,6 @@ impl NodeInfo {
                     existing_node.added_timestamp = edit.modified_timestamp;
                     existing_node.added_signature = edit.modified_signature;
                 }
-                merge_monitors(&mut existing_node.monitoring, monitors);
                 drop(address_map);
                 Self::persist_recovery_snapshot("import existing").await;
                 return Ok(());
@@ -102,16 +86,14 @@ impl NodeInfo {
         }
 
         address_map.insert(edit.address, {
-            let mut node = NodeInfo::new(
+            NodeInfo::new(
                 edit.ip,
                 edit.port,
                 0,
                 edit.modified_by,
                 edit.modified_timestamp,
                 edit.modified_signature,
-            );
-            node.monitoring = monitors;
-            node
+            )
         });
         drop(address_map);
         Self::persist_recovery_snapshot("import add").await;
@@ -170,20 +152,9 @@ impl NodeInfo {
         };
         let modified_timestamp_bytes = edit.modified_timestamp.to_le_bytes();
         let modified_signature_bytes = decode(&edit.modified_signature).unwrap();
-        let monitor_bytes = {
-            let address_map = ADDRESS_MAP.lock().await;
-            address_map
-                .get(&edit.address)
-                .map(|node| {
-                    node.monitoring
-                        .iter()
-                        .filter_map(|monitor| Wallet::short_address_to_bytes(monitor))
-                        .collect::<Vec<_>>()
-                })
-                .unwrap_or_default()
-        };
-        let monitor_count = monitor_bytes.len().min(u16::MAX as usize) as u16;
-        let monitor_count_bytes = monitor_count.to_le_bytes();
+        // Membership broadcasts never carry liveness state. Monitor changes
+        // use their own signed commands and cannot be injected into node adds.
+        let monitor_count_bytes = 0_u16.to_le_bytes();
         let streams = {
             let connections_lock = CONNECTIONS.read().await;
             connections_lock
@@ -193,6 +164,7 @@ impl NodeInfo {
         };
         match streams {
             streams if !streams.is_empty() => {
+                let mut messages = Vec::with_capacity(streams.len());
                 for (peer_key, unlocked_stream) in streams {
                     let Some((peer_ip, _)) = peer_key.rsplit_once(':') else {
                         continue;
@@ -216,11 +188,9 @@ impl NodeInfo {
                     message.extend_from_slice(&modified_timestamp_bytes);
                     message.extend_from_slice(&modified_signature_bytes);
                     message.extend_from_slice(&monitor_count_bytes);
-                    for monitor in monitor_bytes.iter().take(monitor_count as usize) {
-                        message.extend_from_slice(monitor);
-                    }
-                    RpcResponse::send_raw(&unlocked_stream, Some(&peer_key), &message).await;
+                    messages.push((peer_key, unlocked_stream, message));
                 }
+                RpcResponse::broadcast_raw(messages).await;
             }
             _ => {
                 warn!("No active connections found.");
@@ -232,7 +202,7 @@ impl NodeInfo {
         let AddAddressParams {
             map,
             mut edit,
-            monitors,
+            monitors: _monitors,
             blocks_mined: _blocks_mined,
             remote_ip,
             db,
@@ -384,8 +354,6 @@ impl NodeInfo {
                     existing_node.added_by = edit.modified_by.clone();
                     existing_node.added_timestamp = edit.modified_timestamp;
                     existing_node.added_signature = edit.modified_signature.clone();
-                    existing_node.deleted_timestamp = 0_u64;
-                    existing_node.deleted_block = 0_u32;
                     state_changed = true;
                 } else {
                     if existing_node.ip != edit.ip {
@@ -403,9 +371,6 @@ impl NodeInfo {
                         existing_node.added_signature = edit.modified_signature.clone();
                         state_changed = true;
                     }
-                    if merge_monitors(&mut existing_node.monitoring, monitors.clone()) {
-                        state_changed = true;
-                    }
                 }
             }
 
@@ -421,18 +386,17 @@ impl NodeInfo {
                 if !penalize_duplicate_ip {
                     // Persist the new node locally. Network-map entries are bare
                     // IP membership records, separate from live socket keys.
-                    address_map.insert(edit.address.clone(), {
-                        let mut node = NodeInfo::new(
+                    address_map.insert(
+                        edit.address.clone(),
+                        NodeInfo::new(
                             edit.ip.clone(),
                             edit.port,
                             0,
                             edit.modified_by.clone(),
                             edit.modified_timestamp,
                             edit.modified_signature.clone(),
-                        );
-                        node.monitoring = monitors.clone();
-                        node
-                    });
+                        ),
+                    );
                     state_changed = true;
                 }
             }

src/records/memory/network_mapping/mined_counts.rs

@@ -43,30 +43,6 @@ impl NodeInfo {
         }
     }
 
-    pub async fn set_deleted_block_from_mapping(address: &str, deleted_block: u32) {
-        let mut map = ADDRESS_MAP.lock().await;
-        if let Some(node_info) = map.get_mut(address) {
-            // The deletion height is filled in once the chain knows the block
-            // where the delete action becomes active.
-            node_info.deleted_block = deleted_block;
-        }
-    }
-
-    pub async fn set_deleted_metadata_from_mapping(
-        address: &str,
-        deleted_timestamp: u64,
-        deleted_block: u32,
-    ) {
-        {
-            let mut map = ADDRESS_MAP.lock().await;
-            if let Some(node_info) = map.get_mut(address) {
-                node_info.deleted_timestamp = deleted_timestamp;
-                node_info.deleted_block = deleted_block;
-            }
-        }
-        Self::persist_recovery_snapshot("deleted metadata import").await;
-    }
-
     pub async fn rebuild_mined_counts_from_chain(db: &Db) -> Result<(), String> {
         // Recompute node mined counts directly from saved block headers
         // so startup and recovery can rebuild memory-only state.

src/records/memory/network_mapping/monitor.rs

@@ -1,4 +1,9 @@
 use super::*;
+use crate::records::memory::network_mapping::structs::{
+    MONITOR_EVENT_ACTION_OFFSET, MONITOR_EVENT_BLOCK_OFFSET, MONITOR_EVENT_BYTES,
+    MONITOR_EVENT_MONITORED_OFFSET, MONITOR_EVENT_MONITORING_OFFSET,
+    MONITOR_EVENT_SIGNATURE_OFFSET, MONITOR_EVENT_TARGET_IP_OFFSET, MONITOR_EVENT_TIMESTAMP_OFFSET,
+};
 use crate::records::memory::response_channels::reserve_transient_entry_with_context;
 use crate::records::wallet_registry::resolve_pubkey_from_short_address;
 use crate::rpc::command_maps::{RPC_NETWORK_MONITOR_ADD, RPC_NETWORK_MONITOR_REMOVE};
@@ -7,7 +12,102 @@ pub const MONITOR_ACTION_ADD: u8 = 1;
 pub const MONITOR_ACTION_REMOVE: u8 = 2;
 
 lazy_static! {
-    static ref MONITOR_EVENTS_SEEN: Mutex<HashMap<String, u64>> = Mutex::new(HashMap::new());
+    // Keep the newest signed assertion for each monitor relationship. This is
+    // both replay protection and the authenticated state shared with joining peers.
+    static ref MONITOR_EVENT_STATE: Mutex<HashMap<String, SignedMonitorEdit>> =
+        Mutex::new(HashMap::new());
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    fn node(monitors: &[&str]) -> NodeInfo {
+        let mut node = NodeInfo::new(
+            "1.2.3.4".to_string(),
+            50050,
+            0,
+            "sponsor.cltc".to_string(),
+            1,
+            "00".repeat(Wallet::SIGNATURE_LENGTH),
+        );
+        node.monitoring = monitors.iter().map(|monitor| monitor.to_string()).collect();
+        node
+    }
+
+    #[test]
+    fn deletion_cascade_preserves_nodes_with_another_monitor() {
+        let mut map = HashMap::new();
+        map.insert("node-a".to_string(), node(&[]));
+        map.insert("node-b".to_string(), node(&["node-a", "node-c"]));
+        map.insert("node-c".to_string(), node(&["local"]));
+
+        NodeInfo::mark_deleted_and_cascade(&mut map, "node-a", "local", 100, 50);
+
+        assert_eq!(map["node-a"].deleted_block, 50);
+        assert_eq!(map["node-b"].monitoring, vec!["node-c".to_string()]);
+        assert_eq!(map["node-b"].deleted_timestamp, 0);
+    }
+
+    #[test]
+    fn deletion_cascade_removes_offline_nodes_from_their_targets() {
+        let mut map = HashMap::new();
+        map.insert("node-a".to_string(), node(&[]));
+        map.insert("node-b".to_string(), node(&["node-a"]));
+        map.insert("node-c".to_string(), node(&["node-b"]));
+
+        NodeInfo::mark_deleted_and_cascade(&mut map, "node-a", "local", 100, 50);
+
+        for address in ["node-a", "node-b", "node-c"] {
+            assert_eq!(map[address].deleted_timestamp, 100);
+            assert_eq!(map[address].deleted_block, 50);
+            assert!(map[address].monitoring.is_empty());
+        }
+    }
+
+    #[test]
+    fn remove_wins_an_equal_timestamp_monitor_race() {
+        let add = SignedMonitorEdit {
+            action: MONITOR_ACTION_ADD,
+            monitored_address: "target".to_string(),
+            monitoring_address: "monitor".to_string(),
+            target_ip: "1.2.3.4".to_string(),
+            modified_timestamp: 100,
+            modified_block: 50,
+            modified_signature: "a".to_string(),
+        };
+        let mut remove = add.clone();
+        remove.action = MONITOR_ACTION_REMOVE;
+
+        assert!(NodeInfo::monitor_event_order(&remove) > NodeInfo::monitor_event_order(&add));
+    }
+
+    #[test]
+    fn signed_monitor_wire_record_round_trips() {
+        let monitored = "ab13318c26250b048db92920a80a86127c933b0c.cltc".to_string();
+        let monitoring = "7ba97ce531937388ef8bca4d8a4ca54cefbefb87.cltc".to_string();
+        let mut bytes = Vec::with_capacity(MONITOR_EVENT_BYTES);
+        bytes.push(MONITOR_ACTION_REMOVE);
+        bytes.extend_from_slice(&Wallet::short_address_to_bytes(&monitored).unwrap());
+        bytes.extend_from_slice(&Wallet::short_address_to_bytes(&monitoring).unwrap());
+        bytes.extend_from_slice(&ip_to_binary("1.2.3.4"));
+        bytes.extend_from_slice(&100_u64.to_le_bytes());
+        bytes.extend_from_slice(&50_u32.to_le_bytes());
+        bytes.extend_from_slice(&vec![7_u8; Wallet::SIGNATURE_LENGTH]);
+
+        assert_eq!(bytes.len(), MONITOR_EVENT_BYTES);
+        let parsed = NodeInfo::monitor_event_from_bytes(&bytes).unwrap();
+        assert_eq!(parsed.action, MONITOR_ACTION_REMOVE);
+        assert_eq!(parsed.monitored_address, monitored);
+        assert_eq!(parsed.monitoring_address, monitoring);
+        assert_eq!(parsed.target_ip, "1.2.3.4");
+        assert_eq!(parsed.modified_timestamp, 100);
+        assert_eq!(parsed.modified_block, 50);
+        assert_eq!(
+            parsed.modified_signature,
+            crate::encode(vec![7_u8; Wallet::SIGNATURE_LENGTH])
+        );
+    }
 }
 
 impl NodeInfo {
@@ -16,12 +116,13 @@ impl NodeInfo {
         monitored_address: &str,
         monitoring_address: &str,
         target_ip: &str,
-        current_timestamp: u64,
+        modified_timestamp: u64,
+        modified_block: u32,
         wallet: &Arc<Wallet>,
     ) -> String {
         let private_key = &wallet.saved.private_key;
         let data = format!(
-            "{action}{monitored_address}{monitoring_address}{target_ip}{current_timestamp}"
+            "{action}{monitored_address}{monitoring_address}{target_ip}{modified_timestamp}{modified_block}"
         );
         let hashed_data = skein_256_hash_data(&data);
         Wallet::sign_transaction(&hashed_data, private_key).await
@@ -31,7 +132,7 @@ impl NodeInfo {
         address_map: &mut HashMap<String, NodeInfo>,
         deleted_address: &str,
         local_address: &str,
-        current_timestamp: u64,
+        deleted_timestamp: u64,
         deleted_block: u32,
     ) {
         let mut stack = vec![deleted_address.to_string()];
@@ -40,7 +141,7 @@ impl NodeInfo {
             let should_cascade = match address_map.get_mut(&address) {
                 Some(_) if address == local_address => false,
                 Some(node) if node.deleted_timestamp == 0 && node.monitoring.is_empty() => {
-                    node.deleted_timestamp = current_timestamp;
+                    node.deleted_timestamp = deleted_timestamp;
                     node.deleted_block = deleted_block;
                     true
                 }
@@ -53,9 +154,12 @@ impl NodeInfo {
 
             info!(
                 "[network_map] node marked deleted: address={} timestamp={} deleted_block={}",
-                address, current_timestamp, deleted_block
+                address, deleted_timestamp, deleted_block
             );
 
+            // Once a node has no monitors, its own old monitoring claims can
+            // no longer keep other nodes active. Remove those claims locally;
+            // no signature from the now-offline node is required.
             let targets: Vec<String> = address_map
                 .iter_mut()
                 .filter_map(|(target, node)| {
@@ -81,12 +185,13 @@ impl NodeInfo {
             return false;
         };
         let data = format!(
-            "{}{}{}{}{}",
+            "{}{}{}{}{}{}",
             edit.action,
             edit.monitored_address,
             edit.monitoring_address,
             edit.target_ip,
-            edit.modified_timestamp
+            edit.modified_timestamp,
+            edit.modified_block,
         );
         let hashed_data = skein_256_hash_data(&data);
         Wallet::verify_transaction_with_public_key_bytes(
@@ -105,24 +210,47 @@ impl NodeInfo {
         Self::apply_monitor(params, MONITOR_ACTION_REMOVE).await
     }
 
-    fn monitor_event_key(edit: &SignedMonitorEdit) -> String {
+    fn monitor_relation_key(edit: &SignedMonitorEdit) -> String {
         format!(
-            "{}:{}:{}:{}:{}:{}",
-            edit.action,
-            edit.monitored_address,
-            edit.monitoring_address,
-            edit.target_ip,
-            edit.modified_timestamp,
-            edit.modified_signature
+            "{}:{}:{}",
+            edit.monitored_address, edit.monitoring_address, edit.target_ip
         )
     }
 
-    async fn remember_monitor_event(edit: &SignedMonitorEdit) -> bool {
-        let key = Self::monitor_event_key(edit);
-        let mut seen = MONITOR_EVENTS_SEEN.lock().await;
-        let now = Utc::now().timestamp_millis() as u64;
-        seen.retain(|_, timestamp| now.saturating_sub(*timestamp) < 3_600_000);
-        seen.insert(key, now).is_none()
+    fn monitor_event_order(edit: &SignedMonitorEdit) -> (u64, u32, u8, &str) {
+        (
+            edit.modified_timestamp,
+            edit.modified_block,
+            edit.action,
+            &edit.modified_signature,
+        )
+    }
+
+    async fn remember_newest_monitor_event(edit: &SignedMonitorEdit) -> bool {
+        let key = Self::monitor_relation_key(edit);
+        let mut state = MONITOR_EVENT_STATE.lock().await;
+        if state
+            .get(&key)
+            .map(|existing| Self::monitor_event_order(edit) <= Self::monitor_event_order(existing))
+            .unwrap_or(false)
+        {
+            return false;
+        }
+        state.insert(key, edit.clone());
+        true
+    }
+
+    async fn deletion_marker_for(monitored_address: &str) -> Option<(u64, u32)> {
+        MONITOR_EVENT_STATE
+            .lock()
+            .await
+            .values()
+            .filter(|event| {
+                event.monitored_address == monitored_address
+                    && event.action == MONITOR_ACTION_REMOVE
+            })
+            .map(|event| (event.modified_timestamp, event.modified_block))
+            .max()
     }
 
     async fn broadcast_monitor_event(
@@ -145,9 +273,10 @@ impl NodeInfo {
         };
         let target_ip_bytes = ip_to_binary(&edit.target_ip);
         let timestamp_bytes = edit.modified_timestamp.to_le_bytes();
+        let block_bytes = edit.modified_block.to_le_bytes();
         let signature_bytes = match decode(&edit.modified_signature) {
-            Ok(bytes) => bytes,
-            Err(_) => return,
+            Ok(bytes) if bytes.len() == Wallet::SIGNATURE_LENGTH => bytes,
+            _ => return,
         };
         let streams = {
             let connections_lock = CONNECTIONS.read().await;
@@ -157,6 +286,7 @@ impl NodeInfo {
                 .unwrap_or_default()
         };
 
+        let mut messages = Vec::with_capacity(streams.len());
         for (peer_key, unlocked_stream) in streams {
             let Some((peer_ip, _)) = peer_key.rsplit_once(':') else {
                 continue;
@@ -170,7 +300,7 @@ impl NodeInfo {
                 Some(peer_key.clone()),
             )
             .await;
-            let mut message: Vec<u8> = Vec::new();
+            let mut message = Vec::with_capacity(1 + 3 + MONITOR_EVENT_BYTES);
             message.push(message_type);
             message.extend_from_slice(&hashmap_key);
             message.push(edit.action);
@@ -178,9 +308,79 @@ impl NodeInfo {
             message.extend_from_slice(&monitoring_bytes);
             message.extend_from_slice(&target_ip_bytes);
             message.extend_from_slice(&timestamp_bytes);
+            message.extend_from_slice(&block_bytes);
             message.extend_from_slice(&signature_bytes);
-            RpcResponse::send_raw(&unlocked_stream, Some(&peer_key), &message).await;
+            messages.push((peer_key, unlocked_stream, message));
         }
+        RpcResponse::broadcast_raw(messages).await;
+    }
+
+    async fn apply_verified_monitor_edit(
+        edit: &SignedMonitorEdit,
+        db: &Db,
+        local_short: &str,
+    ) -> Result<bool, String> {
+        if !matches!(edit.action, MONITOR_ACTION_ADD | MONITOR_ACTION_REMOVE) {
+            return Err("Invalid monitor action".to_string());
+        }
+        if !Self::verify_monitor_edit(edit, db).await {
+            return Err("Could not validate monitor signature".to_string());
+        }
+
+        // Validate membership before recording the event so an invalid event
+        // cannot poison replay protection for a later valid assertion.
+        {
+            let address_map = ADDRESS_MAP.lock().await;
+            let monitored = address_map
+                .get(&edit.monitored_address)
+                .ok_or_else(|| "monitored address not found".to_string())?;
+            if monitored.ip != edit.target_ip {
+                return Err("monitor target IP mismatch".to_string());
+            }
+        }
+
+        if !Self::remember_newest_monitor_event(edit).await {
+            return Ok(false);
+        }
+
+        let deletion_marker = if edit.action == MONITOR_ACTION_REMOVE {
+            Self::deletion_marker_for(&edit.monitored_address).await
+        } else {
+            None
+        };
+        let mut address_map = ADDRESS_MAP.lock().await;
+        let monitored = address_map
+            .get_mut(&edit.monitored_address)
+            .ok_or_else(|| "monitored address not found".to_string())?;
+
+        match edit.action {
+            MONITOR_ACTION_ADD => {
+                if !monitored.monitoring.contains(&edit.monitoring_address) {
+                    monitored.monitoring.push(edit.monitoring_address.clone());
+                }
+                monitored.deleted_timestamp = 0;
+                monitored.deleted_block = 0;
+            }
+            MONITOR_ACTION_REMOVE => {
+                monitored
+                    .monitoring
+                    .retain(|monitor| monitor != &edit.monitoring_address);
+                if monitored.monitoring.is_empty() {
+                    let (deleted_timestamp, deleted_block) =
+                        deletion_marker.unwrap_or((edit.modified_timestamp, edit.modified_block));
+                    Self::mark_deleted_and_cascade(
+                        &mut address_map,
+                        &edit.monitored_address,
+                        local_short,
+                        deleted_timestamp,
+                        deleted_block,
+                    );
+                }
+            }
+            _ => unreachable!(),
+        }
+
+        Ok(true)
     }
 
     async fn apply_monitor(params: MonitorAddressParams, action: u8) -> RpcResponse {
@@ -192,7 +392,6 @@ impl NodeInfo {
             map,
             ..
         } = params;
-        let current_timestamp = Utc::now().timestamp_millis() as u64;
 
         if edit.action != action {
             return RpcResponse::Binary(b"Error: Invalid monitor action".to_vec());
@@ -200,74 +399,127 @@ impl NodeInfo {
 
         let local_short = wallet.saved.short_address.clone();
         if remote_ip.is_empty() && edit.monitoring_address == local_short {
-            edit.modified_timestamp = current_timestamp;
+            edit.modified_timestamp = Utc::now().timestamp_millis() as u64;
+            edit.modified_block = get_height(&db).saturating_add(1);
             edit.modified_signature = Self::monitor_signature(
                 action,
                 &edit.monitored_address,
                 &edit.monitoring_address,
                 &edit.target_ip,
-                current_timestamp,
+                edit.modified_timestamp,
+                edit.modified_block,
                 &wallet,
             )
             .await;
         }
 
-        if !Self::verify_monitor_edit(&edit, &db).await {
-            return RpcResponse::Binary(b"Error: Could not validate monitor signature".to_vec());
-        }
-
-        let is_new_event = Self::remember_monitor_event(&edit).await;
-        if !is_new_event {
-            return RpcResponse::Binary(b"Success".to_vec());
-        }
-
-        {
-            let mut address_map = ADDRESS_MAP.lock().await;
-            let Some(monitored) = address_map.get_mut(&edit.monitored_address) else {
-                return RpcResponse::Binary(b"Error: monitored address not found".to_vec());
-            };
-            if monitored.ip != edit.target_ip {
-                return RpcResponse::Binary(b"Error: monitor target IP mismatch".to_vec());
-            }
-
-            match action {
-                MONITOR_ACTION_ADD => {
-                    if !monitored.monitoring.contains(&edit.monitoring_address) {
-                        monitored.monitoring.push(edit.monitoring_address.clone());
-                    }
-                    monitored.deleted_timestamp = 0;
-                    monitored.deleted_block = 0;
-                }
-                MONITOR_ACTION_REMOVE => {
-                    let before = monitored.monitoring.len();
-                    monitored
-                        .monitoring
-                        .retain(|monitor| monitor != &edit.monitoring_address);
-                    if before != monitored.monitoring.len() && monitored.monitoring.is_empty() {
-                        let deleted_block = get_height(&db) + 1;
-                        Self::mark_deleted_and_cascade(
-                            &mut address_map,
-                            &edit.monitored_address,
-                            &local_short,
-                            current_timestamp,
-                            deleted_block,
-                        );
-                    }
-                }
-                _ => return RpcResponse::Binary(b"Error: Invalid monitor action".to_vec()),
-            }
+        match Self::apply_verified_monitor_edit(&edit, &db, &local_short).await {
+            Ok(false) => return RpcResponse::Binary(b"Success".to_vec()),
+            Err(err) => return RpcResponse::Binary(format!("Error: {err}").into_bytes()),
+            Ok(true) => {}
         }
 
         Self::persist_recovery_snapshot("monitor update").await;
         Self::broadcast_monitor_event(map, &edit, &remote_ip).await;
-
         RpcResponse::Binary(b"Success".to_vec())
     }
 
-    pub async fn set_monitors_from_mapping(address: &str, monitors: Vec<String>) {
-        let mut address_map = ADDRESS_MAP.lock().await;
-        if let Some(node) = address_map.get_mut(address) {
-            node.monitoring = monitors;
+    pub async fn import_signed_monitor_event(
+        edit: SignedMonitorEdit,
+        db: &Db,
+        local_short: &str,
+    ) -> Result<bool, String> {
+        Self::apply_verified_monitor_edit(&edit, db, local_short).await
+    }
+
+    pub(crate) async fn signed_monitor_state_bytes() -> Vec<u8> {
+        let mut events: Vec<SignedMonitorEdit> =
+            MONITOR_EVENT_STATE.lock().await.values().cloned().collect();
+        events.sort_by(|left, right| {
+            Self::monitor_event_order(left)
+                .cmp(&Self::monitor_event_order(right))
+                .then_with(|| {
+                    Self::monitor_relation_key(left).cmp(&Self::monitor_relation_key(right))
+                })
+        });
+
+        let mut data = Vec::with_capacity(events.len() * MONITOR_EVENT_BYTES);
+        for event in events {
+            let Some(monitored) = Wallet::short_address_to_bytes(&event.monitored_address) else {
+                continue;
+            };
+            let Some(monitoring) = Wallet::short_address_to_bytes(&event.monitoring_address) else {
+                continue;
+            };
+            let Ok(signature) = decode(&event.modified_signature) else {
+                continue;
+            };
+            if signature.len() != Wallet::SIGNATURE_LENGTH {
+                continue;
+            }
+            data.push(event.action);
+            data.extend_from_slice(&monitored);
+            data.extend_from_slice(&monitoring);
+            data.extend_from_slice(&ip_to_binary(&event.target_ip));
+            data.extend_from_slice(&event.modified_timestamp.to_le_bytes());
+            data.extend_from_slice(&event.modified_block.to_le_bytes());
+            data.extend_from_slice(&signature);
         }
+        data
+    }
+
+    pub async fn signed_monitor_state() -> RpcResponse {
+        RpcResponse::Binary(Self::signed_monitor_state_bytes().await)
+    }
+
+    pub(crate) async fn load_monitor_recovery_state(bytes: &[u8]) -> usize {
+        let mut state = MONITOR_EVENT_STATE.lock().await;
+        state.clear();
+        let mut loaded = 0;
+        for chunk in bytes.chunks_exact(MONITOR_EVENT_BYTES) {
+            let Some(edit) = Self::monitor_event_from_bytes(chunk) else {
+                continue;
+            };
+            // A restarted node no longer has the live sockets represented by
+            // old additions. Retain signed removals as historical deletion
+            // evidence; current additions are recreated by operational peers.
+            if edit.action != MONITOR_ACTION_REMOVE {
+                continue;
+            }
+            state.insert(Self::monitor_relation_key(&edit), edit);
+            loaded += 1;
+        }
+        loaded
+    }
+
+    pub fn monitor_event_from_bytes(bytes: &[u8]) -> Option<SignedMonitorEdit> {
+        if bytes.len() != MONITOR_EVENT_BYTES {
+            return None;
+        }
+        let monitored_address = Wallet::bytes_to_short_address(
+            &bytes[MONITOR_EVENT_MONITORED_OFFSET..MONITOR_EVENT_MONITORING_OFFSET],
+        )?;
+        let monitoring_address = Wallet::bytes_to_short_address(
+            &bytes[MONITOR_EVENT_MONITORING_OFFSET..MONITOR_EVENT_TARGET_IP_OFFSET],
+        )?;
+        Some(SignedMonitorEdit {
+            action: bytes[MONITOR_EVENT_ACTION_OFFSET],
+            monitored_address,
+            monitoring_address,
+            target_ip: crate::common::binary_conversions::binary_to_ip(
+                bytes[MONITOR_EVENT_TARGET_IP_OFFSET..MONITOR_EVENT_TIMESTAMP_OFFSET].to_vec(),
+            ),
+            modified_timestamp: u64::from_le_bytes(
+                bytes[MONITOR_EVENT_TIMESTAMP_OFFSET..MONITOR_EVENT_BLOCK_OFFSET]
+                    .try_into()
+                    .ok()?,
+            ),
+            modified_block: u32::from_le_bytes(
+                bytes[MONITOR_EVENT_BLOCK_OFFSET..MONITOR_EVENT_SIGNATURE_OFFSET]
+                    .try_into()
+                    .ok()?,
+            ),
+            modified_signature: crate::encode(&bytes[MONITOR_EVENT_SIGNATURE_OFFSET..]),
+        })
     }
 }

src/records/memory/network_mapping/persistence.rs

@@ -13,6 +13,11 @@ fn recovery_snapshot_path() -> PathBuf {
     PathBuf::from(db_path).join("network_map_recovery.bin")
 }
 
+fn monitor_recovery_snapshot_path() -> PathBuf {
+    let (_, _, _, _, _, _, db_path, _, _) = block_extension_and_paths();
+    PathBuf::from(db_path).join("network_monitor_recovery.bin")
+}
+
 impl NodeInfo {
     pub async fn save_recovery_snapshot() -> Result<(), String> {
         let path = recovery_snapshot_path();
@@ -60,7 +65,13 @@ impl NodeInfo {
 
         crate::tokio::fs::write(&path, data)
             .await
-            .map_err(|err| format!("failed to write network-map recovery snapshot: {err}"))
+            .map_err(|err| format!("failed to write network-map recovery snapshot: {err}"))?;
+        crate::tokio::fs::write(
+            monitor_recovery_snapshot_path(),
+            Self::signed_monitor_state_bytes().await,
+        )
+        .await
+        .map_err(|err| format!("failed to write monitor recovery snapshot: {err}"))
     }
 
     pub async fn load_recovery_snapshot() -> Result<usize, String> {
@@ -140,6 +151,17 @@ impl NodeInfo {
 
         let mut map = ADDRESS_MAP.lock().await;
         *map = recovered;
+        drop(map);
+
+        if let Ok(monitor_bytes) = read(monitor_recovery_snapshot_path()).await {
+            if monitor_bytes.len()
+                % crate::records::memory::network_mapping::structs::MONITOR_EVENT_BYTES
+                != 0
+            {
+                warn!("[network_map] monitor recovery snapshot had trailing partial bytes");
+            }
+            Self::load_monitor_recovery_state(&monitor_bytes).await;
+        }
         Ok(loaded)
     }
 

src/records/memory/network_mapping/queries.rs

@@ -95,10 +95,11 @@ impl NodeInfo {
                 Wallet::short_address_to_bytes(&node_info.added_by).unwrap_or_default();
             let added_timestamp_bytes = node_info.added_timestamp.to_le_bytes();
 
-            let deleted_timestamp_bytes = node_info.deleted_timestamp.to_le_bytes();
-            let deleted_block_bytes = node_info.deleted_block.to_le_bytes();
-            let monitor_count = node_info.monitoring.len().min(u16::MAX as usize) as u16;
-            let monitor_count_bytes = monitor_count.to_le_bytes();
+            // Network-map snapshots contain signed membership only. Liveness
+            // is transferred separately as signed monitor assertions.
+            let deleted_timestamp_bytes = 0_u64.to_le_bytes();
+            let deleted_block_bytes = 0_u32.to_le_bytes();
+            let monitor_count_bytes = 0_u16.to_le_bytes();
 
             let added_signature_bytes = decode(node_info.added_signature.clone()).unwrap();
 
@@ -118,13 +119,6 @@ impl NodeInfo {
             data.extend_from_slice(&deleted_timestamp_bytes);
             data.extend_from_slice(&deleted_block_bytes);
             data.extend_from_slice(&monitor_count_bytes);
-            for monitor in node_info.monitoring.iter().take(monitor_count as usize) {
-                if let Some(monitor_bytes) = Wallet::short_address_to_bytes(monitor) {
-                    data.extend_from_slice(&monitor_bytes);
-                } else {
-                    data.extend_from_slice(&[0u8; Wallet::SHORT_ADDRESS_BYTES_LENGTH]);
-                }
-            }
         }
         RpcResponse::Binary(data)
     }

src/records/memory/network_mapping/structs.rs

@@ -25,6 +25,18 @@ pub const NODE_DELETED_BLOCK_OFFSET: usize = NODE_DELETED_TIMESTAMP_OFFSET + NOD
 pub const NODE_MONITOR_COUNT_OFFSET: usize = NODE_DELETED_BLOCK_OFFSET + NODE_DELETED_BLOCK_BYTES;
 pub const NODE_RECORD_FIXED_BYTES: usize = NODE_MONITOR_COUNT_OFFSET + NODE_MONITOR_COUNT_BYTES;
 
+pub const MONITOR_EVENT_ACTION_OFFSET: usize = 0;
+pub const MONITOR_EVENT_MONITORED_OFFSET: usize = MONITOR_EVENT_ACTION_OFFSET + 1;
+pub const MONITOR_EVENT_MONITORING_OFFSET: usize =
+    MONITOR_EVENT_MONITORED_OFFSET + Wallet::SHORT_ADDRESS_BYTES_LENGTH;
+pub const MONITOR_EVENT_TARGET_IP_OFFSET: usize =
+    MONITOR_EVENT_MONITORING_OFFSET + Wallet::SHORT_ADDRESS_BYTES_LENGTH;
+pub const MONITOR_EVENT_TIMESTAMP_OFFSET: usize = MONITOR_EVENT_TARGET_IP_OFFSET + NODE_IP_BYTES;
+pub const MONITOR_EVENT_BLOCK_OFFSET: usize = MONITOR_EVENT_TIMESTAMP_OFFSET + NODE_TIMESTAMP_BYTES;
+pub const MONITOR_EVENT_SIGNATURE_OFFSET: usize =
+    MONITOR_EVENT_BLOCK_OFFSET + NODE_DELETED_BLOCK_BYTES;
+pub const MONITOR_EVENT_BYTES: usize = MONITOR_EVENT_SIGNATURE_OFFSET + Wallet::SIGNATURE_LENGTH;
+
 // SignedNodeEdit carries the signed node membership payload used by add/delete updates.
 #[derive(Debug, Clone)]
 pub struct SignedNodeEdit {
@@ -43,6 +55,7 @@ pub struct SignedMonitorEdit {
     pub monitoring_address: String,
     pub target_ip: String,
     pub modified_timestamp: u64,
+    pub modified_block: u32,
     pub modified_signature: String,
 }
 

src/records/memory/response_channels.rs

@@ -99,6 +99,21 @@ pub async fn reserve_entry_with_context(
             );
             (tx, rx)
         };
+        let cleanup_map = map.clone();
+        tokio::spawn(async move {
+            // Normal replies retire their UID in route_reply. This fallback
+            // covers send failures and cancellation of the requesting task.
+            crate::sleep(Duration::from_secs(60)).await;
+            let still_active = {
+                let map = cleanup_map.lock().await;
+                map.get(&key)
+                    .map(|entry| entry.expires_at.is_none())
+                    .unwrap_or(false)
+            };
+            if still_active {
+                delete_entry(cleanup_map, key).await;
+            }
+        });
         return (key, tx, rx);
     }
 }
@@ -181,3 +196,20 @@ pub async fn delete_entry(map: Arc<Mutex<Command>>, key: Byte3) {
         channel_pair.expires_at = Some(Instant::now() + Duration::from_secs(30));
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[tokio::test(start_paused = true)]
+    async fn abandoned_response_uid_is_retired_automatically() {
+        let map = Arc::new(Mutex::new(Command::new()));
+        let (uid, _tx, _rx) = reserve_entry(map.clone()).await;
+
+        tokio::task::yield_now().await;
+        tokio::time::advance(Duration::from_secs(61)).await;
+        tokio::task::yield_now().await;
+
+        assert!(is_retired_entry(map, uid).await);
+    }
+}

src/records/memory/structs.rs

@@ -7,6 +7,14 @@ use crate::Mutex;
 use crate::Ordering;
 use crate::TcpStream;
 
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum ConnectionHealth {
+    Connected,
+    Busy,
+    Unresponsive,
+    Closed,
+}
+
 // ConnectionInfo stores the live metadata associated with one tracked peer.
 #[derive(Debug, Clone)]
 pub struct ConnectionInfo {
@@ -18,7 +26,11 @@ pub struct ConnectionInfo {
     pub wallet_short_address: String,
     pub wallet_registry_synced: bool,
     pub network_map_synced: bool,
+    pub local_setup_complete: bool,
+    pub remote_setup_complete: bool,
+    pub local_setup_acknowledged: bool,
     pub ready: bool,
+    pub health: ConnectionHealth,
 }
 
 // ConnectionKey is the stable lookup key used inside the in-memory connection map.
@@ -213,7 +225,11 @@ impl ConnectionInfo {
             wallet_short_address,
             wallet_registry_synced: false,
             network_map_synced: false,
+            local_setup_complete: false,
+            remote_setup_complete: false,
+            local_setup_acknowledged: false,
             ready: false,
+            health: ConnectionHealth::Connected,
         }
     }
 }

src/rpc/client/handshake.rs

@@ -157,8 +157,15 @@ async fn perform_handshake(mut params: Handshake) -> io::Result<()> {
 }
 
 async fn send_and_receive_handshake(params: &mut Handshake, data: &[u8]) -> io::Result<Vec<u8>> {
-    params.stream.write_all(data).await?;
-    params.stream.flush().await?;
+    timeout(
+        Duration::from_secs(HANDSHAKE_RESPONSE_TIMEOUT_SECONDS),
+        async {
+            params.stream.write_all(data).await?;
+            params.stream.flush().await
+        },
+    )
+    .await
+    .map_err(|_| io::Error::new(io::ErrorKind::TimedOut, "Timed out sending handshake"))??;
 
     let mut buffer = vec![0u8; HANDSHAKE_RESPONSE_BYTES];
     let mut total_read = 0usize;

src/rpc/client/handshake_processing.rs

@@ -5,22 +5,18 @@ use crate::config::SETTINGS;
 use crate::encode;
 use crate::io;
 use crate::log::{info, warn};
-use crate::miner::flag::{
-    clear_mining_stop_request, request_mining_stop, set_mining_state, set_node_mode, MiningState,
-    NodeMode,
-};
+use crate::miner::flag::{begin_chain_sync, is_normal_mode};
 use crate::orphans::structs::OrphanCheckup2;
 use crate::orphans::sync_check::sync_checkup;
 use crate::orphans::torrent_candidates::hydrate_torrent_candidates;
 use crate::records::block_height::get_block_height::get_height;
 use crate::records::memory::connections::{
     mark_peer_network_map_synced, mark_peer_operational, mark_peer_wallet_registry_synced,
-    set_reconnect_context, startup_synced_peer_streams, CONNECTIONS,
+    spawn_peer_setup_retry, startup_synced_peer_streams, CONNECTIONS,
 };
-use crate::records::memory::enums::{ClientType, ConnectionType};
 use crate::records::memory::network_mapping::NodeInfo;
 use crate::records::memory::response_channels::Command;
-use crate::records::memory::structs::{Connection, StoreConnectionParams};
+use crate::records::memory::structs::Connection;
 use crate::rpc::client::genesis_compat::ensure_compatible_genesis;
 use crate::rpc::client::handshake::connect_and_handshake;
 use crate::rpc::client::register_wallet::register_connected_wallet;
@@ -31,7 +27,9 @@ use crate::rpc::handshake_constants::{
     HANDSHAKE_ADDRESS_OFFSET, HANDSHAKE_MESSAGE_BYTES, HANDSHAKE_RESPONSE_BYTES,
     HANDSHAKE_SIGNATURE_OFFSET,
 };
-use crate::rpc::server::connection_memory_manager::remove_key_from_memory;
+use crate::rpc::server::connection_memory_manager::{
+    remove_stream_from_memory, write_outgoing_miner_to_memory,
+};
 use crate::rpc::server::rpc_command_loop::start_loop;
 use crate::sled::Db;
 use crate::sleep;
@@ -59,23 +57,18 @@ pub struct BootstrapParams {
 
 pub fn spawn_bootstrap_peer_discovery(params: BootstrapParams) {
     tokio::spawn(async move {
-        let run_startup_sync = params.run_startup_sync;
         if let Err(e) = bootstrap_peer_discovery(params).await {
-            if !run_startup_sync {
-                set_node_mode(NodeMode::Normal);
-                clear_mining_stop_request();
-                set_mining_state(MiningState::Idle);
-            }
             eprintln!("[bootstrap] error: {e}");
         }
     });
 }
 
 pub async fn bootstrap_peer_discovery(mut params: BootstrapParams) -> Result<(), String> {
-    if params.run_startup_sync {
-        set_node_mode(NodeMode::Syncing);
-        request_mining_stop();
-    }
+    let chain_sync_guard = if params.run_startup_sync {
+        Some(begin_chain_sync().await)
+    } else {
+        None
+    };
     let (_, _, local_endpoint) = get_ip_and_port().await;
     let max = SETTINGS.outgoing_connections;
     let mut current_key = params.connections_key.clone();
@@ -89,7 +82,7 @@ pub async fn bootstrap_peer_discovery(mut params: BootstrapParams) -> Result<(),
             let connections = CONNECTIONS.read().await;
             connections
                 .as_ref()
-                .map(|connection| connection.count_outgoing_connections())
+                .map(|connection| connection.count_ready_outgoing_connections())
                 .unwrap_or(0)
         };
         if outgoing_connections >= max as usize {
@@ -221,13 +214,15 @@ pub async fn bootstrap_peer_discovery(mut params: BootstrapParams) -> Result<(),
 
     info!("[sync] post-sync checks complete, mining grace period started");
     for (peer_key, _) in startup_synced_peer_streams().await {
-        mark_peer_operational(&peer_key, params.map.clone()).await;
+        if !mark_peer_operational(&peer_key, params.map.clone()).await {
+            spawn_peer_setup_retry(peer_key, params.map.clone());
+        }
     }
     sleep(Duration::from_secs(15)).await;
     info!("[sync] mining grace period complete, mining resuming");
-    set_node_mode(NodeMode::Normal);
-    clear_mining_stop_request();
-    set_mining_state(MiningState::Idle);
+    if let Some(guard) = chain_sync_guard {
+        guard.finish();
+    }
     Ok(())
 }
 
@@ -296,27 +291,17 @@ pub async fn process_handshake_response(
     // create and arc mutex of the stream
     let stream = Arc::new(Mutex::new(params.stream));
     let connections_key = params.addr.clone();
-    let socket_parts: Vec<&str> = params.addr.split(':').collect();
-    if socket_parts.len() == 2 {
-        let ip = socket_parts[0];
-        let port: u16 = socket_parts[1].parse().unwrap_or(0);
-        set_reconnect_context(params.db.clone(), params.wallet.clone(), params.map.clone()).await;
-        let mut conn = CONNECTIONS.write().await;
-        if let Some(manager) = conn.as_mut() {
-            if !manager.store_connection(StoreConnectionParams {
-                connection_type: ConnectionType::Outgoing,
-                ip: ip.to_string(),
-                port,
-                stream: Arc::clone(&stream),
-                client_type: ClientType::Miner,
-                wallet_short_address: returned_short_address.clone(),
-                command_map: params.map.clone(),
-            }) {
-                return Err(io::Error::other(
-                    "The connection is already in the connection manager Please wait 10 minutes and try again",
-                ));
-            }
-        }
+    if !write_outgoing_miner_to_memory(
+        &params.addr,
+        stream.clone(),
+        returned_short_address,
+        params.map.clone(),
+    )
+    .await
+    {
+        return Err(io::Error::other(
+            "The connection is already in the connection manager Please wait 10 minutes and try again",
+        ));
     }
     let listener_stream = Arc::clone(&stream);
     tokio::spawn(start_loop(
@@ -337,7 +322,7 @@ pub async fn process_handshake_response(
     )
     .await
     {
-        remove_key_from_memory(&connections_key).await;
+        remove_stream_from_memory(&stream).await;
         return Err(io::Error::other(format!(
             "Wallet registry sync failed after handshake: {err}"
         )));
@@ -353,7 +338,7 @@ pub async fn process_handshake_response(
     )
     .await
     {
-        remove_key_from_memory(&connections_key).await;
+        remove_stream_from_memory(&stream).await;
         return Err(io::Error::other(format!(
             "Wallet registration failed after handshake: {err}"
         )));
@@ -366,9 +351,9 @@ pub async fn process_handshake_response(
     )
     .await
     .map_err(|err| {
-        let connections_key = connections_key.clone();
+        let failed_stream = stream.clone();
         tokio::spawn(async move {
-            remove_key_from_memory(&connections_key).await;
+            remove_stream_from_memory(&failed_stream).await;
         });
         io::Error::other(format!(
             "Remote height lookup failed before network join: {err}"
@@ -382,9 +367,9 @@ pub async fn process_handshake_response(
     )
     .await
     .map_err(|err| {
-        let connections_key = connections_key.clone();
+        let failed_stream = stream.clone();
         tokio::spawn(async move {
-            remove_key_from_memory(&connections_key).await;
+            remove_stream_from_memory(&failed_stream).await;
         });
         io::Error::other(format!("Genesis compatibility check failed: {err}"))
     })?;
@@ -412,9 +397,9 @@ pub async fn process_handshake_response(
     )
     .await
     .map_err(|err| {
-        let connections_key = connections_key.clone();
+        let failed_stream = stream.clone();
         tokio::spawn(async move {
-            remove_key_from_memory(&connections_key).await;
+            remove_stream_from_memory(&failed_stream).await;
         });
         io::Error::other(format!("Network join/map sync failed: {err}"))
     })?;
@@ -433,8 +418,10 @@ pub async fn process_handshake_response(
 
         spawn_bootstrap_peer_discovery(bsparams);
     } else {
-        if crate::miner::flag::is_normal_mode() {
-            mark_peer_operational(&connections_key, params.map.clone()).await;
+        if is_normal_mode() {
+            if !mark_peer_operational(&connections_key, params.map.clone()).await {
+                spawn_peer_setup_retry(connections_key.clone(), params.map.clone());
+            }
         }
     }
     Ok(())

src/rpc/command_maps.rs

@@ -67,6 +67,8 @@ pub const RPC_NETWORK_MONITOR_ADD: u8 = 50;
 pub const RPC_NETWORK_MONITOR_REMOVE: u8 = 51;
 pub const RPC_STORAGE_LOOKUP_COST: u8 = 52;
 pub const RPC_STORAGE_LOOKUP: u8 = 53;
+pub const RPC_NETWORK_MONITOR_STATE: u8 = 54;
+pub const RPC_SETUP_COMPLETE: u8 = 55;
 pub const RPC_REPLY: u8 = 255;
 pub const MAX_RPC_REPLY_BYTES: usize = 64 * 1024 * 1024;
 

src/rpc/commands/add_network_node.rs

@@ -52,16 +52,8 @@ pub async fn add_network_node(
     let monitor_count =
         read_bytes_from_stream::read_u16_from_stream(connections_key, stream_locked.clone()).await?
             as usize;
-    let mut monitors = Vec::with_capacity(monitor_count);
-    for _ in 0..monitor_count {
-        let monitor_bytes = read_bytes_from_stream::read_short_address_from_stream(
-            connections_key,
-            stream_locked.clone(),
-        )
-        .await?;
-        if let Some(monitor) = Wallet::bytes_to_short_address(&monitor_bytes) {
-            monitors.push(monitor);
-        }
+    if monitor_count != 0 {
+        return Err("error: Node membership updates cannot contain monitor state".to_string());
     }
     let remote_ip = read_bytes_from_stream::read_caller_ip(stream_locked).await?;
 
@@ -77,7 +69,7 @@ pub async fn add_network_node(
             modified_timestamp: added_timestamp,
             modified_signature: added_signature,
         },
-        monitors,
+        monitors: Vec::new(),
         blocks_mined: 0_u8,
         remote_ip,
         db: db.clone(),

src/rpc/commands/mod.rs

@@ -21,6 +21,7 @@ pub mod memory_by_signature;
 pub mod network_info;
 pub mod network_monitor_add;
 pub mod network_monitor_remove;
+pub mod network_monitor_state;
 pub mod nft_list;
 pub mod nft_lookup;
 pub mod receive_torrent;

src/rpc/commands/network_monitor_add.rs

@@ -42,6 +42,9 @@ pub async fn network_monitor_add(
     let modified_timestamp =
         read_bytes_from_stream::read_u64_from_stream(connections_key, stream_locked.clone())
             .await?;
+    let modified_block =
+        read_bytes_from_stream::read_u32_from_stream(connections_key, stream_locked.clone())
+            .await?;
     let modified_signature =
         read_bytes_from_stream::read_signature_from_stream(connections_key, stream_locked.clone())
             .await?;
@@ -62,6 +65,7 @@ pub async fn network_monitor_add(
             monitoring_address,
             target_ip,
             modified_timestamp,
+            modified_block,
             modified_signature,
         },
         remote_ip,

src/rpc/commands/network_monitor_remove.rs

@@ -42,6 +42,9 @@ pub async fn network_monitor_remove(
     let modified_timestamp =
         read_bytes_from_stream::read_u64_from_stream(connections_key, stream_locked.clone())
             .await?;
+    let modified_block =
+        read_bytes_from_stream::read_u32_from_stream(connections_key, stream_locked.clone())
+            .await?;
     let modified_signature =
         read_bytes_from_stream::read_signature_from_stream(connections_key, stream_locked.clone())
             .await?;
@@ -62,6 +65,7 @@ pub async fn network_monitor_remove(
             monitoring_address,
             target_ip,
             modified_timestamp,
+            modified_block,
             modified_signature,
         },
         remote_ip,

src/rpc/commands/network_monitor_state.rs

@@ -0,0 +1,6 @@
+use crate::records::memory::network_mapping::NodeInfo;
+use crate::rpc::responses::RpcResponse;
+
+pub async fn network_monitor_state() -> RpcResponse {
+    NodeInfo::signed_monitor_state().await
+}

src/rpc/commands/receive_torrent.rs

@@ -1,11 +1,7 @@
 use crate::common::check_genesis::genesis_checkup;
 use crate::common::skein::skein_128_hash_bytes;
 use crate::log::{error, info, warn};
-use crate::miner::flag::{
-    clear_mining_stop_request, is_normal_mode, is_reorganizing_mode, is_syncing_mode,
-    request_mining_stop, set_mining_state, set_node_mode, wait_for_mining_idle, MiningState,
-    NodeMode,
-};
+use crate::miner::flag::{begin_chain_sync, is_normal_mode, is_reorganizing_mode, is_syncing_mode};
 use crate::orphans::checkup_state::{
     finish_orphan_check, request_orphan_recheck, try_begin_orphan_check,
 };
@@ -79,9 +75,7 @@ async fn run_live_catchup_sync(
     info!(
         "[sync] live catch-up started: local_height={local_height} remote_height={remote_height}"
     );
-    set_node_mode(NodeMode::Syncing);
-    request_mining_stop();
-    wait_for_mining_idle().await;
+    let chain_sync_guard = begin_chain_sync().await;
 
     let sync_result = node_syncing(
         stream.clone(),
@@ -124,9 +118,7 @@ async fn run_live_catchup_sync(
         }
     }
 
-    set_node_mode(NodeMode::Normal);
-    clear_mining_stop_request();
-    set_mining_state(MiningState::Idle);
+    chain_sync_guard.finish();
     if sync_result.is_ok() {
         info!("[sync] live catch-up complete, normal mode restored");
     }

src/rpc/commands/tx_submit.rs

@@ -40,6 +40,7 @@ async fn broadcast_tx(tx_bytes: Vec<u8>, map: Arc<Mutex<Command>>) {
     // Broadcast newly accepted mempool transactions only to miner peers,
     // since those are the nodes that need transaction fan-out most.
     let nodes = get_live_node_broadcast_peers().await;
+    let mut messages = Vec::with_capacity(nodes.len());
 
     for (key, stream) in nodes {
         let client_type = get_client_type_from_memory(&key)
@@ -61,8 +62,9 @@ async fn broadcast_tx(tx_bytes: Vec<u8>, map: Arc<Mutex<Command>>) {
         message.push(RPC_SUBMIT_TRANSACTION);
         message.extend_from_slice(&uid);
         message.extend_from_slice(&tx_bytes);
-        RpcResponse::send_raw(&stream, Some(&key), &message).await;
+        messages.push((key, stream, message));
     }
+    RpcResponse::broadcast_raw(messages).await;
 }
 
 macro_rules! submit_storage_value {

src/rpc/commands/wallet_register.rs

@@ -33,6 +33,7 @@ async fn broadcast_wallet_registration(
             .unwrap_or_default()
     };
 
+    let mut messages = Vec::with_capacity(streams.len());
     for (peer_key, unlocked_stream) in streams {
         if let Some((peer_ip, _)) = peer_key.rsplit_once(':') {
             if !remote_ip.is_empty() && peer_ip == remote_ip {
@@ -59,8 +60,9 @@ async fn broadcast_wallet_registration(
         message.extend_from_slice(&public_key_bytes);
         message.extend_from_slice(&signature_bytes);
 
-        RpcResponse::send_raw(&unlocked_stream, Some(&peer_key), &message).await;
+        messages.push((peer_key, unlocked_stream, message));
     }
+    RpcResponse::broadcast_raw(messages).await;
 }
 
 pub async fn register(

src/rpc/read_bytes_from_stream.rs

@@ -11,7 +11,8 @@ use crate::{timeout, AsyncReadExt, AsyncWriteExt, Duration};
 const STREAM_READ_TIMEOUT_SECONDS: u64 = 30;
 
 async fn read_exact_from_stream(
-    key: &str,
+    _key: &str,
+    stream_locked: &Arc<Mutex<TcpStream>>,
     stream: &mut TcpStream,
     buffer: &mut [u8],
 ) -> Result<(), String> {
@@ -28,14 +29,14 @@ async fn read_exact_from_stream(
             if let Err(e) = stream.shutdown().await {
                 warn!("Error shutting down stream: {e}");
             }
-            connection_memory_manager::remove_key_from_memory(key).await;
+            connection_memory_manager::remove_stream_from_memory(stream_locked).await;
             Err(err.to_string())
         }
         Err(_) => {
             if let Err(e) = stream.shutdown().await {
                 warn!("Error shutting down stream: {e}");
             }
-            connection_memory_manager::remove_key_from_memory(key).await;
+            connection_memory_manager::remove_stream_from_memory(stream_locked).await;
             Err(format!(
                 "Timed out reading from stream after {STREAM_READ_TIMEOUT_SECONDS} seconds"
             ))
@@ -63,7 +64,7 @@ pub async fn read_first_byte(
     // if the peer disconnects mid-read.
     let mut stream = stream_locked.lock().await;
     let mut buffer = [0u8; 1];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(u8::from_le_bytes(buffer))
 }
@@ -76,7 +77,7 @@ pub async fn read_uid_from_stream(
     // can be matched through the shared hashmap.
     let mut stream = stream_locked.lock().await;
     let mut uid_bytes = [0u8; 3];
-    read_exact_from_stream(key, &mut stream, &mut uid_bytes).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut uid_bytes).await?;
     let uid = u32::from_le_bytes([0, uid_bytes[0], uid_bytes[1], uid_bytes[2]]);
     drop(stream);
     Ok((uid, uid_bytes))
@@ -88,7 +89,7 @@ pub async fn read_u8_from_stream(
 ) -> Result<u8, String> {
     let mut stream = stream_locked.lock().await;
     let mut buffer = [0u8; 1];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(u8::from_le_bytes(buffer))
 }
@@ -99,7 +100,7 @@ pub async fn read_u16_from_stream(
 ) -> Result<u16, String> {
     let mut stream = stream_locked.lock().await;
     let mut buffer = [0u8; 2];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(u16::from_le_bytes(buffer))
 }
@@ -110,7 +111,7 @@ pub async fn read_u32_from_stream(
 ) -> Result<u32, String> {
     let mut stream = stream_locked.lock().await;
     let mut buffer = [0u8; 4];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(u32::from_le_bytes(buffer))
 }
@@ -121,7 +122,7 @@ pub async fn read_u64_from_stream(
 ) -> Result<u64, String> {
     let mut stream = stream_locked.lock().await;
     let mut buffer = [0u8; 8];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(u64::from_le_bytes(buffer))
 }
@@ -132,7 +133,7 @@ pub async fn read_u128_from_stream(
 ) -> Result<u128, String> {
     let mut stream = stream_locked.lock().await;
     let mut buffer = [0u8; 16];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(u128::from_le_bytes(buffer))
 }
@@ -146,7 +147,7 @@ pub async fn read_usize_from_stream(
     // leaving interpretation to the calling command handler.
     let mut stream = stream_locked.lock().await;
     let mut buffer = vec![0u8; bytes];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(buffer)
 }
@@ -157,7 +158,7 @@ pub async fn read_public_key_from_stream(
 ) -> Result<Vec<u8>, String> {
     let mut stream = stream_locked.lock().await;
     let mut buffer = vec![0u8; Wallet::PUBLIC_KEY_LENGTH];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(buffer)
 }
@@ -170,7 +171,7 @@ pub async fn read_short_address_from_stream(
     // wire and stay as raw bytes until the command handler validates them.
     let mut stream = stream_locked.lock().await;
     let mut buffer = vec![0u8; Wallet::SHORT_ADDRESS_BYTES_LENGTH];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     Ok(buffer)
 }
@@ -192,7 +193,7 @@ pub async fn read_ip_from_stream(
     // the node list and peer-management RPC calls.
     let mut stream = stream_locked.lock().await;
     let mut buffer = vec![0u8; 16];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     let ip = binary_to_ip(buffer.to_vec());
     Ok(ip)
@@ -206,7 +207,7 @@ pub async fn read_signature_from_stream(
     // into hex strings for downstream verification logic.
     let mut stream = stream_locked.lock().await;
     let mut buffer = vec![0u8; Wallet::SIGNATURE_LENGTH];
-    read_exact_from_stream(key, &mut stream, &mut buffer).await?;
+    read_exact_from_stream(key, &stream_locked, &mut stream, &mut buffer).await?;
     drop(stream);
     let signature = encode(buffer);
     Ok(signature)

src/rpc/responses.rs

@@ -1,12 +1,21 @@
 use crate::log::warn;
+use crate::records::memory::connections::set_stream_health;
+use crate::records::memory::structs::ConnectionHealth;
 use crate::rpc::command_maps::{MAX_RPC_REPLY_BYTES, RPC_REPLY};
-use crate::rpc::server::connection_memory_manager::remove_key_from_memory;
+use crate::rpc::server::connection_memory_manager::remove_stream_from_memory;
+use crate::timeout;
 use crate::Arc;
 use crate::AsyncWriteExt;
+use crate::Duration;
 use crate::Mutex;
 use crate::Serialize;
 use crate::TcpStream;
 use std::io::ErrorKind;
+use tokio::sync::Semaphore;
+use tokio::task::JoinSet;
+
+const NETWORK_WRITE_TIMEOUT: Duration = Duration::from_secs(10);
+const MAX_CONCURRENT_BROADCAST_WRITES: usize = 16;
 
 #[derive(Debug, Serialize)]
 pub enum RpcResponse {
@@ -25,27 +34,79 @@ impl RpcResponse {
         )
     }
 
+    async fn write_bytes(
+        stream: &Arc<Mutex<TcpStream>>,
+        connections_key: Option<&str>,
+        bytes: &[u8],
+    ) -> bool {
+        let peer = connections_key.unwrap_or("unknown");
+        let mut stream_guard = match timeout(NETWORK_WRITE_TIMEOUT, stream.lock()).await {
+            Ok(guard) => guard,
+            Err(_) => {
+                set_stream_health(stream, ConnectionHealth::Busy).await;
+                warn!("[rpc] timed out waiting for stream lock: peer={peer}");
+                return false;
+            }
+        };
+
+        let write_result = timeout(NETWORK_WRITE_TIMEOUT, async {
+            stream_guard.write_all(bytes).await?;
+            stream_guard.flush().await
+        })
+        .await;
+
+        match write_result {
+            Ok(Ok(())) => {
+                drop(stream_guard);
+                set_stream_health(stream, ConnectionHealth::Connected).await;
+                true
+            }
+            Ok(Err(err)) => {
+                if !Self::is_expected_disconnect(&err) {
+                    warn!("Error sending binary response to {peer}: {err:?}");
+                }
+                let _ = stream_guard.shutdown().await;
+                drop(stream_guard);
+                set_stream_health(stream, ConnectionHealth::Closed).await;
+                remove_stream_from_memory(stream).await;
+                false
+            }
+            Err(_) => {
+                drop(stream_guard);
+                set_stream_health(stream, ConnectionHealth::Unresponsive).await;
+                warn!("[rpc] socket write timed out: peer={peer}");
+                false
+            }
+        }
+    }
+
     pub async fn send_raw(
         stream: &Arc<Mutex<TcpStream>>,
         connections_key: Option<&str>,
         bytes: &[u8],
-    ) {
+    ) -> bool {
         // Raw sends are used for messages that already include their
         // command framing and payload layout.
-        let mut stream_guard = stream.lock().await;
-        if let Err(err) = stream_guard.write_all(bytes).await {
-            if !Self::is_expected_disconnect(&err) {
-                warn!("Error sending binary response: {err:?}");
-            }
-            if let Some(key) = connections_key {
-                remove_key_from_memory(key).await;
-            }
-            let _ = stream_guard.shutdown().await;
-            return;
+        Self::write_bytes(stream, connections_key, bytes).await
+    }
+
+    pub async fn broadcast_raw(messages: Vec<(String, Arc<Mutex<TcpStream>>, Vec<u8>)>) {
+        let permits = Arc::new(Semaphore::new(MAX_CONCURRENT_BROADCAST_WRITES));
+        let mut writes = JoinSet::new();
+
+        for (peer_key, stream, message) in messages {
+            let permits = permits.clone();
+            writes.spawn(async move {
+                let Ok(_permit) = permits.acquire_owned().await else {
+                    return;
+                };
+                Self::send_raw(&stream, Some(&peer_key), &message).await;
+            });
         }
-        if let Err(err) = stream_guard.flush().await {
-            if !Self::is_expected_disconnect(&err) {
-                warn!("Error flushing stream: {err:?}");
+
+        while let Some(result) = writes.join_next().await {
+            if let Err(err) = result {
+                warn!("[rpc] broadcast write task failed: {err}");
             }
         }
     }
@@ -55,11 +116,9 @@ impl RpcResponse {
         stream: &Arc<Mutex<TcpStream>>,
         connections_key: Option<&str>,
         uid: u32,
-    ) {
+    ) -> bool {
         // Standard RPC responses are framed with command 255 and the
         // request UID so the waiting caller can route the reply.
-        let mut stream_guard = stream.lock().await;
-
         let uid_bytes = uid.to_le_bytes()[1..4].to_vec();
         let command = RPC_REPLY;
         match self {
@@ -82,21 +141,7 @@ impl RpcResponse {
                 response_bytes.extend_from_slice(&uid_bytes);
                 response_bytes.extend_from_slice(&bytes_as_slice);
                 response_bytes.extend_from_slice(payload);
-                if let Err(err) = stream_guard.write_all(&response_bytes).await {
-                    if !Self::is_expected_disconnect(&err) {
-                        warn!("Error sending binary response: {err:?}");
-                    }
-                    if let Some(key) = connections_key {
-                        remove_key_from_memory(key).await;
-                    }
-                    let _ = stream_guard.shutdown().await;
-                    return;
-                }
-            }
-        }
-        if let Err(err) = stream_guard.flush().await {
-            if !Self::is_expected_disconnect(&err) {
-                warn!("Error flushing stream: {err:?}");
+                return Self::write_bytes(stream, connections_key, &response_bytes).await;
             }
         }
     }

src/rpc/server/command_loop_state.rs

@@ -107,10 +107,8 @@ pub async fn next_incoming_command(
         && !peer_accepts_live_relay(connections_key).await
     {
         warn!(
-            "[rpc] closing unsynced miner peer that sent relay command: peer={connections_key} cmd={command}"
+            "[rpc] setup-time relay received before peer became operational: peer={connections_key} cmd={command}"
         );
-        remove_stream_from_memory(&stream_locked).await;
-        return Ok(None);
     }
 
     // Replies and miner sync traffic belong to expected node-to-node

src/rpc/server/connection_memory_manager.rs

@@ -1,9 +1,18 @@
-use crate::records::memory::connections::{spawn_retry_dropped_outgoing, CONNECTIONS};
+use crate::common::binary_conversions::{binary_to_ip, ip_to_binary};
+use crate::log::warn;
+use crate::records::memory::connections::{
+    set_stream_health, spawn_retry_dropped_outgoing, CONNECTIONS,
+};
 use crate::records::memory::enums::{ClientType, ConnectionType};
-use crate::records::memory::response_channels::Command;
-use crate::records::memory::structs::StoreConnectionParams;
+use crate::records::memory::response_channels::{
+    delete_entry, reserve_entry_with_context, Command,
+};
+use crate::records::memory::structs::{ConnectionHealth, StoreConnectionParams};
+use crate::rpc::command_maps::RPC_BLOCK_HEIGHT;
+use crate::timeout;
 use crate::Arc;
 use crate::AsyncWriteExt;
+use crate::Duration;
 use crate::Mutex;
 use crate::TcpStream;
 
@@ -20,6 +29,194 @@ fn split_ip_port_key(value: &str) -> Option<(String, u16)> {
     Some((ip, port))
 }
 
+async fn miner_stream_health(
+    key: &str,
+    stream: Arc<Mutex<TcpStream>>,
+    command_map: Arc<Mutex<Command>>,
+) -> ConnectionHealth {
+    let (checkup_key, _checkup_tx, checkup_rx_mutex) = reserve_entry_with_context(
+        command_map.clone(),
+        Some(RPC_BLOCK_HEIGHT),
+        Some(key.to_string()),
+    )
+    .await;
+
+    let mut message = Vec::with_capacity(4);
+    message.push(RPC_BLOCK_HEIGHT);
+    message.extend_from_slice(&checkup_key);
+
+    let write_result = {
+        let Ok(mut stream_guard) = stream.try_lock() else {
+            delete_entry(command_map, checkup_key).await;
+            set_stream_health(&stream, ConnectionHealth::Busy).await;
+            return ConnectionHealth::Busy;
+        };
+        timeout(Duration::from_secs(5), async {
+            stream_guard.write_all(&message).await?;
+            stream_guard.flush().await
+        })
+        .await
+    };
+
+    match write_result {
+        Ok(Ok(())) => {}
+        Ok(Err(err)) => {
+            warn!("[connection_manager] stale duplicate probe write failed for {key}: {err}");
+            delete_entry(command_map, checkup_key).await;
+            set_stream_health(&stream, ConnectionHealth::Closed).await;
+            return ConnectionHealth::Closed;
+        }
+        Err(_) => {
+            warn!("[connection_manager] stale duplicate probe write timed out for {key}");
+            delete_entry(command_map, checkup_key).await;
+            set_stream_health(&stream, ConnectionHealth::Unresponsive).await;
+            return ConnectionHealth::Unresponsive;
+        }
+    }
+
+    let response_result = {
+        let mut checkup_rx = checkup_rx_mutex.lock().await;
+        timeout(Duration::from_secs(5), checkup_rx.recv()).await
+    };
+    delete_entry(command_map, checkup_key).await;
+
+    let health = if matches!(response_result, Ok(Some(_))) {
+        ConnectionHealth::Connected
+    } else {
+        ConnectionHealth::Unresponsive
+    };
+    set_stream_health(&stream, health).await;
+    health
+}
+
+async fn purge_stale_duplicate_miner(ip: &str, command_map: Arc<Mutex<Command>>) -> bool {
+    let ip_bytes = ip_to_binary(ip);
+    let duplicate = {
+        let guard = CONNECTIONS.read().await;
+        guard.as_ref().and_then(|connection| {
+            connection
+                .connection_map
+                .iter()
+                .find_map(|(connection_key, connection_info)| {
+                    if connection_key.ip != ip_bytes
+                        || ClientType::from_bytes(&connection_info.client_type)
+                            != Some(ClientType::Miner)
+                    {
+                        return None;
+                    }
+                    let connection_type =
+                        ConnectionType::from_bytes(&connection_key.connection_type)?;
+                    let ip = binary_to_ip(connection_key.ip.clone());
+                    Some((
+                        connection_type,
+                        ip,
+                        connection_key.port,
+                        Arc::clone(&connection_info.stream),
+                        connection_info.health,
+                    ))
+                })
+        })
+    };
+
+    let Some((connection_type, duplicate_ip, duplicate_port, duplicate_stream, recorded_health)) =
+        duplicate
+    else {
+        return false;
+    };
+
+    let duplicate_key = format!("{duplicate_ip}:{duplicate_port}");
+    match recorded_health {
+        ConnectionHealth::Connected | ConnectionHealth::Busy => {
+            let health =
+                miner_stream_health(&duplicate_key, duplicate_stream.clone(), command_map).await;
+            if matches!(health, ConnectionHealth::Connected | ConnectionHealth::Busy) {
+                return false;
+            }
+        }
+        ConnectionHealth::Unresponsive | ConnectionHealth::Closed => {}
+    }
+
+    let mut guard = CONNECTIONS.write().await;
+    let Some(connection) = guard.as_mut() else {
+        return false;
+    };
+    let duplicate_ip_bytes = ip_to_binary(&duplicate_ip);
+    let still_same_stream =
+        connection
+            .connection_map
+            .iter()
+            .any(|(connection_key, connection_info)| {
+                connection_key.ip == duplicate_ip_bytes
+                    && connection_key.port == duplicate_port
+                    && ConnectionType::from_bytes(&connection_key.connection_type)
+                        == Some(connection_type)
+                    && Arc::ptr_eq(&connection_info.stream, &duplicate_stream)
+            });
+
+    if !still_same_stream {
+        return false;
+    }
+
+    warn!(
+        "[connection_manager] removing stale duplicate miner stream before accepting reconnect: peer={duplicate_key}"
+    );
+    connection.drop_connection(connection_type, duplicate_ip, duplicate_port);
+    true
+}
+
+pub async fn write_outgoing_miner_to_memory(
+    received_ip_port: &str,
+    stream: Arc<Mutex<TcpStream>>,
+    wallet_short_address: String,
+    command_map: Arc<Mutex<Command>>,
+) -> bool {
+    let Some((ip, port)) = split_ip_port_key(received_ip_port) else {
+        return false;
+    };
+
+    let added = {
+        let mut guard = CONNECTIONS.write().await;
+        guard
+            .as_mut()
+            .map(|connection| {
+                connection.store_connection(StoreConnectionParams {
+                    connection_type: ConnectionType::Outgoing,
+                    ip: ip.clone(),
+                    port,
+                    stream: stream.clone(),
+                    client_type: ClientType::Miner,
+                    wallet_short_address: wallet_short_address.clone(),
+                    command_map: command_map.clone(),
+                })
+            })
+            .unwrap_or(false)
+    };
+    if added {
+        return true;
+    }
+
+    if !purge_stale_duplicate_miner(&ip, command_map.clone()).await {
+        return false;
+    }
+
+    CONNECTIONS
+        .write()
+        .await
+        .as_mut()
+        .map(|connection| {
+            connection.store_connection(StoreConnectionParams {
+                connection_type: ConnectionType::Outgoing,
+                ip,
+                port,
+                stream,
+                client_type: ClientType::Miner,
+                wallet_short_address,
+                command_map,
+            })
+        })
+        .unwrap_or(false)
+}
+
 // write an incoming connection to memory
 // client type defines if connection
 // is another node or a client connecting
@@ -52,26 +249,51 @@ pub async fn write_to_memory(
             port,
             stream: stream.clone(),
             client_type,
-            wallet_short_address,
-            command_map,
+            wallet_short_address: wallet_short_address.clone(),
+            command_map: command_map.clone(),
         });
 
         *connection_instance = Some(connection);
+        drop(connection_instance);
+        if !added
+            && client_type == ClientType::Miner
+            && purge_stale_duplicate_miner(&ip, command_map.clone()).await
+        {
+            let mut retry_connection_instance = CONNECTIONS.write().await;
+            if let Some(mut retry_connection) = retry_connection_instance.take() {
+                let retry_added = retry_connection.store_connection(StoreConnectionParams {
+                    connection_type: ConnectionType::Incoming,
+                    ip: ip.clone(),
+                    port,
+                    stream: stream.clone(),
+                    client_type,
+                    wallet_short_address,
+                    command_map,
+                });
+                *retry_connection_instance = Some(retry_connection);
+                if retry_added {
+                    drop(retry_connection_instance);
+                    return received_ip_port.to_string();
+                }
+            }
+        }
         if added {
             // Return the original ip:port string as the connection key
             // used by stream readers and cleanup paths.
-            drop(connection_instance);
             received_ip_port.to_string()
         } else {
             // Duplicate streams are told why they are being closed before
             // the socket is shut down.
-            drop(connection_instance);
             let duplicate_message =
                 "The connection is already in the connection manager Please wait 10 minutes and try again";
-            let mut stream_guard = stream.lock().await;
-            let _ = stream_guard.write_all(duplicate_message.as_bytes()).await;
-            let _ = stream_guard.flush().await;
-            let _ = stream_guard.shutdown().await;
+            if let Ok(mut stream_guard) = timeout(Duration::from_secs(10), stream.lock()).await {
+                let _ = timeout(Duration::from_secs(10), async {
+                    let _ = stream_guard.write_all(duplicate_message.as_bytes()).await;
+                    let _ = stream_guard.flush().await;
+                    stream_guard.shutdown().await
+                })
+                .await;
+            }
             "false".to_string()
         }
     } else {
@@ -79,28 +301,6 @@ pub async fn write_to_memory(
     }
 }
 
-// delete a connection from memory
-pub async fn remove_key_from_memory(key: &str) {
-    let mut connection_instance = CONNECTIONS.write().await;
-    // A key can represent either direction, so cleanup removes both
-    // incoming and outgoing entries with the same endpoint.
-    let Some((ip, port)) = split_ip_port_key(key) else {
-        return;
-    };
-    if let Some(connection) = connection_instance.as_mut() {
-        connection.drop_connection(ConnectionType::Incoming, ip.clone(), port);
-        let dropped_outgoing =
-            connection.drop_connection(ConnectionType::Outgoing, ip.clone(), port);
-        if dropped_outgoing
-            .as_ref()
-            .and_then(|connection_info| ClientType::from_bytes(&connection_info.client_type))
-            == Some(ClientType::Miner)
-        {
-            spawn_retry_dropped_outgoing(ip, port);
-        }
-    }
-}
-
 pub async fn remove_stream_from_memory(stream: &Arc<Mutex<TcpStream>>) {
     let mut connection_instance = CONNECTIONS.write().await;
     let Some(connection) = connection_instance.as_mut() else {

src/rpc/server/handshake.rs

@@ -1,15 +1,13 @@
 use crate::common::check_genesis::genesis_checkup;
 use crate::log::{error, warn};
-use crate::miner::flag::{
-    clear_mining_stop_request, request_mining_stop, set_mining_state, set_node_mode, MiningState,
-    NodeMode,
-};
+use crate::miner::flag::{begin_chain_sync, ChainOperationGuard};
 use crate::orphans::structs::OrphanCheckup2;
 use crate::orphans::sync_check::sync_checkup;
 use crate::orphans::torrent_candidates::hydrate_torrent_candidates;
 use crate::records::block_height::get_block_height::get_height;
 use crate::records::memory::connections::{
     mark_peer_network_map_synced, mark_peer_operational, mark_peer_wallet_registry_synced,
+    spawn_peer_setup_retry,
 };
 use crate::records::memory::response_channels::generate_uid;
 use crate::records::memory::response_channels::Command;
@@ -19,7 +17,7 @@ use crate::rpc::client::register_wallet::register_connected_wallet;
 use crate::rpc::client::syncing::node_syncing;
 use crate::rpc::client::wallet_registry_sync::sync_wallet_registry_with_retries;
 use crate::rpc::responses::RpcResponse;
-use crate::rpc::server::connection_memory_manager::{remove_key_from_memory, write_to_memory};
+use crate::rpc::server::connection_memory_manager::{remove_stream_from_memory, write_to_memory};
 use crate::rpc::server::handshake_processing::{combine_and_send_data, parse_received_data};
 use crate::rpc::server::handshake_verifications::{connection_count, perform_handshake_tests};
 use crate::rpc::server::structs::{CombineAndSendDataParams, HandshakeTestParams};
@@ -42,9 +40,13 @@ use crate::Utc;
 async fn drop_failed_handshake(stream: &Arc<Mutex<TcpStream>>) {
     // Failed handshakes are never stored in connection memory, but the
     // accepted TCP socket should still be closed immediately.
-    let mut stream_guard = stream.lock().await;
-    let _ = stream_guard.flush().await;
-    let _ = stream_guard.shutdown().await;
+    if let Ok(mut stream_guard) = crate::timeout(Duration::from_secs(10), stream.lock()).await {
+        let _ = crate::timeout(Duration::from_secs(10), async {
+            let _ = stream_guard.flush().await;
+            stream_guard.shutdown().await
+        })
+        .await;
+    }
 }
 
 async fn get_connection_counts() -> (u8, u8) {
@@ -62,7 +64,29 @@ async fn sync_incoming_peer_before_operational(
     wallet: Arc<Wallet>,
     map: Arc<Mutex<Command>>,
     connections_key: &str,
-) -> Result<bool, String> {
+) -> Result<(bool, Option<ChainOperationGuard>), String> {
+    let initial_local_height = get_height(db);
+    let initial_remote_height =
+        request_remote_height(stream.clone(), map.clone(), connections_key.to_string()).await?;
+    ensure_compatible_genesis(
+        stream.clone(),
+        map.clone(),
+        connections_key.to_string(),
+        initial_remote_height,
+    )
+    .await?;
+    let initial_local_genesis_exists = genesis_checkup().await;
+
+    if initial_local_genesis_exists && initial_remote_height < initial_local_height {
+        warn!(
+            "[startup] incoming peer is behind local chain; keeping connection passive until peer catches up: local_height={initial_local_height} remote_height={initial_remote_height}"
+        );
+        return Ok((false, None));
+    }
+
+    // Wait for any existing startup, live catch-up, or orphan operation to
+    // finish before this incoming peer can mutate the canonical chain.
+    let chain_sync_guard = begin_chain_sync().await;
     let local_height = get_height(db);
     let remote_height =
         request_remote_height(stream.clone(), map.clone(), connections_key.to_string()).await?;
@@ -77,16 +101,12 @@ async fn sync_incoming_peer_before_operational(
 
     if local_genesis_exists && remote_height < local_height {
         warn!(
-            "[startup] incoming peer is behind local chain; keeping connection passive until peer catches up: local_height={local_height} remote_height={remote_height}"
+            "[startup] incoming peer fell behind while waiting for chain access; keeping connection passive: local_height={local_height} remote_height={remote_height}"
         );
-        return Ok(false);
+        return Ok((false, None));
     }
 
     if !local_genesis_exists || remote_height > local_height + 10 {
-        set_node_mode(NodeMode::Syncing);
-        request_mining_stop();
-        set_mining_state(MiningState::Idle);
-
         node_syncing(
             stream.clone(),
             db,
@@ -145,7 +165,7 @@ async fn sync_incoming_peer_before_operational(
             .map_err(|err| format!("Incoming post-sync orphan check error: {err}"))?;
     }
 
-    Ok(true)
+    Ok((true, Some(chain_sync_guard)))
 }
 
 fn spawn_incoming_peer_promotion_watcher(
@@ -175,8 +195,9 @@ fn spawn_incoming_peer_promotion_watcher(
                 };
 
             if remote_height >= local_height {
-                mark_peer_operational(&connections_key, map.clone()).await;
-                break;
+                if mark_peer_operational(&connections_key, map.clone()).await {
+                    break;
+                }
             }
         }
     });
@@ -201,7 +222,7 @@ async fn complete_incoming_miner_setup(
     .await
     {
         error!("[startup] incoming peer wallet registry sync failed: {err}");
-        remove_key_from_memory(connections_key).await;
+        remove_stream_from_memory(&stream).await;
         return;
     }
     mark_peer_wallet_registry_synced(connections_key).await;
@@ -216,7 +237,7 @@ async fn complete_incoming_miner_setup(
     .await
     {
         error!("[startup] incoming peer wallet registration failed: {err}");
-        remove_key_from_memory(connections_key).await;
+        remove_stream_from_memory(&stream).await;
         return;
     }
 
@@ -228,7 +249,7 @@ async fn complete_incoming_miner_setup(
                 error!(
                 "[startup] incoming peer remote height lookup failed before network join: {err}"
             );
-                remove_key_from_memory(connections_key).await;
+                remove_stream_from_memory(&stream).await;
                 return;
             }
         };
@@ -241,7 +262,7 @@ async fn complete_incoming_miner_setup(
     .await
     {
         error!("[startup] incoming peer genesis compatibility check failed: {err}");
-        remove_key_from_memory(connections_key).await;
+        remove_stream_from_memory(&stream).await;
         return;
     }
     let self_add_allowed =
@@ -266,7 +287,7 @@ async fn complete_incoming_miner_setup(
         .await
         {
             error!("[startup] incoming peer network map sync failed: {err}");
-            remove_key_from_memory(connections_key).await;
+            remove_stream_from_memory(&stream).await;
             return;
         }
         if let Err(err) = get_network_mapping_for_address(
@@ -280,13 +301,13 @@ async fn complete_incoming_miner_setup(
         .await
         {
             error!("[startup] incoming peer local network record import failed: {err}");
-            remove_key_from_memory(connections_key).await;
+            remove_stream_from_memory(&stream).await;
             return;
         }
     }
     mark_peer_network_map_synced(connections_key).await;
 
-    let operational = match sync_incoming_peer_before_operational(
+    let (operational, chain_sync_guard) = match sync_incoming_peer_before_operational(
         stream.clone(),
         db,
         wallet.clone(),
@@ -298,17 +319,19 @@ async fn complete_incoming_miner_setup(
         Ok(operational) => operational,
         Err(err) => {
             error!("[startup] incoming peer chain sync failed: {err}");
-            remove_key_from_memory(connections_key).await;
+            remove_stream_from_memory(&stream).await;
             return;
         }
     };
 
     if operational {
-        mark_peer_operational(connections_key, map.clone()).await;
+        if !mark_peer_operational(connections_key, map.clone()).await {
+            spawn_peer_setup_retry(connections_key.to_string(), map.clone());
+        }
         sleep(Duration::from_secs(15)).await;
-        set_node_mode(NodeMode::Normal);
-        clear_mining_stop_request();
-        set_mining_state(MiningState::Idle);
+        if let Some(guard) = chain_sync_guard {
+            guard.finish();
+        }
     } else {
         spawn_incoming_peer_promotion_watcher(
             stream.clone(),

src/rpc/server/handshake_processing.rs

@@ -7,7 +7,7 @@ use crate::rpc::handshake_constants::{
     HANDSHAKE_REQUEST_BYTES, HANDSHAKE_RESPONSE_BYTES, HANDSHAKE_SIGNATURE_OFFSET,
     HANDSHAKE_TIME_OFFSET,
 };
-use crate::rpc::server::connection_memory_manager::remove_key_from_memory;
+use crate::rpc::server::connection_memory_manager::remove_stream_from_memory;
 use crate::rpc::server::rpc_command_loop::start_loop;
 use crate::rpc::server::structs::CombineAndSendDataParams;
 use crate::wallets::structures::Wallet;
@@ -108,7 +108,7 @@ pub async fn return_handshake(
     public_key: &str,
     message: &str,
     signed_message: &str,
-    connections_key: &str,
+    _connections_key: &str,
 ) -> Result<(), String> {
     // convert to binary/bytes
     let public_key_bin = Wallet::normalize_saved_public_key_bytes(public_key)
@@ -123,17 +123,24 @@ pub async fn return_handshake(
     data.extend_from_slice(&signed_bin);
     data.extend(&public_key_bin);
 
-    // get stream lock
-    let mut stream_guard = stream.lock().await;
+    // Handshake writes are bounded just like normal RPC writes so a peer
+    // that stops reading cannot hold this setup task indefinitely.
+    let mut stream_guard = crate::timeout(crate::Duration::from_secs(10), stream.lock())
+        .await
+        .map_err(|_| "error: Timed out waiting for handshake stream".to_string())?;
 
-    // write to stream
-    if let Err(err) = stream_guard.write_all(&data).await {
-        let _ = remove_key_from_memory(connections_key).await;
+    let write_result = crate::timeout(crate::Duration::from_secs(10), async {
+        stream_guard.write_all(&data).await?;
+        stream_guard.flush().await
+    })
+    .await;
+    if let Ok(Err(err)) = write_result {
+        remove_stream_from_memory(&stream).await;
         error!("Error writing to stream: {err:?}");
         return Err("error: Error writing to stream".to_string());
-    } else if let Err(err) = stream_guard.flush().await {
-        error!("Error flushing stream: {err:?}");
-        return Err("error: Error flushing stream".to_string());
+    } else if write_result.is_err() {
+        remove_stream_from_memory(&stream).await;
+        return Err("error: Timed out writing handshake response".to_string());
     }
 
     // drop lock

src/rpc/server/rpc_command_loop.rs

@@ -1,12 +1,11 @@
 use crate::common::binary_conversions::binary_to_string;
 use crate::encode;
 use crate::log::warn;
+use crate::records::memory::connections::mark_peer_remote_setup_complete;
 use crate::records::memory::enums::ClientType;
 use crate::records::memory::response_channels::Command;
 use crate::rpc::server::command_loop_state::next_incoming_command;
-use crate::rpc::server::connection_memory_manager::{
-    remove_key_from_memory, remove_stream_from_memory,
-};
+use crate::rpc::server::connection_memory_manager::remove_stream_from_memory;
 use crate::rpc::*;
 use crate::sled::Db;
 use crate::wallets::structures::Wallet;
@@ -558,7 +557,7 @@ pub async fn start_loop(
                     .send(&stream_locked, Some(&connections_key), uid)
                     .await;
                 if should_drop_rejected_miner {
-                    remove_key_from_memory(&connections_key).await;
+                    remove_stream_from_memory(&stream_locked).await;
                     break 'outer Ok(());
                 }
             }
@@ -592,6 +591,32 @@ pub async fn start_loop(
                     .send(&stream_locked, Some(&connections_key), uid)
                     .await;
             }
+            54 => {
+                // return the latest signed state for every known monitor relationship
+                let (uid, _) = read_bytes_from_stream::read_uid_from_stream(
+                    &connections_key,
+                    stream_locked.clone(),
+                )
+                .await?;
+                let result = commands::network_monitor_state::network_monitor_state().await;
+                result
+                    .send(&stream_locked, Some(&connections_key), uid)
+                    .await;
+            }
+            55 => {
+                // Both sides announce setup completion independently. Receipt
+                // records the remote half and the response acknowledges it.
+                let (uid, _) = read_bytes_from_stream::read_uid_from_stream(
+                    &connections_key,
+                    stream_locked.clone(),
+                )
+                .await?;
+                mark_peer_remote_setup_complete(&connections_key, map.clone()).await;
+                let result = responses::RpcResponse::Binary(b"setup_complete_ack".to_vec());
+                result
+                    .send(&stream_locked, Some(&connections_key), uid)
+                    .await;
+            }
             30 => {
                 // request node list
                 let (uid, _) = read_bytes_from_stream::read_uid_from_stream(

src/startup/connections.rs

@@ -4,7 +4,9 @@ use crate::miner::flag::{
     clear_mining_stop_request, is_normal_mode, set_mining_state, set_node_mode, MiningState,
     NodeMode,
 };
-use crate::records::memory::connections::peer_connection_count;
+use crate::records::memory::connections::{
+    peer_connection_count, ready_outgoing_connection_count, refill_outgoing_connections_once,
+};
 use crate::records::memory::response_channels::Command;
 use crate::rpc::client::handshake::connect_and_handshake;
 use crate::rpc::client::structs::Connect;
@@ -144,22 +146,32 @@ pub fn spawn_isolated_bootstrap_recovery(db: Db, wallet: Arc<Wallet>, map: Arc<M
         loop {
             sleep(Duration::from_secs(60)).await;
 
-            if !is_normal_mode() || peer_connection_count().await > 0 {
+            if !is_normal_mode() {
                 continue;
             }
 
-            info!("[reconnect] no operational peers remain; retrying bootstrap recovery");
-            match attempt_bootstrap_connections(
-                db.clone(),
-                wallet.clone(),
-                map.clone(),
-                "reconnect",
-            )
-            .await
-            {
-                Ok(true) => {}
-                Ok(false) => {}
-                Err(err) => warn!("[reconnect] bootstrap recovery aborted: {err}"),
+            if peer_connection_count().await == 0 {
+                info!("[reconnect] no operational peers remain; retrying bootstrap recovery");
+                match attempt_bootstrap_connections(
+                    db.clone(),
+                    wallet.clone(),
+                    map.clone(),
+                    "reconnect",
+                )
+                .await
+                {
+                    Ok(true) => {}
+                    Ok(false) => {}
+                    Err(err) => warn!("[reconnect] bootstrap recovery aborted: {err}"),
+                }
+                continue;
+            }
+
+            // A topology pass is bounded: it walks the current active map once,
+            // tries each eligible endpoint at most once, then sleeps until the
+            // next interval even when the configured limit exceeds network size.
+            if ready_outgoing_connection_count().await < outgoing_connections as usize {
+                refill_outgoing_connections_once().await;
             }
         }
     });

src/startup/network_broadcast.rs

@@ -2,13 +2,16 @@ use crate::common::binary_conversions::{binary_to_ip, binary_to_string, ip_to_bi
 use crate::common::network_startup::get_ip_and_port;
 use crate::log::warn;
 use crate::records::memory::network_mapping::structs::{
-    SignedNodeEdit, NODE_ADDED_BY_OFFSET, NODE_ADDED_SIGNATURE_OFFSET, NODE_ADDED_TIMESTAMP_OFFSET,
-    NODE_BLOCKS_MINED_OFFSET, NODE_DELETED_BLOCK_OFFSET, NODE_DELETED_TIMESTAMP_OFFSET,
-    NODE_IP_OFFSET, NODE_MONITOR_COUNT_OFFSET, NODE_PORT_OFFSET, NODE_RECORD_FIXED_BYTES,
+    SignedNodeEdit, MONITOR_EVENT_BYTES, NODE_ADDED_BY_OFFSET, NODE_ADDED_SIGNATURE_OFFSET,
+    NODE_ADDED_TIMESTAMP_OFFSET, NODE_BLOCKS_MINED_OFFSET, NODE_DELETED_BLOCK_OFFSET,
+    NODE_DELETED_TIMESTAMP_OFFSET, NODE_IP_OFFSET, NODE_MONITOR_COUNT_OFFSET, NODE_PORT_OFFSET,
+    NODE_RECORD_FIXED_BYTES,
 };
 use crate::records::memory::network_mapping::NodeInfo;
 use crate::records::memory::response_channels::{reserve_entry, Command};
-use crate::rpc::command_maps::{RPC_ADD_NETWORK_NODE, RPC_REQUEST_NODE_LIST};
+use crate::rpc::command_maps::{
+    RPC_ADD_NETWORK_NODE, RPC_NETWORK_MONITOR_STATE, RPC_REQUEST_NODE_LIST,
+};
 use crate::rpc::responses::RpcResponse;
 use crate::sled::Db;
 use crate::timeout;
@@ -109,17 +112,25 @@ pub async fn get_network_mapping(
     unlocked_stream: Arc<Mutex<TcpStream>>,
     command_map: Arc<Mutex<Command>>,
     db: &Db,
-    _wallet: Arc<Wallet>,
+    wallet: Arc<Wallet>,
     connections_key: &str,
 ) -> Result<(), String> {
-    get_network_mapping_inner(unlocked_stream, command_map, db, connections_key, None).await
+    get_network_mapping_inner(
+        unlocked_stream,
+        command_map,
+        db,
+        wallet,
+        connections_key,
+        None,
+    )
+    .await
 }
 
 pub async fn get_network_mapping_for_address(
     unlocked_stream: Arc<Mutex<TcpStream>>,
     command_map: Arc<Mutex<Command>>,
     db: &Db,
-    _wallet: Arc<Wallet>,
+    wallet: Arc<Wallet>,
     connections_key: &str,
     only_address: &str,
 ) -> Result<(), String> {
@@ -127,6 +138,7 @@ pub async fn get_network_mapping_for_address(
         unlocked_stream,
         command_map,
         db,
+        wallet,
         connections_key,
         Some(only_address),
     )
@@ -137,6 +149,7 @@ async fn get_network_mapping_inner(
     unlocked_stream: Arc<Mutex<TcpStream>>,
     command_map: Arc<Mutex<Command>>,
     db: &Db,
+    wallet: Arc<Wallet>,
     connections_key: &str,
     only_address: Option<&str>,
 ) -> Result<(), String> {
@@ -166,6 +179,9 @@ async fn get_network_mapping_inner(
                 .try_into()
                 .unwrap(),
         ) as usize;
+        if monitor_count != 0 {
+            return Err("network membership response contained unsigned monitor state".to_string());
+        }
         let record_bytes =
             NODE_RECORD_FIXED_BYTES + (monitor_count * Wallet::SHORT_ADDRESS_BYTES_LENGTH);
         if buffer.len() < record_bytes {
@@ -206,14 +222,10 @@ async fn get_network_mapping_inner(
                 .try_into()
                 .unwrap(),
         );
-        let mut monitors = Vec::with_capacity(monitor_count);
-        for monitor_index in 0..monitor_count {
-            let start =
-                NODE_RECORD_FIXED_BYTES + monitor_index * Wallet::SHORT_ADDRESS_BYTES_LENGTH;
-            let end = start + Wallet::SHORT_ADDRESS_BYTES_LENGTH;
-            if let Some(monitor) = Wallet::bytes_to_short_address(&chunk[start..end]) {
-                monitors.push(monitor);
-            }
+        if deleted_timestamp != 0 || deleted_block != 0 {
+            return Err(
+                "network membership response contained unsigned deletion state".to_string(),
+            );
         }
 
         if only_address
@@ -236,23 +248,74 @@ async fn get_network_mapping_inner(
                 modified_timestamp: added_timestamp,
                 modified_signature: added_signature,
             },
-            monitors,
             blocks_mined,
         )
         .await
         {
             warn!("[network_map] skipped imported node record {address}: {err}");
         }
-
-        if deleted_timestamp > 0 {
-            NodeInfo::set_deleted_metadata_from_mapping(&address, deleted_timestamp, deleted_block)
-                .await;
-        }
     }
 
     if !buffer.is_empty() {
         return Err("network mapping response had trailing partial bytes".to_string());
     }
 
+    get_signed_monitor_state(
+        unlocked_stream,
+        command_map,
+        db,
+        &wallet.saved.short_address,
+        connections_key,
+        only_address,
+    )
+    .await?;
+
+    Ok(())
+}
+
+async fn get_signed_monitor_state(
+    unlocked_stream: Arc<Mutex<TcpStream>>,
+    command_map: Arc<Mutex<Command>>,
+    db: &Db,
+    local_short: &str,
+    connections_key: &str,
+    only_address: Option<&str>,
+) -> Result<(), String> {
+    let (uid, _tx, rx) = reserve_entry(command_map).await;
+    let mut message = Vec::with_capacity(4);
+    message.push(RPC_NETWORK_MONITOR_STATE);
+    message.extend_from_slice(&uid);
+    RpcResponse::send_raw(&unlocked_stream, Some(connections_key), &message).await;
+
+    let mut rx = rx.lock().await;
+    let buffer = timeout(Duration::from_secs(30), rx.recv())
+        .await
+        .map_err(|_| "timed out waiting for signed monitor state".to_string())?
+        .ok_or_else(|| "signed monitor state response channel closed".to_string())?;
+    if buffer.len() % MONITOR_EVENT_BYTES != 0 {
+        return Err("signed monitor state response had trailing partial bytes".to_string());
+    }
+
+    let mut changed = false;
+    for chunk in buffer.chunks_exact(MONITOR_EVENT_BYTES) {
+        let Some(edit) = NodeInfo::monitor_event_from_bytes(chunk) else {
+            warn!("[network_map] skipped malformed signed monitor event");
+            continue;
+        };
+        if only_address
+            .map(|address| edit.monitored_address != address)
+            .unwrap_or(false)
+        {
+            continue;
+        }
+        match NodeInfo::import_signed_monitor_event(edit, db, local_short).await {
+            Ok(imported) => changed |= imported,
+            Err(err) => warn!("[network_map] skipped signed monitor event: {err}"),
+        }
+    }
+
+    if changed {
+        NodeInfo::persist_recovery_snapshot("signed monitor state import").await;
+    }
     Ok(())
 }

src/startup/node_runtime.rs

@@ -10,6 +10,7 @@ use crate::miner::genesis::create_genesis_transaction;
 use crate::miner::mining::mine_block;
 use crate::panic;
 use crate::records::memory::chain_state::rebuild_chain_state_cache;
+use crate::records::memory::connections::initialize_node_runtime_context;
 use crate::records::memory::mempool::{init_db, setup_mempool};
 use crate::records::memory::network_mapping::NodeInfo;
 use crate::records::memory::response_channels::Command;
@@ -143,6 +144,9 @@ pub async fn run_unlocked_node(wallet: Arc<Wallet>, install_shutdown: bool) -> R
 
     let wallet_for_server = wallet.clone();
     let map: Arc<Mutex<Command>> = Arc::new(Mutex::new(HashMap::new()));
+    // Install shared node dependencies before either incoming or outgoing
+    // handshakes can promote a connection and broadcast monitor state.
+    initialize_node_runtime_context(db.clone(), wallet.clone(), map.clone())?;
     let map_cloned = Arc::clone(&map);
     let db_server = db.clone();
     let verification_service = Arc::new(initialize_global_verification_service());

src/torrent/create_metadata.rs

@@ -160,6 +160,7 @@ pub async fn broadcast_new_torrent_to_peers(
 
     // Send the torrent to the same live node-peer set used by transaction relay.
     let peers = get_live_node_broadcast_peers().await;
+    let mut messages = Vec::with_capacity(peers.len());
     for (connections_key, stream) in peers {
         // Each peer gets a short-lived reply mapping so the ack can be drained.
         let uid_bytes = reserve_transient_entry_with_context(
@@ -176,6 +177,7 @@ pub async fn broadcast_new_torrent_to_peers(
         message.extend_from_slice(&block_height.to_le_bytes()); // Block height
         message.extend_from_slice(torrent_bytes); // Torrent file contents
 
-        RpcResponse::send_raw(&stream, Some(&connections_key), &message).await;
+        messages.push((connections_key, stream, message));
     }
+    RpcResponse::broadcast_raw(messages).await;
 }