Add stop node command to solana-gossip (#3928)

Cargo.lock

@@ -2483,6 +2483,7 @@ dependencies = [
  "clap 2.33.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "env_logger 0.6.1 (registry+https://github.com/rust-lang/crates.io-index)",
  "solana 0.14.0",
+ "solana-client 0.14.0",
  "solana-netutil 0.14.0",
  "solana-sdk 0.14.0",
 ]

book/src/testnet-participation.md

@@ -41,7 +41,7 @@ Inspect the blockexplorer at [http://beta.testnet.solana.com/](http://beta.testn
 
 Run the following command to join the gossip network and view all the other nodes in the cluster:
 ```bash
-$ solana-gossip --network beta.testnet.solana.com:8001
+$ solana-gossip --network beta.testnet.solana.com:8001 spy
 ```
 
 View the [metrics dashboard](

ci/localnet-sanity.sh

@@ -306,8 +306,7 @@ while [[ $iteration -le $iterations ]]; do
     set -x
     client_id=/tmp/client-id.json-$$
     $solana_keygen -o $client_id || exit $?
-    $solana_gossip \
+    $solana_gossip spy --num-nodes-exactly $numNodes || exit $?
-      --num-nodes-exactly $numNodes || exit $?
     rm -rf $client_id
   ) || flag_error
 

gossip/Cargo.toml

@@ -12,6 +12,7 @@ homepage = "https://solana.com/"
 clap = "2.33.0"
 env_logger = "0.6.1"
 solana = { path = "../core", version = "0.14.0" }
+solana-client = { path = "../client", version = "0.14.0"  }
 solana-netutil = { path = "../netutil", version = "0.14.0" }
 solana-sdk = { path = "../sdk", version = "0.14.0" }
 

gossip/src/main.rs

@@ -1,7 +1,9 @@
 //! A command-line executable for monitoring a cluster's gossip plane.
 
-use clap::{crate_description, crate_name, crate_version, App, Arg};
+use clap::{crate_description, crate_name, crate_version, App, AppSettings, Arg, SubCommand};
+use solana::contact_info::ContactInfo;
 use solana::gossip_service::discover;
+use solana_client::rpc_client::RpcClient;
 use solana_sdk::pubkey::Pubkey;
 use std::error;
 use std::net::SocketAddr;
@@ -22,6 +24,7 @@ fn main() -> Result<(), Box<dyn error::Error>> {
     let matches = App::new(crate_name!())
         .about(crate_description!())
         .version(crate_version!())
+        .setting(AppSettings::SubcommandRequiredElseHelp)
         .arg(
             Arg::with_name("network")
                 .short("n")
@@ -31,38 +34,58 @@ fn main() -> Result<(), Box<dyn error::Error>> {
                 .default_value(&network_string)
                 .help("Rendezvous with the cluster at this gossip entry point"),
         )
-        .arg(
+        .subcommand(
-            Arg::with_name("num_nodes")
+            SubCommand::with_name("spy")
-                .short("N")
+                .about("Monitor the gossip network")
-                .long("num-nodes")
+                .setting(AppSettings::DisableVersion)
-                .value_name("NUM")
+                .arg(
-                .takes_value(true)
+                    Arg::with_name("num_nodes")
-                .conflicts_with("num_nodes_exactly")
+                        .short("N")
-                .help("Wait for at least NUM nodes to converge"),
+                        .long("num-nodes")
+                        .value_name("NUM")
+                        .takes_value(true)
+                        .conflicts_with("num_nodes_exactly")
+                        .help("Wait for at least NUM nodes to converge"),
+                )
+                .arg(
+                    Arg::with_name("num_nodes_exactly")
+                        .short("E")
+                        .long("num-nodes-exactly")
+                        .value_name("NUM")
+                        .takes_value(true)
+                        .help("Wait for exactly NUM nodes to converge"),
+                )
+                .arg(
+                    Arg::with_name("node_pubkey")
+                        .short("p")
+                        .long("pubkey")
+                        .value_name("PUBKEY")
+                        .takes_value(true)
+                        .validator(pubkey_validator)
+                        .help("Public key of a specific node to wait for"),
+                )
+                .arg(
+                    Arg::with_name("timeout")
+                        .long("timeout")
+                        .value_name("SECS")
+                        .takes_value(true)
+                        .help(
+                            "Maximum time to wait for cluster to converge [default: wait forever]",
+                        ),
+                ),
         )
-        .arg(
+        .subcommand(
-            Arg::with_name("num_nodes_exactly")
+            SubCommand::with_name("stop")
-                .short("E")
+                .about("Send stop request to a node")
-                .long("num-nodes-exactly")
+                .setting(AppSettings::DisableVersion)
-                .value_name("NUM")
+                .arg(
-                .takes_value(true)
+                    Arg::with_name("node_pubkey")
-                .help("Wait for exactly NUM nodes to converge"),
+                        .index(1)
-        )
+                        .required(true)
-        .arg(
+                        .value_name("PUBKEY")
-            Arg::with_name("node_pubkey")
+                        .validator(pubkey_validator)
-                .short("p")
+                        .help("Public key of a specific node to stop"),
-                .long("pubkey")
+                ),
-                .value_name("PUBKEY")
-                .takes_value(true)
-                .validator(pubkey_validator)
-                .help("Public key of a specific node to wait for"),
-        )
-        .arg(
-            Arg::with_name("timeout")
-                .long("timeout")
-                .value_name("SECS")
-                .takes_value(true)
-                .help("Maximum time to wait for cluster to converge [default: wait forever]"),
         )
         .get_matches();
 
@@ -72,48 +95,74 @@ fn main() -> Result<(), Box<dyn error::Error>> {
             exit(1)
         });
     }
+    match matches.subcommand() {
+        ("spy", Some(matches)) => {
+            let num_nodes_exactly = matches
+                .value_of("num_nodes_exactly")
+                .map(|num| num.to_string().parse().unwrap());
+            let num_nodes = matches
+                .value_of("num_nodes")
+                .map(|num| num.to_string().parse().unwrap())
+                .or(num_nodes_exactly);
+            let timeout = matches
+                .value_of("timeout")
+                .map(|secs| secs.to_string().parse().unwrap());
+            let pubkey = matches
+                .value_of("node_pubkey")
+                .map(|pubkey_str| pubkey_str.parse::<Pubkey>().unwrap());
 
-    let num_nodes_exactly = matches
+            let nodes = discover(&network_addr, num_nodes, timeout, pubkey)?;
-        .value_of("num_nodes_exactly")
-        .map(|num| num.to_string().parse().unwrap());
-    let num_nodes = matches
-        .value_of("num_nodes")
-        .map(|num| num.to_string().parse().unwrap())
-        .or(num_nodes_exactly);
-    let timeout = matches
-        .value_of("timeout")
-        .map(|secs| secs.to_string().parse().unwrap());
-    let pubkey = matches
-        .value_of("node_pubkey")
-        .map(|pubkey_str| pubkey_str.parse::<Pubkey>().unwrap());
 
-    let nodes = discover(&network_addr, num_nodes, timeout, pubkey)?;
+            if timeout.is_some() {
-
+                if let Some(num) = num_nodes {
-    if timeout.is_some() {
+                    if nodes.len() < num {
-        if let Some(num) = num_nodes {
+                        let add = if num_nodes_exactly.is_some() {
-            if nodes.len() < num {
+                            ""
-                let add = if num_nodes_exactly.is_some() {
+                        } else {
-                    ""
+                            " or more"
-                } else {
+                        };
-                    " or more"
+                        eprintln!(
-                };
+                            "Error: Insufficient nodes discovered.  Expecting {}{}",
+                            num, add,
+                        );
+                    }
+                }
+                if let Some(node) = pubkey {
+                    if nodes.iter().find(|x| x.id == node).is_none() {
+                        eprintln!("Error: Could not find node {:?}", node);
+                    }
+                }
+            }
+            if num_nodes_exactly.is_some() && nodes.len() > num_nodes_exactly.unwrap() {
                 eprintln!(
-                    "Error: Insufficient nodes discovered.  Expecting {}{}",
+                    "Error: Extra nodes discovered.  Expecting exactly {}",
-                    num, add,
+                    num_nodes_exactly.unwrap()
                 );
             }
         }
-        if let Some(node) = pubkey {
+        ("stop", Some(matches)) => {
-            if nodes.iter().find(|x| x.id == node).is_none() {
+            let pubkey = matches
-                eprintln!("Error: Could not find node {:?}", node);
+                .value_of("node_pubkey")
+                .unwrap()
+                .parse::<Pubkey>()
+                .unwrap();
+            let nodes = discover(&network_addr, None, None, Some(pubkey))?;
+            let node = nodes.iter().find(|x| x.id == pubkey).unwrap();
+
+            if !ContactInfo::is_valid_address(&node.rpc) {
+                eprintln!("Error: RPC service is not enabled on node {:?}", pubkey);
+            }
+            println!("\nSending stop request to node {:?}", pubkey);
+
+            let result = RpcClient::new_socket(node.rpc).fullnode_exit()?;
+            if result {
+                println!("Stop signal accepted");
+            } else {
+                eprintln!("Error: Stop signal ignored");
             }
         }
+        _ => unreachable!(),
     }
-    if num_nodes_exactly.is_some() && nodes.len() > num_nodes_exactly.unwrap() {
+
-        eprintln!(
-            "Error: Extra nodes discovered.  Expecting exactly {}",
-            num_nodes_exactly.unwrap()
-        );
-    }
     Ok(())
 }

net/remote/remote-node.sh

@@ -81,12 +81,13 @@ local|tar)
     fi
     ./multinode-demo/drone.sh > drone.log 2>&1 &
 
-    maybePublicAddress=
+    args=()
     if $publicNetwork; then
-      maybePublicAddress="--public-address"
+      args+=(--public-address)
     fi
+    args+=(--enable-rpc-exit)
 
-    ./multinode-demo/bootstrap-leader.sh $maybePublicAddress > bootstrap-leader.log 2>&1 &
+    ./multinode-demo/bootstrap-leader.sh "${args[@]}" > bootstrap-leader.log 2>&1 &
     ln -sTf bootstrap-leader.log fullnode.log
     ;;
   fullnode|blockstreamer)
@@ -116,6 +117,7 @@ local|tar)
     fi
 
     args+=(
+      --enable-rpc-exit
       --gossip-port 8001
       --rpc-port 8899
     )
@@ -147,7 +149,9 @@ local|tar)
       # Confirm the blockexplorer is now globally accessible
       curl --head "$(curl ifconfig.io)"
     fi
-    ./multinode-demo/fullnode.sh "${args[@]}" "$entrypointIp":~/solana "$entrypointIp:8001" > fullnode.log 2>&1 &
+
+    args+=("$entrypointIp":~/solana "$entrypointIp:8001")
+    ./multinode-demo/fullnode.sh "${args[@]}" > fullnode.log 2>&1 &
     ;;
   *)
     echo "Error: unknown node type: $nodeType"

net/remote/remote-sanity.sh

@@ -89,9 +89,8 @@ echo "+++ $entrypointIp: node count ($numNodes expected)"
     nodeArg="num-nodes-exactly"
   fi
 
-  timeout 2m $solana_gossip \
+  timeout 2m $solana_gossip --network "$entrypointIp:8001" \
-    --network "$entrypointIp:8001" \
+    spy --$nodeArg "$numNodes" \
-    --$nodeArg "$numNodes" \
 )
 
 echo "--- RPC API: getTransactionCount"