Files
neon/safekeeper/tests/misc_test.rs
Arthur Petukhovsky 4be2223a4c Discrete event simulation for safekeepers (#5804)
This PR contains the first version of a
[FoundationDB-like](https://www.youtube.com/watch?v=4fFDFbi3toc)
simulation testing for safekeeper and walproposer.

### desim

This is a core "framework" for running determenistic simulation. It
operates on threads, allowing to test syncronous code (like walproposer).

`libs/desim/src/executor.rs` contains implementation of a determenistic
thread execution. This is achieved by blocking all threads, and each
time allowing only a single thread to make an execution step. All
executor's threads are blocked using `yield_me(after_ms)` function. This
function is called when a thread wants to sleep or wait for an external
notification (like blocking on a channel until it has a ready message).

`libs/desim/src/chan.rs` contains implementation of a channel (basic
sync primitive). It has unlimited capacity and any thread can push or
read messages to/from it.

`libs/desim/src/network.rs` has a very naive implementation of a network
(only reliable TCP-like connections are supported for now), that can
have arbitrary delays for each package and failure injections for
breaking connections with some probability.

`libs/desim/src/world.rs` ties everything together, to have a concept of
virtual nodes that can have network connections between them.

### walproposer_sim

Has everything to run walproposer and safekeepers in a simulation.

`safekeeper.rs` reimplements all necesary stuff from `receive_wal.rs`,
`send_wal.rs` and `timelines_global_map.rs`.

`walproposer_api.rs` implements all walproposer callback to use
simulation library.

`simulation.rs` defines a schedule – a set of events like `restart <sk>`
or `write_wal` that should happen at time `<ts>`. It also has code to
spawn walproposer/safekeeper threads and provide config to them.

### tests

`simple_test.rs` has tests that just start walproposer and 3 safekeepers
together in a simulation, and tests that they are not crashing right
away.

`misc_test.rs` has tests checking more advanced simulation cases, like
crashing or restarting threads, testing memory deallocation, etc.

`random_test.rs` is the main test, it checks thousands of random seeds
(schedules) for correctness. It roughly corresponds to running a real
python integration test in an environment with very unstable network and
cpu, but in a determenistic way (each seed results in the same execution
log) and much much faster.

Closes #547

---------

Co-authored-by: Arseny Sher <sher-ars@yandex.ru>
2024-02-12 20:29:57 +00:00

156 lines
4.6 KiB
Rust

use std::sync::Arc;
use tracing::{info, warn};
use utils::lsn::Lsn;
use crate::walproposer_sim::{
log::{init_logger, init_tracing_logger},
simulation::{generate_network_opts, generate_schedule, Schedule, TestAction, TestConfig},
};
pub mod walproposer_sim;
// Test that simulation supports restarting (crashing) safekeepers.
#[test]
fn crash_safekeeper() {
let clock = init_logger();
let config = TestConfig::new(Some(clock));
let test = config.start(1337);
let lsn = test.sync_safekeepers().unwrap();
assert_eq!(lsn, Lsn(0));
info!("Sucessfully synced empty safekeepers at 0/0");
let mut wp = test.launch_walproposer(lsn);
// Write some WAL and crash safekeeper 0 without waiting for replication.
test.poll_for_duration(30);
wp.write_tx(3);
test.servers[0].restart();
// Wait some time, so that walproposer can reconnect.
test.poll_for_duration(2000);
}
// Test that walproposer can be crashed (stopped).
#[test]
fn test_simple_restart() {
let clock = init_logger();
let config = TestConfig::new(Some(clock));
let test = config.start(1337);
let lsn = test.sync_safekeepers().unwrap();
assert_eq!(lsn, Lsn(0));
info!("Sucessfully synced empty safekeepers at 0/0");
let mut wp = test.launch_walproposer(lsn);
test.poll_for_duration(30);
wp.write_tx(3);
test.poll_for_duration(100);
wp.stop();
drop(wp);
let lsn = test.sync_safekeepers().unwrap();
info!("Sucessfully synced safekeepers at {}", lsn);
}
// Test runnning a simple schedule, restarting everything a several times.
#[test]
fn test_simple_schedule() -> anyhow::Result<()> {
let clock = init_logger();
let mut config = TestConfig::new(Some(clock));
config.network.keepalive_timeout = Some(100);
let test = config.start(1337);
let schedule: Schedule = vec![
(0, TestAction::RestartWalProposer),
(50, TestAction::WriteTx(5)),
(100, TestAction::RestartSafekeeper(0)),
(100, TestAction::WriteTx(5)),
(110, TestAction::RestartSafekeeper(1)),
(110, TestAction::WriteTx(5)),
(120, TestAction::RestartSafekeeper(2)),
(120, TestAction::WriteTx(5)),
(201, TestAction::RestartWalProposer),
(251, TestAction::RestartSafekeeper(0)),
(251, TestAction::RestartSafekeeper(1)),
(251, TestAction::RestartSafekeeper(2)),
(251, TestAction::WriteTx(5)),
(255, TestAction::WriteTx(5)),
(1000, TestAction::WriteTx(5)),
];
test.run_schedule(&schedule)?;
info!("Test finished, stopping all threads");
test.world.deallocate();
Ok(())
}
// Test that simulation can process 10^4 transactions.
#[test]
fn test_many_tx() -> anyhow::Result<()> {
let clock = init_logger();
let config = TestConfig::new(Some(clock));
let test = config.start(1337);
let mut schedule: Schedule = vec![];
for i in 0..100 {
schedule.push((i * 10, TestAction::WriteTx(100)));
}
test.run_schedule(&schedule)?;
info!("Test finished, stopping all threads");
test.world.stop_all();
let events = test.world.take_events();
info!("Events: {:?}", events);
let last_commit_lsn = events
.iter()
.filter_map(|event| {
if event.data.starts_with("commit_lsn;") {
let lsn: u64 = event.data.split(';').nth(1).unwrap().parse().unwrap();
return Some(lsn);
}
None
})
.last()
.unwrap();
let initdb_lsn = 21623024;
let diff = last_commit_lsn - initdb_lsn;
info!("Last commit lsn: {}, diff: {}", last_commit_lsn, diff);
// each tx is at least 8 bytes, it's written a 100 times for in a loop for 100 times
assert!(diff > 100 * 100 * 8);
Ok(())
}
// Checks that we don't have nasty circular dependencies, preventing Arc from deallocating.
// This test doesn't really assert anything, you need to run it manually to check if there
// is any issue.
#[test]
fn test_res_dealloc() -> anyhow::Result<()> {
let clock = init_tracing_logger(true);
let mut config = TestConfig::new(Some(clock));
let seed = 123456;
config.network = generate_network_opts(seed);
let test = config.start(seed);
warn!("Running test with seed {}", seed);
let schedule = generate_schedule(seed);
info!("schedule: {:?}", schedule);
test.run_schedule(&schedule).unwrap();
test.world.stop_all();
let world = test.world.clone();
drop(test);
info!("world strong count: {}", Arc::strong_count(&world));
world.deallocate();
info!("world strong count: {}", Arc::strong_count(&world));
Ok(())
}