1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494
//! Contains helpers for reaping Hosts that haven't received a heartbeat within a configured amount
//! of time and actors and providers on hosts that no longer exist
use std::collections::HashMap;
use chrono::{Duration, Utc};
use tokio::{task::JoinHandle, time};
use tracing::{debug, error, info, instrument, trace, warn};
use super::{Actor, Host, Provider, Store};
/// A struct that can reap various pieces of data from the given store
pub struct Reaper<S> {
store: S,
interval: Duration,
handles: HashMap<String, JoinHandle<()>>,
}
impl<S: Store + Clone + Send + Sync + 'static> Reaper<S> {
/// Creates a new reaper using the given store configured to check for reaping every
/// `check_interval` for all passed lattice IDs. This reaper will immediately begin executing
/// spawned tasks. When the reaper is dropped, it will stop polling all tasks. This function
/// will panic if you pass it a duration that is larger than the maximum value accepted by the
/// `chrono` library. As this is a rare case, we don't actually return an error and panic
/// instead
///
/// The reaper will wait for 2 * `check_interval` before removing anything. For example, if
/// `check_interval` is set to 30s, then after 30s, the item is considered to be in a "warning"
/// state. This isn't actually reflected in state right now, but it will be logged. When the
/// next tick fires (around 60s total), then the item will be removed from the store
pub fn new(
store: S,
check_interval: std::time::Duration,
lattices_to_observe: impl IntoIterator<Item = String>,
) -> Reaper<S> {
let interval = Duration::from_std(check_interval)
.expect("The given duration is out of bounds for a max duration value");
let cloned_store = store.clone();
let handles = lattices_to_observe.into_iter().map(move |id| {
(
id.clone(),
tokio::spawn(
Undertaker {
store: cloned_store.clone(),
lattice_id: id,
interval,
}
.reap(),
),
)
});
Reaper {
store,
interval,
handles: handles.collect(),
}
}
/// Adds a new lattice to be reaped
pub fn observe(&mut self, lattice_id: &str) {
// If the handle exists and is still running, just leave it
if let Some(handle) = self.handles.get(lattice_id) {
if !handle.is_finished() {
return;
}
}
self.handles.insert(
lattice_id.to_owned(),
tokio::spawn(
Undertaker {
store: self.store.clone(),
lattice_id: lattice_id.to_owned(),
interval: self.interval,
}
.reap(),
),
);
}
/// Stops observing the given lattice
pub fn remove(&mut self, lattice_id: &str) {
if let Some(handle) = self.handles.remove(lattice_id) {
handle.abort();
}
}
}
struct Undertaker<S> {
store: S,
lattice_id: String,
interval: Duration,
}
impl<S: Store + Clone + Send + Sync + 'static> Undertaker<S> {
#[instrument(level = "debug", skip(self), fields(lattice_id = %self.lattice_id, check_interval = %self.interval))]
async fn reap(self) {
debug!("Starting reaper");
// SAFETY: We created this Duration from a std Duration, so it should unwrap back just fine
let mut ticker = time::interval(self.interval.to_std().unwrap());
ticker.set_missed_tick_behavior(time::MissedTickBehavior::Delay);
loop {
ticker.tick().await;
trace!("Tick fired, running reap tasks");
// We want to reap hosts first so that the state is up to date for reaping actors and providers
self.reap_hosts().await;
// Now get the current list of hosts
let hosts = match self.store.list::<Host>(&self.lattice_id).await {
Ok(n) => n,
Err(e) => {
error!(error = %e, "Error when fetching hosts from store. Will retry on next tick");
continue;
}
};
// Reap actors and providers simultaneously
futures::join!(self.reap_actors(&hosts), self.reap_providers(&hosts));
trace!("Completed reap tasks");
}
}
#[instrument(level = "debug", skip(self), fields(lattice_id = %self.lattice_id))]
async fn reap_hosts(&self) {
let hosts = match self.store.list::<Host>(&self.lattice_id).await {
Ok(n) => n,
Err(e) => {
error!(error = %e, "Error when fetching hosts from store. Will retry on next tick");
return;
}
};
let hosts_to_remove = hosts.into_iter().filter_map(|(id, host)| {
let elapsed = Utc::now() - host.last_seen;
if elapsed > (self.interval * 2) {
info!(%id, friendly_name = %host.friendly_name, "Host has not been seen for 2 intervals. Will reap node");
Some(id)
} else if elapsed > self.interval {
info!(%id, friendly_name = %host.friendly_name, "Host has not been seen for 1 interval. Next check will reap node from store");
None
} else {
None
}
});
if let Err(e) = self
.store
.delete_many::<Host, _, _>(&self.lattice_id, hosts_to_remove)
.await
{
error!(error = %e, "Error when deleting hosts from store. Will retry on next tick")
}
}
#[instrument(level = "debug", skip(self, hosts), fields(lattice_id = %self.lattice_id))]
async fn reap_actors(&self, hosts: &HashMap<String, Host>) {
let actors = match self.store.list::<Actor>(&self.lattice_id).await {
Ok(n) => n,
Err(e) => {
error!(error = %e, "Error when fetching actors from store. Will retry on next tick");
return;
}
};
let (actors_to_remove, actors_to_update): (HashMap<String, Actor>, HashMap<String, Actor>) =
actors
.into_iter()
.filter_map(|(id, mut actor)| {
let current_num_hosts = actor.instances.len();
// Only keep the instances where the host exists and the actor is in its map
actor
.instances
.retain(|host_id, _| hosts.get(host_id).map(|host| host.actors.contains_key(&actor.id)).unwrap_or(false));
// Now for the remaining instances, make sure the number of instances is equal
// to what is observed on the host, otherwise truncate.
// NOTE: If for some reason we start using instance IDs, than things will need
// to be updated so we can clear the instance ID
let mut did_truncate = false;
for (host_id, instances) in actor.instances.iter_mut() {
if let Some(host) = hosts.get(host_id) {
// This unwrap shouldn't happen because we just retained the instances
// that have the actor ID in their list. If it does, we unwrap to
// current length so it just skips this logic for now
let current_num_instances = *host.actors.get(&actor.id).unwrap_or(&instances.len());
if instances.len() > current_num_instances {
debug!(%id, %host_id, num_instances = %instances.len(), %current_num_instances, "Number of instances for actor is greater than number of instances observed on host. Truncating to correct number");
*instances = instances.drain().take(current_num_instances).collect();
did_truncate = true;
}
// If we have less instances than the host, then it just means the host
// heartbeat will update them down the line
}
}
// If we got rid of something or truncated instances, that means this needs to update
((current_num_hosts != actor.instances.len()) || did_truncate).then_some((id, actor))
})
.partition(|(_, actor)| actor.instances.is_empty());
debug!(to_remove = %actors_to_remove.len(), to_update = %actors_to_update.len(), "Filtered out list of actors to update and reap");
if let Err(e) = self
.store
.store_many(&self.lattice_id, actors_to_update)
.await
{
warn!(error = %e, "Error when storing updated actors. Will retry on next tick");
return;
}
if let Err(e) = self
.store
.delete_many::<Actor, _, _>(&self.lattice_id, actors_to_remove.keys())
.await
{
warn!(error = %e, "Error when deleting actors from store. Will retry on next tick")
}
}
#[instrument(level = "debug", skip(self, hosts), fields(lattice_id = %self.lattice_id))]
async fn reap_providers(&self, hosts: &HashMap<String, Host>) {
let providers = match self.store.list::<Provider>(&self.lattice_id).await {
Ok(n) => n,
Err(e) => {
error!(error = %e, "Error when fetching actors from store. Will retry on next tick");
return;
}
};
let (providers_to_remove, providers_to_update): (
HashMap<String, Provider>,
HashMap<String, Provider>,
) = providers
.into_iter()
.filter_map(|(id, mut provider)| {
let current_num_hosts = provider.hosts.len();
// Only keep the instances where the host exists
provider
.hosts
.retain(|host_id, _| hosts.contains_key(host_id));
// If we got rid of something, that means this needs to update
(current_num_hosts != provider.hosts.len()).then_some((id, provider))
})
.partition(|(_, provider)| provider.hosts.is_empty());
debug!(to_remove = %providers_to_remove.len(), to_update = %providers_to_update.len(), "Filtered out list of providers to update and reap");
if let Err(e) = self
.store
.store_many(&self.lattice_id, providers_to_update)
.await
{
warn!(error = %e, "Error when storing updated providers. Will retry on next tick");
return;
}
if let Err(e) = self
.store
.delete_many::<Provider, _, _>(&self.lattice_id, providers_to_remove.keys())
.await
{
warn!(error = %e, "Error when deleting providers from store. Will retry on next tick")
}
}
}
#[cfg(test)]
mod test {
use super::*;
use std::{
collections::{BTreeMap, HashSet},
sync::Arc,
};
use crate::{
storage::{ProviderStatus, ReadStore, WadmActorInfo},
test_util::TestStore,
};
#[tokio::test]
async fn test_reaping() {
let store = Arc::new(TestStore::default());
let lattice_id = "reaper";
let actor_id = "testactor";
let host1_id = "host1";
let host2_id = "host2";
// Prepopulate the store
store
.store_many(
lattice_id,
[
(
actor_id.to_string(),
Actor {
id: actor_id.to_string(),
instances: HashMap::from([
(
host1_id.to_string(),
HashSet::from_iter([WadmActorInfo {
annotations: BTreeMap::default(),
count: 1,
}]),
),
(
host2_id.to_string(),
HashSet::from_iter([WadmActorInfo {
annotations: BTreeMap::default(),
count: 1,
}]),
),
]),
..Default::default()
},
),
(
"idontexist".to_string(),
Actor {
id: "idontexist".to_string(),
instances: HashMap::from([(
host1_id.to_string(),
HashSet::from_iter([WadmActorInfo {
annotations: BTreeMap::default(),
count: 1,
}]),
)]),
..Default::default()
},
),
],
)
.await
.unwrap();
store
.store(
lattice_id,
"fakeprovider".to_string(),
Provider {
id: "fakeprovider".to_string(),
hosts: HashMap::from([(host1_id.to_string(), ProviderStatus::Running)]),
..Default::default()
},
)
.await
.unwrap();
store
.store_many(
lattice_id,
[
(
host1_id.to_string(),
Host {
actors: HashMap::from([(actor_id.to_string(), 1)]),
providers: HashSet::default(),
id: host1_id.to_string(),
last_seen: Utc::now(),
..Default::default()
},
),
(
host2_id.to_string(),
Host {
actors: HashMap::from([(actor_id.to_string(), 1)]),
providers: HashSet::default(),
id: host2_id.to_string(),
// Make this host stick around for longer
last_seen: Utc::now() + Duration::milliseconds(600),
..Default::default()
},
),
],
)
.await
.unwrap();
let reap_interval = std::time::Duration::from_millis(500);
// Interval + wiggle
let wait = reap_interval + std::time::Duration::from_millis(100);
let _reaper = Reaper::new(store.clone(), reap_interval, [lattice_id.to_owned()]);
// Wait for first node to be reaped (two ticks)
tokio::time::sleep(wait * 2).await;
// Now check that the providers, actors, and hosts were reaped
let hosts = store.list::<Host>(lattice_id).await.unwrap();
assert_eq!(hosts.len(), 1, "Only one host should be left");
let actors = store.list::<Actor>(lattice_id).await.unwrap();
assert_eq!(actors.len(), 1, "Only one actor should remain in the store");
actors
.get(actor_id)
.expect("Should have the correct actor in the store");
assert!(
store.list::<Provider>(lattice_id).await.unwrap().is_empty(),
"No providers should exist"
);
}
#[tokio::test]
async fn test_stale_actor() {
let store = Arc::new(TestStore::default());
let lattice_id = "reaper";
let actor_id = "testactor";
let host1_id = "host1";
let host2_id = "host2";
// Prepopulate the store
store
.store(
lattice_id,
actor_id.to_string(),
Actor {
id: actor_id.to_string(),
instances: HashMap::from([
(
host1_id.to_string(),
HashSet::from_iter([WadmActorInfo {
annotations: BTreeMap::default(),
count: 1,
}]),
),
(
host2_id.to_string(),
HashSet::from_iter([WadmActorInfo {
annotations: BTreeMap::default(),
count: 1,
}]),
),
]),
..Default::default()
},
)
.await
.unwrap();
store
.store_many(
lattice_id,
[
(
host1_id.to_string(),
Host {
actors: HashMap::from([(actor_id.to_string(), 1)]),
providers: HashSet::default(),
id: host1_id.to_string(),
last_seen: Utc::now() + Duration::milliseconds(600),
..Default::default()
},
),
(
host2_id.to_string(),
Host {
actors: HashMap::default(),
providers: HashSet::default(),
id: host2_id.to_string(),
last_seen: Utc::now() + Duration::milliseconds(600),
..Default::default()
},
),
],
)
.await
.unwrap();
let reap_interval = std::time::Duration::from_millis(50);
// Interval + wiggle
let wait = std::time::Duration::from_millis(70);
let _reaper = Reaper::new(store.clone(), reap_interval, [lattice_id.to_owned()]);
// Wait for first tick
tokio::time::sleep(wait).await;
// Make sure we only have one instance of the actor left
let actors = store.list::<Actor>(lattice_id).await.unwrap();
let actor = actors
.get(actor_id)
.expect("Should have the correct actor in the store");
assert_eq!(
actor.instances.len(),
1,
"Only one host should remain in instances"
);
assert_eq!(
actor
.instances
.get(host1_id)
.expect("Should have instance left on the correct host")
.len(),
1,
"Only one instance should remain on host"
);
}
}