solana/src/retransmit_stage.rs

132 lines
4.0 KiB
Rust
Raw Normal View History

//! The `retransmit_stage` retransmits blobs between validators
2018-10-08 19:55:54 -07:00
use cluster_info::ClusterInfo;
use counter::Counter;
use entry::Entry;
use influx_db_client as influxdb;
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
use leader_scheduler::LeaderScheduler;
use log::Level;
use metrics;
use result::{Error, Result};
use service::Service;
use std::net::UdpSocket;
use std::sync::atomic::{AtomicBool, AtomicUsize};
use std::sync::mpsc::RecvTimeoutError;
use std::sync::mpsc::{channel, Receiver};
use std::sync::{Arc, RwLock};
use std::thread::{self, Builder, JoinHandle};
use std::time::Duration;
2018-08-09 12:31:34 -07:00
use streamer::BlobReceiver;
use sys_info::hostname;
2018-09-07 15:00:26 -07:00
use window::SharedWindow;
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
use window_service::window_service;
2018-10-08 19:55:54 -07:00
fn retransmit(
cluster_info: &Arc<RwLock<ClusterInfo>>,
r: &BlobReceiver,
sock: &UdpSocket,
) -> Result<()> {
let timer = Duration::new(1, 0);
let mut dq = r.recv_timeout(timer)?;
while let Ok(mut nq) = r.try_recv() {
dq.append(&mut nq);
}
metrics::submit(
influxdb::Point::new("retransmit-stage")
.add_field(
"host",
influxdb::Value::String(hostname().unwrap_or_else(|_| "?".to_string())),
).add_field("count", influxdb::Value::Integer(dq.len() as i64))
.to_owned(),
);
for b in &mut dq {
2018-10-08 19:55:54 -07:00
ClusterInfo::retransmit(&cluster_info, b, sock)?;
}
Ok(())
}
/// Service to retransmit messages from the leader to layer 1 nodes.
2018-10-08 19:55:54 -07:00
/// See `cluster_info` for network layer definitions.
/// # Arguments
/// * `sock` - Socket to read from. Read timeout is set to 1.
/// * `exit` - Boolean to signal system exit.
2018-10-08 19:55:54 -07:00
/// * `cluster_info` - This structure needs to be updated and populated by the bank and via gossip.
/// * `recycler` - Blob recycler.
/// * `r` - Receive channel for blobs to be retransmitted to all the layer 1 nodes.
2018-10-08 19:55:54 -07:00
fn retransmitter(
sock: Arc<UdpSocket>,
cluster_info: Arc<RwLock<ClusterInfo>>,
r: BlobReceiver,
) -> JoinHandle<()> {
Builder::new()
.name("solana-retransmitter".to_string())
.spawn(move || {
trace!("retransmitter started");
loop {
2018-10-08 19:55:54 -07:00
if let Err(e) = retransmit(&cluster_info, &r, &sock) {
match e {
Error::RecvTimeoutError(RecvTimeoutError::Disconnected) => break,
Error::RecvTimeoutError(RecvTimeoutError::Timeout) => (),
_ => {
inc_new_counter_info!("streamer-retransmit-error", 1, 1);
}
}
}
}
trace!("exiting retransmitter");
2018-09-14 16:25:14 -07:00
}).unwrap()
}
pub struct RetransmitStage {
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
thread_hdls: Vec<JoinHandle<()>>,
}
impl RetransmitStage {
pub fn new(
2018-10-08 19:55:54 -07:00
cluster_info: &Arc<RwLock<ClusterInfo>>,
window: SharedWindow,
tick_height: u64,
2018-06-27 12:35:58 -07:00
entry_height: u64,
retransmit_socket: Arc<UdpSocket>,
repair_socket: Arc<UdpSocket>,
2018-06-27 11:33:56 -07:00
fetch_stage_receiver: BlobReceiver,
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
leader_scheduler: Arc<RwLock<LeaderScheduler>>,
) -> (Self, Receiver<Vec<Entry>>) {
let (retransmit_sender, retransmit_receiver) = channel();
2018-10-08 19:55:54 -07:00
let t_retransmit =
retransmitter(retransmit_socket, cluster_info.clone(), retransmit_receiver);
let (entry_sender, entry_receiver) = channel();
let done = Arc::new(AtomicBool::new(false));
2018-09-07 15:00:26 -07:00
let t_window = window_service(
2018-10-08 19:55:54 -07:00
cluster_info.clone(),
window,
tick_height,
2018-06-27 12:35:58 -07:00
entry_height,
0,
fetch_stage_receiver,
entry_sender,
retransmit_sender,
repair_socket,
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
leader_scheduler,
done,
);
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
let thread_hdls = vec![t_retransmit, t_window];
(RetransmitStage { thread_hdls }, entry_receiver)
}
}
impl Service for RetransmitStage {
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
type JoinReturnType = ();
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
fn join(self) -> thread::Result<()> {
for thread_hdl in self.thread_hdls {
thread_hdl.join()?;
}
Leader scheduler plumbing (#1440) * Added LeaderScheduler module and tests * plumbing for LeaderScheduler in Fullnode + tests. Add vote processing for active set to ReplicateStage and WriteStage * Add LeaderScheduler plumbing for Tvu, window, and tests * Fix bank and switch tests to use new LeaderScheduler * move leader rotation check from window service to replicate stage * Add replicate_stage leader rotation exit test * removed leader scheduler from the window service and associated modules/tests * Corrected is_leader calculation in repair() function in window.rs * Integrate LeaderScheduler with write_stage for leader to validator transitions * Integrated LeaderScheduler with BroadcastStage * Removed gossip leader rotation from crdt * Add multi validator, leader test * Comments and cleanup * Remove unneeded checks from broadcast stage * Fix case where a validator/leader need to immediately transition on startup after reading ledger and seeing they are not in the correct role * Set new leader in validator -> validator transitions * Clean up for PR comments, refactor LeaderScheduler from process_entry/process_ledger_tail * Cleaned out LeaderScheduler options, implemented LeaderScheduler strategy that only picks the bootstrap leader to support existing tests, drone/airdrops * Ignore test_full_leader_validator_network test due to bug where the next leader in line fails to get the last entry before rotation (b/c it hasn't started up yet). Added a test test_dropped_handoff_recovery go track this bug
2018-10-10 16:49:41 -07:00
Ok(())
}
}