Line data Source code
1 : //! This module implements the pageserver-global disk-usage-based layer eviction task.
2 : //!
3 : //! # Mechanics
4 : //!
5 : //! Function `launch_disk_usage_global_eviction_task` starts a pageserver-global background
6 : //! loop that evicts layers in response to a shortage of available bytes
7 : //! in the $repo/tenants directory's filesystem.
8 : //!
9 : //! The loop runs periodically at a configurable `period`.
10 : //!
11 : //! Each loop iteration uses `statvfs` to determine filesystem-level space usage.
12 : //! It compares the returned usage data against two different types of thresholds.
13 : //! The iteration tries to evict layers until app-internal accounting says we should be below the thresholds.
14 : //! We cross-check this internal accounting with the real world by making another `statvfs` at the end of the iteration.
15 : //! We're good if that second statvfs shows that we're _actually_ below the configured thresholds.
16 : //! If we're still above one or more thresholds, we emit a warning log message, leaving it to the operator to investigate further.
17 : //!
18 : //! # Eviction Policy
19 : //!
20 : //! There are two thresholds:
21 : //! `max_usage_pct` is the relative available space, expressed in percent of the total filesystem space.
22 : //! If the actual usage is higher, the threshold is exceeded.
23 : //! `min_avail_bytes` is the absolute available space in bytes.
24 : //! If the actual usage is lower, the threshold is exceeded.
25 : //! If either of these thresholds is exceeded, the system is considered to have "disk pressure", and eviction
26 : //! is performed on the next iteration, to release disk space and bring the usage below the thresholds again.
27 : //! The iteration evicts layers in LRU fashion, but, with a weak reservation per tenant.
28 : //! The reservation is to keep the most recently accessed X bytes per tenant resident.
29 : //! If we cannot relieve pressure by evicting layers outside of the reservation, we
30 : //! start evicting layers that are part of the reservation, LRU first.
31 : //!
32 : //! The value for the per-tenant reservation is referred to as `tenant_min_resident_size`
33 : //! throughout the code, but, no actual variable carries that name.
34 : //! The per-tenant default value is the `max(tenant's layer file sizes, regardless of local or remote)`.
35 : //! The idea is to allow at least one layer to be resident per tenant, to ensure it can make forward progress
36 : //! during page reconstruction.
37 : //! An alternative default for all tenants can be specified in the `tenant_config` section of the config.
38 : //! Lastly, each tenant can have an override in their respective tenant config (`min_resident_size_override`).
39 :
40 : // Implementation notes:
41 : // - The `#[allow(dead_code)]` above various structs are to suppress warnings about only the Debug impl
42 : // reading these fields. We use the Debug impl for semi-structured logging, though.
43 :
44 : use std::{
45 : sync::Arc,
46 : time::{Duration, SystemTime},
47 : };
48 :
49 : use anyhow::Context;
50 : use pageserver_api::shard::TenantShardId;
51 : use remote_storage::GenericRemoteStorage;
52 : use serde::{Deserialize, Serialize};
53 : use tokio::time::Instant;
54 : use tokio_util::sync::CancellationToken;
55 : use tracing::{debug, error, info, instrument, warn, Instrument};
56 : use utils::serde_percent::Percent;
57 : use utils::{completion, id::TimelineId};
58 :
59 : use crate::{
60 : config::PageServerConf,
61 : metrics::disk_usage_based_eviction::METRICS,
62 : task_mgr::{self, BACKGROUND_RUNTIME},
63 : tenant::{
64 : mgr::TenantManager,
65 : remote_timeline_client::LayerFileMetadata,
66 : secondary::SecondaryTenant,
67 : storage_layer::{AsLayerDesc, EvictionError, Layer, LayerName},
68 : },
69 : CancellableTask, DiskUsageEvictionTask,
70 : };
71 :
72 10 : #[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
73 : pub struct DiskUsageEvictionTaskConfig {
74 : pub max_usage_pct: Percent,
75 : pub min_avail_bytes: u64,
76 : #[serde(with = "humantime_serde")]
77 : pub period: Duration,
78 : #[cfg(feature = "testing")]
79 : pub mock_statvfs: Option<crate::statvfs::mock::Behavior>,
80 : /// Select sorting for evicted layers
81 : #[serde(default)]
82 : pub eviction_order: EvictionOrder,
83 : }
84 :
85 : /// Selects the sort order for eviction candidates *after* per tenant `min_resident_size`
86 : /// partitioning.
87 0 : #[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize, Deserialize)]
88 : #[serde(tag = "type", content = "args")]
89 : pub enum EvictionOrder {
90 : /// Order the layers to be evicted by how recently they have been accessed relatively within
91 : /// the set of resident layers of a tenant.
92 : RelativeAccessed {
93 : /// Determines if the tenant with most layers should lose first.
94 : ///
95 : /// Having this enabled is currently the only reasonable option, because the order in which
96 : /// we read tenants is deterministic. If we find the need to use this as `false`, we need
97 : /// to ensure nondeterminism by adding in a random number to break the
98 : /// `relative_last_activity==0.0` ties.
99 : #[serde(default = "default_highest_layer_count_loses_first")]
100 : highest_layer_count_loses_first: bool,
101 : },
102 : }
103 :
104 : impl Default for EvictionOrder {
105 6 : fn default() -> Self {
106 6 : Self::RelativeAccessed {
107 6 : highest_layer_count_loses_first: true,
108 6 : }
109 6 : }
110 : }
111 :
112 0 : fn default_highest_layer_count_loses_first() -> bool {
113 0 : true
114 0 : }
115 :
116 : impl EvictionOrder {
117 0 : fn sort(&self, candidates: &mut [(MinResidentSizePartition, EvictionCandidate)]) {
118 0 : use EvictionOrder::*;
119 0 :
120 0 : match self {
121 0 : RelativeAccessed { .. } => candidates.sort_unstable_by_key(|(partition, candidate)| {
122 0 : (*partition, candidate.relative_last_activity)
123 0 : }),
124 0 : }
125 0 : }
126 :
127 : /// Called to fill in the [`EvictionCandidate::relative_last_activity`] while iterating tenants
128 : /// layers in **most** recently used order.
129 40 : fn relative_last_activity(&self, total: usize, index: usize) -> finite_f32::FiniteF32 {
130 40 : use EvictionOrder::*;
131 40 :
132 40 : match self {
133 40 : RelativeAccessed {
134 40 : highest_layer_count_loses_first,
135 : } => {
136 : // keeping the -1 or not decides if every tenant should lose their least recently accessed
137 : // layer OR if this should happen in the order of having highest layer count:
138 40 : let fudge = if *highest_layer_count_loses_first {
139 : // relative_last_activity vs. tenant layer count:
140 : // - 0.1..=1.0 (10 layers)
141 : // - 0.01..=1.0 (100 layers)
142 : // - 0.001..=1.0 (1000 layers)
143 : //
144 : // leading to evicting less of the smallest tenants.
145 20 : 0
146 : } else {
147 : // use full 0.0..=1.0 range, which means even the smallest tenants could always lose a
148 : // layer. the actual ordering is unspecified: for 10k tenants on a pageserver it could
149 : // be that less than 10k layer evictions is enough, so we would not need to evict from
150 : // all tenants.
151 : //
152 : // as the tenant ordering is now deterministic this could hit the same tenants
153 : // disproportionetly on multiple invocations. alternative could be to remember how many
154 : // layers did we evict last time from this tenant, and inject that as an additional
155 : // fudge here.
156 20 : 1
157 : };
158 :
159 40 : let total = total.checked_sub(fudge).filter(|&x| x > 1).unwrap_or(1);
160 40 : let divider = total as f32;
161 40 :
162 40 : // most recently used is always (total - 0) / divider == 1.0
163 40 : // least recently used depends on the fudge:
164 40 : // - (total - 1) - (total - 1) / total => 0 / total
165 40 : // - total - (total - 1) / total => 1 / total
166 40 : let distance = (total - index) as f32;
167 40 :
168 40 : finite_f32::FiniteF32::try_from_normalized(distance / divider)
169 40 : .unwrap_or_else(|val| {
170 0 : tracing::warn!(%fudge, "calculated invalid relative_last_activity for i={index}, total={total}: {val}");
171 0 : finite_f32::FiniteF32::ZERO
172 40 : })
173 40 : }
174 40 : }
175 40 : }
176 : }
177 :
178 : #[derive(Default)]
179 : pub struct State {
180 : /// Exclude http requests and background task from running at the same time.
181 : mutex: tokio::sync::Mutex<()>,
182 : }
183 :
184 0 : pub fn launch_disk_usage_global_eviction_task(
185 0 : conf: &'static PageServerConf,
186 0 : storage: GenericRemoteStorage,
187 0 : state: Arc<State>,
188 0 : tenant_manager: Arc<TenantManager>,
189 0 : background_jobs_barrier: completion::Barrier,
190 0 : ) -> Option<DiskUsageEvictionTask> {
191 0 : let Some(task_config) = &conf.disk_usage_based_eviction else {
192 0 : info!("disk usage based eviction task not configured");
193 0 : return None;
194 : };
195 :
196 0 : info!("launching disk usage based eviction task");
197 :
198 0 : let cancel = CancellationToken::new();
199 0 : let task = BACKGROUND_RUNTIME.spawn(task_mgr::exit_on_panic_or_error(
200 0 : "disk usage based eviction",
201 0 : {
202 0 : let cancel = cancel.clone();
203 0 : async move {
204 : // wait until initial load is complete, because we cannot evict from loading tenants.
205 : tokio::select! {
206 : _ = cancel.cancelled() => { return anyhow::Ok(()); },
207 : _ = background_jobs_barrier.wait() => { }
208 : };
209 :
210 0 : disk_usage_eviction_task(&state, task_config, &storage, tenant_manager, cancel)
211 0 : .await;
212 0 : anyhow::Ok(())
213 0 : }
214 0 : },
215 0 : ));
216 0 :
217 0 : Some(DiskUsageEvictionTask(CancellableTask { cancel, task }))
218 0 : }
219 :
220 0 : #[instrument(skip_all)]
221 : async fn disk_usage_eviction_task(
222 : state: &State,
223 : task_config: &DiskUsageEvictionTaskConfig,
224 : storage: &GenericRemoteStorage,
225 : tenant_manager: Arc<TenantManager>,
226 : cancel: CancellationToken,
227 : ) {
228 : scopeguard::defer! {
229 : info!("disk usage based eviction task finishing");
230 : };
231 :
232 : use crate::tenant::tasks::random_init_delay;
233 : {
234 : if random_init_delay(task_config.period, &cancel)
235 : .await
236 : .is_err()
237 : {
238 : return;
239 : }
240 : }
241 :
242 : let mut iteration_no = 0;
243 : loop {
244 : iteration_no += 1;
245 : let start = Instant::now();
246 :
247 0 : async {
248 0 : let res = disk_usage_eviction_task_iteration(
249 0 : state,
250 0 : task_config,
251 0 : storage,
252 0 : &tenant_manager,
253 0 : &cancel,
254 0 : )
255 0 : .await;
256 :
257 0 : match res {
258 0 : Ok(()) => {}
259 0 : Err(e) => {
260 0 : // these stat failures are expected to be very rare
261 0 : warn!("iteration failed, unexpected error: {e:#}");
262 : }
263 : }
264 0 : }
265 : .instrument(tracing::info_span!("iteration", iteration_no))
266 : .await;
267 :
268 : let sleep_until = start + task_config.period;
269 : if tokio::time::timeout_at(sleep_until, cancel.cancelled())
270 : .await
271 : .is_ok()
272 : {
273 : break;
274 : }
275 : }
276 : }
277 :
278 : pub trait Usage: Clone + Copy + std::fmt::Debug {
279 : fn has_pressure(&self) -> bool;
280 : fn add_available_bytes(&mut self, bytes: u64);
281 : }
282 :
283 0 : async fn disk_usage_eviction_task_iteration(
284 0 : state: &State,
285 0 : task_config: &DiskUsageEvictionTaskConfig,
286 0 : storage: &GenericRemoteStorage,
287 0 : tenant_manager: &Arc<TenantManager>,
288 0 : cancel: &CancellationToken,
289 0 : ) -> anyhow::Result<()> {
290 0 : let tenants_dir = tenant_manager.get_conf().tenants_path();
291 0 : let usage_pre = filesystem_level_usage::get(&tenants_dir, task_config)
292 0 : .context("get filesystem-level disk usage before evictions")?;
293 0 : let res = disk_usage_eviction_task_iteration_impl(
294 0 : state,
295 0 : storage,
296 0 : usage_pre,
297 0 : tenant_manager,
298 0 : task_config.eviction_order,
299 0 : cancel,
300 0 : )
301 0 : .await;
302 0 : match res {
303 0 : Ok(outcome) => {
304 0 : debug!(?outcome, "disk_usage_eviction_iteration finished");
305 0 : match outcome {
306 0 : IterationOutcome::NoPressure | IterationOutcome::Cancelled => {
307 0 : // nothing to do, select statement below will handle things
308 0 : }
309 0 : IterationOutcome::Finished(outcome) => {
310 : // Verify with statvfs whether we made any real progress
311 0 : let after = filesystem_level_usage::get(&tenants_dir, task_config)
312 0 : // It's quite unlikely to hit the error here. Keep the code simple and bail out.
313 0 : .context("get filesystem-level disk usage after evictions")?;
314 :
315 0 : debug!(?after, "disk usage");
316 :
317 0 : if after.has_pressure() {
318 : // Don't bother doing an out-of-order iteration here now.
319 : // In practice, the task period is set to a value in the tens-of-seconds range,
320 : // which will cause another iteration to happen soon enough.
321 : // TODO: deltas between the three different usages would be helpful,
322 : // consider MiB, GiB, TiB
323 0 : warn!(?outcome, ?after, "disk usage still high");
324 : } else {
325 0 : info!(?outcome, ?after, "disk usage pressure relieved");
326 : }
327 : }
328 : }
329 : }
330 0 : Err(e) => {
331 0 : error!("disk_usage_eviction_iteration failed: {:#}", e);
332 : }
333 : }
334 :
335 0 : Ok(())
336 0 : }
337 :
338 : #[derive(Debug, Serialize)]
339 : #[allow(clippy::large_enum_variant)]
340 : pub enum IterationOutcome<U> {
341 : NoPressure,
342 : Cancelled,
343 : Finished(IterationOutcomeFinished<U>),
344 : }
345 :
346 : #[derive(Debug, Serialize)]
347 : pub struct IterationOutcomeFinished<U> {
348 : /// The actual usage observed before we started the iteration.
349 : before: U,
350 : /// The expected value for `after`, according to internal accounting, after phase 1.
351 : planned: PlannedUsage<U>,
352 : /// The outcome of phase 2, where we actually do the evictions.
353 : ///
354 : /// If all layers that phase 1 planned to evict _can_ actually get evicted, this will
355 : /// be the same as `planned`.
356 : assumed: AssumedUsage<U>,
357 : }
358 :
359 : #[derive(Debug, Serialize)]
360 : struct AssumedUsage<U> {
361 : /// The expected value for `after`, after phase 2.
362 : projected_after: U,
363 : /// The layers we failed to evict during phase 2.
364 : failed: LayerCount,
365 : }
366 :
367 : #[derive(Debug, Serialize)]
368 : struct PlannedUsage<U> {
369 : respecting_tenant_min_resident_size: U,
370 : fallback_to_global_lru: Option<U>,
371 : }
372 :
373 : #[derive(Debug, Default, Serialize)]
374 : struct LayerCount {
375 : file_sizes: u64,
376 : count: usize,
377 : }
378 :
379 0 : pub(crate) async fn disk_usage_eviction_task_iteration_impl<U: Usage>(
380 0 : state: &State,
381 0 : _storage: &GenericRemoteStorage,
382 0 : usage_pre: U,
383 0 : tenant_manager: &Arc<TenantManager>,
384 0 : eviction_order: EvictionOrder,
385 0 : cancel: &CancellationToken,
386 0 : ) -> anyhow::Result<IterationOutcome<U>> {
387 : // use tokio's mutex to get a Sync guard (instead of std::sync::Mutex)
388 0 : let _g = state
389 0 : .mutex
390 0 : .try_lock()
391 0 : .map_err(|_| anyhow::anyhow!("iteration is already executing"))?;
392 :
393 0 : debug!(?usage_pre, "disk usage");
394 :
395 0 : if !usage_pre.has_pressure() {
396 0 : return Ok(IterationOutcome::NoPressure);
397 0 : }
398 0 :
399 0 : warn!(
400 : ?usage_pre,
401 0 : "running disk usage based eviction due to pressure"
402 : );
403 :
404 0 : let (candidates, collection_time) = {
405 0 : let started_at = std::time::Instant::now();
406 0 : match collect_eviction_candidates(tenant_manager, eviction_order, cancel).await? {
407 : EvictionCandidates::Cancelled => {
408 0 : return Ok(IterationOutcome::Cancelled);
409 : }
410 0 : EvictionCandidates::Finished(partitioned) => (partitioned, started_at.elapsed()),
411 0 : }
412 0 : };
413 0 :
414 0 : METRICS.layers_collected.inc_by(candidates.len() as u64);
415 0 :
416 0 : tracing::info!(
417 0 : elapsed_ms = collection_time.as_millis(),
418 0 : total_layers = candidates.len(),
419 0 : "collection completed"
420 : );
421 :
422 : // Debug-log the list of candidates
423 0 : let now = SystemTime::now();
424 0 : for (i, (partition, candidate)) in candidates.iter().enumerate() {
425 0 : let nth = i + 1;
426 0 : let total_candidates = candidates.len();
427 0 : let size = candidate.layer.get_file_size();
428 0 : let rel = candidate.relative_last_activity;
429 0 : debug!(
430 0 : "cand {nth}/{total_candidates}: size={size}, rel_last_activity={rel}, no_access_for={}us, partition={partition:?}, {}/{}/{}",
431 0 : now.duration_since(candidate.last_activity_ts)
432 0 : .unwrap()
433 0 : .as_micros(),
434 0 : candidate.layer.get_tenant_shard_id(),
435 0 : candidate.layer.get_timeline_id(),
436 0 : candidate.layer.get_name(),
437 : );
438 : }
439 :
440 : // phase1: select victims to relieve pressure
441 : //
442 : // Walk through the list of candidates, until we have accumulated enough layers to get
443 : // us back under the pressure threshold. 'usage_planned' is updated so that it tracks
444 : // how much disk space would be used after evicting all the layers up to the current
445 : // point in the list.
446 : //
447 : // If we get far enough in the list that we start to evict layers that are below
448 : // the tenant's min-resident-size threshold, print a warning, and memorize the disk
449 : // usage at that point, in 'usage_planned_min_resident_size_respecting'.
450 :
451 0 : let (evicted_amount, usage_planned) =
452 0 : select_victims(&candidates, usage_pre).into_amount_and_planned();
453 0 :
454 0 : METRICS.layers_selected.inc_by(evicted_amount as u64);
455 0 :
456 0 : // phase2: evict layers
457 0 :
458 0 : let mut js = tokio::task::JoinSet::new();
459 0 : let limit = 1000;
460 0 :
461 0 : let mut evicted = candidates.into_iter().take(evicted_amount).fuse();
462 0 : let mut consumed_all = false;
463 0 :
464 0 : // After the evictions, `usage_assumed` is the post-eviction usage,
465 0 : // according to internal accounting.
466 0 : let mut usage_assumed = usage_pre;
467 0 : let mut evictions_failed = LayerCount::default();
468 0 :
469 0 : let evict_layers = async move {
470 : loop {
471 0 : let next = if js.len() >= limit || consumed_all {
472 0 : js.join_next().await
473 0 : } else if !js.is_empty() {
474 : // opportunistically consume ready result, one per each new evicted
475 0 : futures::future::FutureExt::now_or_never(js.join_next()).and_then(|x| x)
476 : } else {
477 0 : None
478 : };
479 :
480 0 : if let Some(next) = next {
481 0 : match next {
482 0 : Ok(Ok(file_size)) => {
483 0 : METRICS.layers_evicted.inc();
484 0 : usage_assumed.add_available_bytes(file_size);
485 0 : }
486 : Ok(Err((
487 0 : file_size,
488 0 : EvictionError::NotFound
489 0 : | EvictionError::Downloaded
490 0 : | EvictionError::Timeout,
491 0 : ))) => {
492 0 : evictions_failed.file_sizes += file_size;
493 0 : evictions_failed.count += 1;
494 0 : }
495 0 : Err(je) if je.is_cancelled() => unreachable!("not used"),
496 0 : Err(je) if je.is_panic() => { /* already logged */ }
497 0 : Err(je) => tracing::error!("unknown JoinError: {je:?}"),
498 : }
499 0 : }
500 :
501 0 : if consumed_all && js.is_empty() {
502 0 : break;
503 0 : }
504 :
505 : // calling again when consumed_all is fine as evicted is fused.
506 0 : let Some((_partition, candidate)) = evicted.next() else {
507 0 : if !consumed_all {
508 0 : tracing::info!("all evictions started, waiting");
509 0 : consumed_all = true;
510 0 : }
511 0 : continue;
512 : };
513 :
514 0 : match candidate.layer {
515 0 : EvictionLayer::Attached(layer) => {
516 0 : let file_size = layer.layer_desc().file_size;
517 0 : js.spawn(async move {
518 0 : // have a low eviction waiting timeout because our LRU calculations go stale fast;
519 0 : // also individual layer evictions could hang because of bugs and we do not want to
520 0 : // pause disk_usage_based_eviction for such.
521 0 : let timeout = std::time::Duration::from_secs(5);
522 0 :
523 0 : match layer.evict_and_wait(timeout).await {
524 0 : Ok(()) => Ok(file_size),
525 0 : Err(e) => Err((file_size, e)),
526 : }
527 0 : });
528 0 : }
529 0 : EvictionLayer::Secondary(layer) => {
530 0 : let file_size = layer.metadata.file_size;
531 0 :
532 0 : js.spawn(async move {
533 0 : layer
534 0 : .secondary_tenant
535 0 : .evict_layer(layer.timeline_id, layer.name)
536 0 : .await;
537 0 : Ok(file_size)
538 0 : });
539 0 : }
540 : }
541 0 : tokio::task::yield_now().await;
542 : }
543 :
544 0 : (usage_assumed, evictions_failed)
545 0 : };
546 :
547 0 : let started_at = std::time::Instant::now();
548 0 :
549 0 : let evict_layers = async move {
550 0 : let mut evict_layers = std::pin::pin!(evict_layers);
551 0 :
552 0 : let maximum_expected = std::time::Duration::from_secs(10);
553 :
554 0 : let res = tokio::time::timeout(maximum_expected, &mut evict_layers).await;
555 0 : let tuple = if let Ok(tuple) = res {
556 0 : tuple
557 : } else {
558 0 : let elapsed = started_at.elapsed();
559 0 : tracing::info!(elapsed_ms = elapsed.as_millis(), "still ongoing");
560 0 : evict_layers.await
561 : };
562 :
563 0 : let elapsed = started_at.elapsed();
564 0 : tracing::info!(elapsed_ms = elapsed.as_millis(), "completed");
565 0 : tuple
566 0 : };
567 :
568 0 : let evict_layers =
569 0 : evict_layers.instrument(tracing::info_span!("evict_layers", layers=%evicted_amount));
570 :
571 0 : let (usage_assumed, evictions_failed) = tokio::select! {
572 : tuple = evict_layers => { tuple },
573 : _ = cancel.cancelled() => {
574 : // dropping joinset will abort all pending evict_and_waits and that is fine, our
575 : // requests will still stand
576 : return Ok(IterationOutcome::Cancelled);
577 : }
578 : };
579 :
580 0 : Ok(IterationOutcome::Finished(IterationOutcomeFinished {
581 0 : before: usage_pre,
582 0 : planned: usage_planned,
583 0 : assumed: AssumedUsage {
584 0 : projected_after: usage_assumed,
585 0 : failed: evictions_failed,
586 0 : },
587 0 : }))
588 0 : }
589 :
590 : #[derive(Clone)]
591 : pub(crate) struct EvictionSecondaryLayer {
592 : pub(crate) secondary_tenant: Arc<SecondaryTenant>,
593 : pub(crate) timeline_id: TimelineId,
594 : pub(crate) name: LayerName,
595 : pub(crate) metadata: LayerFileMetadata,
596 : }
597 :
598 : /// Full [`Layer`] objects are specific to tenants in attached mode. This type is a layer
599 : /// of indirection to store either a `Layer`, or a reference to a secondary tenant and a layer name.
600 : #[derive(Clone)]
601 : pub(crate) enum EvictionLayer {
602 : Attached(Layer),
603 : Secondary(EvictionSecondaryLayer),
604 : }
605 :
606 : impl From<Layer> for EvictionLayer {
607 0 : fn from(value: Layer) -> Self {
608 0 : Self::Attached(value)
609 0 : }
610 : }
611 :
612 : impl EvictionLayer {
613 0 : pub(crate) fn get_tenant_shard_id(&self) -> &TenantShardId {
614 0 : match self {
615 0 : Self::Attached(l) => &l.layer_desc().tenant_shard_id,
616 0 : Self::Secondary(sl) => sl.secondary_tenant.get_tenant_shard_id(),
617 : }
618 0 : }
619 :
620 0 : pub(crate) fn get_timeline_id(&self) -> &TimelineId {
621 0 : match self {
622 0 : Self::Attached(l) => &l.layer_desc().timeline_id,
623 0 : Self::Secondary(sl) => &sl.timeline_id,
624 : }
625 0 : }
626 :
627 0 : pub(crate) fn get_name(&self) -> LayerName {
628 0 : match self {
629 0 : Self::Attached(l) => l.layer_desc().layer_name(),
630 0 : Self::Secondary(sl) => sl.name.clone(),
631 : }
632 0 : }
633 :
634 0 : pub(crate) fn get_file_size(&self) -> u64 {
635 0 : match self {
636 0 : Self::Attached(l) => l.layer_desc().file_size,
637 0 : Self::Secondary(sl) => sl.metadata.file_size,
638 : }
639 0 : }
640 : }
641 :
642 : #[derive(Clone)]
643 : pub(crate) struct EvictionCandidate {
644 : pub(crate) layer: EvictionLayer,
645 : pub(crate) last_activity_ts: SystemTime,
646 : pub(crate) relative_last_activity: finite_f32::FiniteF32,
647 : }
648 :
649 : impl std::fmt::Display for EvictionLayer {
650 0 : fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
651 0 : match self {
652 0 : Self::Attached(l) => l.fmt(f),
653 0 : Self::Secondary(sl) => {
654 0 : write!(f, "{}/{}", sl.timeline_id, sl.name)
655 : }
656 : }
657 0 : }
658 : }
659 :
660 : #[derive(Default)]
661 : pub(crate) struct DiskUsageEvictionInfo {
662 : /// Timeline's largest layer (remote or resident)
663 : pub max_layer_size: Option<u64>,
664 : /// Timeline's resident layers
665 : pub resident_layers: Vec<EvictionCandidate>,
666 : }
667 :
668 : impl std::fmt::Debug for EvictionCandidate {
669 0 : fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
670 0 : // format the tv_sec, tv_nsec into rfc3339 in case someone is looking at it
671 0 : // having to allocate a string to this is bad, but it will rarely be formatted
672 0 : let ts = chrono::DateTime::<chrono::Utc>::from(self.last_activity_ts);
673 0 : let ts = ts.to_rfc3339_opts(chrono::SecondsFormat::Nanos, true);
674 0 : struct DisplayIsDebug<'a, T>(&'a T);
675 0 : impl<'a, T: std::fmt::Display> std::fmt::Debug for DisplayIsDebug<'a, T> {
676 0 : fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
677 0 : write!(f, "{}", self.0)
678 0 : }
679 0 : }
680 0 : f.debug_struct("LocalLayerInfoForDiskUsageEviction")
681 0 : .field("layer", &DisplayIsDebug(&self.layer))
682 0 : .field("last_activity", &ts)
683 0 : .finish()
684 0 : }
685 : }
686 :
687 : #[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord)]
688 : enum MinResidentSizePartition {
689 : Above,
690 : Below,
691 : }
692 :
693 : enum EvictionCandidates {
694 : Cancelled,
695 : Finished(Vec<(MinResidentSizePartition, EvictionCandidate)>),
696 : }
697 :
698 : /// Gather the eviction candidates.
699 : ///
700 : /// The returned `Ok(EvictionCandidates::Finished(candidates))` is sorted in eviction
701 : /// order. A caller that evicts in that order, until pressure is relieved, implements
702 : /// the eviction policy outlined in the module comment.
703 : ///
704 : /// # Example with EvictionOrder::AbsoluteAccessed
705 : ///
706 : /// Imagine that there are two tenants, A and B, with five layers each, a-e.
707 : /// Each layer has size 100, and both tenant's min_resident_size is 150.
708 : /// The eviction order would be
709 : ///
710 : /// ```text
711 : /// partition last_activity_ts tenant/layer
712 : /// Above 18:30 A/c
713 : /// Above 19:00 A/b
714 : /// Above 18:29 B/c
715 : /// Above 19:05 B/b
716 : /// Above 20:00 B/a
717 : /// Above 20:03 A/a
718 : /// Below 20:30 A/d
719 : /// Below 20:40 B/d
720 : /// Below 20:45 B/e
721 : /// Below 20:58 A/e
722 : /// ```
723 : ///
724 : /// Now, if we need to evict 300 bytes to relieve pressure, we'd evict `A/c, A/b, B/c`.
725 : /// They are all in the `Above` partition, so, we respected each tenant's min_resident_size.
726 : ///
727 : /// But, if we need to evict 900 bytes to relieve pressure, we'd evict
728 : /// `A/c, A/b, B/c, B/b, B/a, A/a, A/d, B/d, B/e`, reaching into the `Below` partition
729 : /// after exhauting the `Above` partition.
730 : /// So, we did not respect each tenant's min_resident_size.
731 : ///
732 : /// # Example with EvictionOrder::RelativeAccessed
733 : ///
734 : /// ```text
735 : /// partition relative_age last_activity_ts tenant/layer
736 : /// Above 0/4 18:30 A/c
737 : /// Above 0/4 18:29 B/c
738 : /// Above 1/4 19:00 A/b
739 : /// Above 1/4 19:05 B/b
740 : /// Above 2/4 20:00 B/a
741 : /// Above 2/4 20:03 A/a
742 : /// Below 3/4 20:30 A/d
743 : /// Below 3/4 20:40 B/d
744 : /// Below 4/4 20:45 B/e
745 : /// Below 4/4 20:58 A/e
746 : /// ```
747 : ///
748 : /// With tenants having the same number of layers the picture does not change much. The same with
749 : /// A having many more layers **resident** (not all of them listed):
750 : ///
751 : /// ```text
752 : /// Above 0/100 18:30 A/c
753 : /// Above 0/4 18:29 B/c
754 : /// Above 1/100 19:00 A/b
755 : /// Above 2/100 20:03 A/a
756 : /// Above 3/100 20:03 A/nth_3
757 : /// Above 4/100 20:03 A/nth_4
758 : /// ...
759 : /// Above 1/4 19:05 B/b
760 : /// Above 25/100 20:04 A/nth_25
761 : /// ...
762 : /// Above 2/4 20:00 B/a
763 : /// Above 50/100 20:10 A/nth_50
764 : /// ...
765 : /// Below 3/4 20:40 B/d
766 : /// Below 99/100 20:30 A/nth_99
767 : /// Below 4/4 20:45 B/e
768 : /// Below 100/100 20:58 A/nth_100
769 : /// ```
770 : ///
771 : /// Now it's easier to see that because A has grown fast it has more layers to get evicted. What is
772 : /// difficult to see is what happens on the next round assuming the evicting 23 from the above list
773 : /// relieves the pressure (22 A layers gone, 1 B layers gone) but a new fast growing tenant C has
774 : /// appeared:
775 : ///
776 : /// ```text
777 : /// Above 0/87 20:04 A/nth_23
778 : /// Above 0/3 19:05 B/b
779 : /// Above 0/50 20:59 C/nth_0
780 : /// Above 1/87 20:04 A/nth_24
781 : /// Above 1/50 21:00 C/nth_1
782 : /// Above 2/87 20:04 A/nth_25
783 : /// ...
784 : /// Above 16/50 21:02 C/nth_16
785 : /// Above 1/3 20:00 B/a
786 : /// Above 27/87 20:10 A/nth_50
787 : /// ...
788 : /// Below 2/3 20:40 B/d
789 : /// Below 49/50 21:05 C/nth_49
790 : /// Below 86/87 20:30 A/nth_99
791 : /// Below 3/3 20:45 B/e
792 : /// Below 50/50 21:05 C/nth_50
793 : /// Below 87/87 20:58 A/nth_100
794 : /// ```
795 : ///
796 : /// Now relieving pressure with 23 layers would cost:
797 : /// - tenant A 14 layers
798 : /// - tenant B 1 layer
799 : /// - tenant C 8 layers
800 0 : async fn collect_eviction_candidates(
801 0 : tenant_manager: &Arc<TenantManager>,
802 0 : eviction_order: EvictionOrder,
803 0 : cancel: &CancellationToken,
804 0 : ) -> anyhow::Result<EvictionCandidates> {
805 : const LOG_DURATION_THRESHOLD: std::time::Duration = std::time::Duration::from_secs(10);
806 :
807 : // get a snapshot of the list of tenants
808 0 : let tenants = tenant_manager
809 0 : .list_tenants()
810 0 : .context("get list of tenants")?;
811 :
812 : // TODO: avoid listing every layer in every tenant: this loop can block the executor,
813 : // and the resulting data structure can be huge.
814 : // (https://github.com/neondatabase/neon/issues/6224)
815 0 : let mut candidates = Vec::new();
816 :
817 0 : for (tenant_id, _state, _gen) in tenants {
818 0 : if cancel.is_cancelled() {
819 0 : return Ok(EvictionCandidates::Cancelled);
820 0 : }
821 0 : let tenant = match tenant_manager.get_attached_tenant_shard(tenant_id) {
822 0 : Ok(tenant) if tenant.is_active() => tenant,
823 : Ok(_) => {
824 0 : debug!(tenant_id=%tenant_id.tenant_id, shard_id=%tenant_id.shard_slug(), "Tenant shard is not active");
825 0 : continue;
826 : }
827 0 : Err(e) => {
828 0 : // this can happen if tenant has lifecycle transition after we fetched it
829 0 : debug!("failed to get tenant: {e:#}");
830 0 : continue;
831 : }
832 : };
833 :
834 0 : if tenant.cancel.is_cancelled() {
835 0 : info!(%tenant_id, "Skipping tenant for eviction, it is shutting down");
836 0 : continue;
837 0 : }
838 0 :
839 0 : let started_at = std::time::Instant::now();
840 0 :
841 0 : // collect layers from all timelines in this tenant
842 0 : //
843 0 : // If one of the timelines becomes `!is_active()` during the iteration,
844 0 : // for example because we're shutting down, then `max_layer_size` can be too small.
845 0 : // That's OK. This code only runs under a disk pressure situation, and being
846 0 : // a little unfair to tenants during shutdown in such a situation is tolerable.
847 0 : let mut tenant_candidates = Vec::new();
848 0 : let mut max_layer_size = 0;
849 0 : for tl in tenant.list_timelines() {
850 0 : if !tl.is_active() {
851 0 : continue;
852 0 : }
853 0 : let info = tl.get_local_layers_for_disk_usage_eviction().await;
854 0 : debug!(tenant_id=%tl.tenant_shard_id.tenant_id, shard_id=%tl.tenant_shard_id.shard_slug(), timeline_id=%tl.timeline_id, "timeline resident layers count: {}", info.resident_layers.len());
855 :
856 0 : tenant_candidates.extend(info.resident_layers.into_iter());
857 0 : max_layer_size = max_layer_size.max(info.max_layer_size.unwrap_or(0));
858 0 :
859 0 : if cancel.is_cancelled() {
860 0 : return Ok(EvictionCandidates::Cancelled);
861 0 : }
862 : }
863 :
864 : // `min_resident_size` defaults to maximum layer file size of the tenant.
865 : // This ensures that each tenant can have at least one layer resident at a given time,
866 : // ensuring forward progress for a single Timeline::get in that tenant.
867 : // It's a questionable heuristic since, usually, there are many Timeline::get
868 : // requests going on for a tenant, and, at least in Neon prod, the median
869 : // layer file size is much smaller than the compaction target size.
870 : // We could be better here, e.g., sum of all L0 layers + most recent L1 layer.
871 : // That's what's typically used by the various background loops.
872 : //
873 : // The default can be overridden with a fixed value in the tenant conf.
874 : // A default override can be put in the default tenant conf in the pageserver.toml.
875 0 : let min_resident_size = if let Some(s) = tenant.get_min_resident_size_override() {
876 0 : debug!(
877 0 : tenant_id=%tenant.tenant_shard_id().tenant_id,
878 0 : shard_id=%tenant.tenant_shard_id().shard_slug(),
879 0 : overridden_size=s,
880 0 : "using overridden min resident size for tenant"
881 : );
882 0 : s
883 : } else {
884 0 : debug!(
885 0 : tenant_id=%tenant.tenant_shard_id().tenant_id,
886 0 : shard_id=%tenant.tenant_shard_id().shard_slug(),
887 0 : max_layer_size,
888 0 : "using max layer size as min_resident_size for tenant",
889 : );
890 0 : max_layer_size
891 : };
892 :
893 : // Sort layers most-recently-used first, then partition by
894 : // cumsum above/below min_resident_size.
895 0 : tenant_candidates
896 0 : .sort_unstable_by_key(|layer_info| std::cmp::Reverse(layer_info.last_activity_ts));
897 0 : let mut cumsum: i128 = 0;
898 0 :
899 0 : let total = tenant_candidates.len();
900 0 :
901 0 : let tenant_candidates =
902 0 : tenant_candidates
903 0 : .into_iter()
904 0 : .enumerate()
905 0 : .map(|(i, mut candidate)| {
906 0 : // as we iterate this reverse sorted list, the most recently accessed layer will always
907 0 : // be 1.0; this is for us to evict it last.
908 0 : candidate.relative_last_activity =
909 0 : eviction_order.relative_last_activity(total, i);
910 :
911 0 : let partition = if cumsum > min_resident_size as i128 {
912 0 : MinResidentSizePartition::Above
913 : } else {
914 0 : MinResidentSizePartition::Below
915 : };
916 0 : cumsum += i128::from(candidate.layer.get_file_size());
917 0 :
918 0 : (partition, candidate)
919 0 : });
920 0 :
921 0 : METRICS
922 0 : .tenant_layer_count
923 0 : .observe(tenant_candidates.len() as f64);
924 0 :
925 0 : candidates.extend(tenant_candidates);
926 0 :
927 0 : let elapsed = started_at.elapsed();
928 0 : METRICS
929 0 : .tenant_collection_time
930 0 : .observe(elapsed.as_secs_f64());
931 0 :
932 0 : if elapsed > LOG_DURATION_THRESHOLD {
933 0 : tracing::info!(
934 0 : tenant_id=%tenant.tenant_shard_id().tenant_id,
935 0 : shard_id=%tenant.tenant_shard_id().shard_slug(),
936 0 : elapsed_ms = elapsed.as_millis(),
937 0 : "collection took longer than threshold"
938 : );
939 0 : }
940 : }
941 :
942 : // Note: the same tenant ID might be hit twice, if it transitions from attached to
943 : // secondary while we run. That is okay: when we eventually try and run the eviction,
944 : // the `Gate` on the object will ensure that whichever one has already been shut down
945 : // will not delete anything.
946 :
947 0 : let mut secondary_tenants = Vec::new();
948 0 : tenant_manager.foreach_secondary_tenants(
949 0 : |_tenant_shard_id: &TenantShardId, state: &Arc<SecondaryTenant>| {
950 0 : secondary_tenants.push(state.clone());
951 0 : },
952 0 : );
953 :
954 0 : for tenant in secondary_tenants {
955 : // for secondary tenants we use a sum of on_disk layers and already evicted layers. this is
956 : // to prevent repeated disk usage based evictions from completely draining less often
957 : // updating secondaries.
958 0 : let (mut layer_info, total_layers) = tenant.get_layers_for_eviction();
959 0 :
960 0 : debug_assert!(
961 0 : total_layers >= layer_info.resident_layers.len(),
962 0 : "total_layers ({total_layers}) must be at least the resident_layers.len() ({})",
963 0 : layer_info.resident_layers.len()
964 : );
965 :
966 0 : let started_at = std::time::Instant::now();
967 0 :
968 0 : layer_info
969 0 : .resident_layers
970 0 : .sort_unstable_by_key(|layer_info| std::cmp::Reverse(layer_info.last_activity_ts));
971 0 :
972 0 : let tenant_candidates =
973 0 : layer_info
974 0 : .resident_layers
975 0 : .into_iter()
976 0 : .enumerate()
977 0 : .map(|(i, mut candidate)| {
978 0 : candidate.relative_last_activity =
979 0 : eviction_order.relative_last_activity(total_layers, i);
980 0 : (
981 0 : // Secondary locations' layers are always considered above the min resident size,
982 0 : // i.e. secondary locations are permitted to be trimmed to zero layers if all
983 0 : // the layers have sufficiently old access times.
984 0 : MinResidentSizePartition::Above,
985 0 : candidate,
986 0 : )
987 0 : });
988 0 :
989 0 : METRICS
990 0 : .tenant_layer_count
991 0 : .observe(tenant_candidates.len() as f64);
992 0 : candidates.extend(tenant_candidates);
993 0 :
994 0 : tokio::task::yield_now().await;
995 :
996 0 : let elapsed = started_at.elapsed();
997 0 :
998 0 : METRICS
999 0 : .tenant_collection_time
1000 0 : .observe(elapsed.as_secs_f64());
1001 0 :
1002 0 : if elapsed > LOG_DURATION_THRESHOLD {
1003 0 : tracing::info!(
1004 0 : tenant_id=%tenant.tenant_shard_id().tenant_id,
1005 0 : shard_id=%tenant.tenant_shard_id().shard_slug(),
1006 0 : elapsed_ms = elapsed.as_millis(),
1007 0 : "collection took longer than threshold"
1008 : );
1009 0 : }
1010 : }
1011 :
1012 0 : debug_assert!(MinResidentSizePartition::Above < MinResidentSizePartition::Below,
1013 0 : "as explained in the function's doc comment, layers that aren't in the tenant's min_resident_size are evicted first");
1014 :
1015 0 : eviction_order.sort(&mut candidates);
1016 0 :
1017 0 : Ok(EvictionCandidates::Finished(candidates))
1018 0 : }
1019 :
1020 : /// Given a pre-sorted vec of all layers in the system, select the first N which are enough to
1021 : /// relieve pressure.
1022 : ///
1023 : /// Returns the amount of candidates selected, with the planned usage.
1024 0 : fn select_victims<U: Usage>(
1025 0 : candidates: &[(MinResidentSizePartition, EvictionCandidate)],
1026 0 : usage_pre: U,
1027 0 : ) -> VictimSelection<U> {
1028 0 : let mut usage_when_switched = None;
1029 0 : let mut usage_planned = usage_pre;
1030 0 : let mut evicted_amount = 0;
1031 :
1032 0 : for (i, (partition, candidate)) in candidates.iter().enumerate() {
1033 0 : if !usage_planned.has_pressure() {
1034 0 : break;
1035 0 : }
1036 0 :
1037 0 : if partition == &MinResidentSizePartition::Below && usage_when_switched.is_none() {
1038 0 : usage_when_switched = Some((usage_planned, i));
1039 0 : }
1040 :
1041 0 : usage_planned.add_available_bytes(candidate.layer.get_file_size());
1042 0 : evicted_amount += 1;
1043 : }
1044 :
1045 0 : VictimSelection {
1046 0 : amount: evicted_amount,
1047 0 : usage_pre,
1048 0 : usage_when_switched,
1049 0 : usage_planned,
1050 0 : }
1051 0 : }
1052 :
1053 : struct VictimSelection<U> {
1054 : amount: usize,
1055 : usage_pre: U,
1056 : usage_when_switched: Option<(U, usize)>,
1057 : usage_planned: U,
1058 : }
1059 :
1060 : impl<U: Usage> VictimSelection<U> {
1061 0 : fn into_amount_and_planned(self) -> (usize, PlannedUsage<U>) {
1062 0 : debug!(
1063 : evicted_amount=%self.amount,
1064 0 : "took enough candidates for pressure to be relieved"
1065 : );
1066 :
1067 0 : if let Some((usage_planned, candidate_no)) = self.usage_when_switched.as_ref() {
1068 0 : warn!(usage_pre=?self.usage_pre, ?usage_planned, candidate_no, "tenant_min_resident_size-respecting LRU would not relieve pressure, evicting more following global LRU policy");
1069 0 : }
1070 :
1071 0 : let planned = match self.usage_when_switched {
1072 0 : Some((respecting_tenant_min_resident_size, _)) => PlannedUsage {
1073 0 : respecting_tenant_min_resident_size,
1074 0 : fallback_to_global_lru: Some(self.usage_planned),
1075 0 : },
1076 0 : None => PlannedUsage {
1077 0 : respecting_tenant_min_resident_size: self.usage_planned,
1078 0 : fallback_to_global_lru: None,
1079 0 : },
1080 : };
1081 :
1082 0 : (self.amount, planned)
1083 0 : }
1084 : }
1085 :
1086 : /// A totally ordered f32 subset we can use with sorting functions.
1087 : pub(crate) mod finite_f32 {
1088 :
1089 : /// A totally ordered f32 subset we can use with sorting functions.
1090 : #[derive(Clone, Copy, PartialEq)]
1091 : pub struct FiniteF32(f32);
1092 :
1093 : impl std::fmt::Debug for FiniteF32 {
1094 0 : fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
1095 0 : std::fmt::Debug::fmt(&self.0, f)
1096 0 : }
1097 : }
1098 :
1099 : impl std::fmt::Display for FiniteF32 {
1100 0 : fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
1101 0 : std::fmt::Display::fmt(&self.0, f)
1102 0 : }
1103 : }
1104 :
1105 : impl std::cmp::Eq for FiniteF32 {}
1106 :
1107 : impl std::cmp::PartialOrd for FiniteF32 {
1108 0 : fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
1109 0 : Some(self.cmp(other))
1110 0 : }
1111 : }
1112 :
1113 : impl std::cmp::Ord for FiniteF32 {
1114 0 : fn cmp(&self, other: &Self) -> std::cmp::Ordering {
1115 0 : self.0.total_cmp(&other.0)
1116 0 : }
1117 : }
1118 :
1119 : impl TryFrom<f32> for FiniteF32 {
1120 : type Error = f32;
1121 :
1122 0 : fn try_from(value: f32) -> Result<Self, Self::Error> {
1123 0 : if value.is_finite() {
1124 0 : Ok(FiniteF32(value))
1125 : } else {
1126 0 : Err(value)
1127 : }
1128 0 : }
1129 : }
1130 :
1131 : impl From<FiniteF32> for f32 {
1132 40 : fn from(value: FiniteF32) -> f32 {
1133 40 : value.0
1134 40 : }
1135 : }
1136 :
1137 : impl FiniteF32 {
1138 : pub const ZERO: FiniteF32 = FiniteF32(0.0);
1139 :
1140 40 : pub fn try_from_normalized(value: f32) -> Result<Self, f32> {
1141 40 : if (0.0..=1.0).contains(&value) {
1142 : // -0.0 is within the range, make sure it is assumed 0.0..=1.0
1143 40 : let value = value.abs();
1144 40 : Ok(FiniteF32(value))
1145 : } else {
1146 0 : Err(value)
1147 : }
1148 40 : }
1149 :
1150 40 : pub fn into_inner(self) -> f32 {
1151 40 : self.into()
1152 40 : }
1153 : }
1154 : }
1155 :
1156 : mod filesystem_level_usage {
1157 : use anyhow::Context;
1158 : use camino::Utf8Path;
1159 :
1160 : use crate::statvfs::Statvfs;
1161 :
1162 : use super::DiskUsageEvictionTaskConfig;
1163 :
1164 : #[derive(Debug, Clone, Copy)]
1165 : pub struct Usage<'a> {
1166 : config: &'a DiskUsageEvictionTaskConfig,
1167 :
1168 : /// Filesystem capacity
1169 : total_bytes: u64,
1170 : /// Free filesystem space
1171 : avail_bytes: u64,
1172 : }
1173 :
1174 : impl super::Usage for Usage<'_> {
1175 14 : fn has_pressure(&self) -> bool {
1176 14 : let usage_pct =
1177 14 : (100.0 * (1.0 - ((self.avail_bytes as f64) / (self.total_bytes as f64)))) as u64;
1178 14 :
1179 14 : let pressures = [
1180 14 : (
1181 14 : "min_avail_bytes",
1182 14 : self.avail_bytes < self.config.min_avail_bytes,
1183 14 : ),
1184 14 : (
1185 14 : "max_usage_pct",
1186 14 : usage_pct >= self.config.max_usage_pct.get() as u64,
1187 14 : ),
1188 14 : ];
1189 14 :
1190 28 : pressures.into_iter().any(|(_, has_pressure)| has_pressure)
1191 14 : }
1192 :
1193 12 : fn add_available_bytes(&mut self, bytes: u64) {
1194 12 : self.avail_bytes += bytes;
1195 12 : }
1196 : }
1197 :
1198 0 : pub fn get<'a>(
1199 0 : tenants_dir: &Utf8Path,
1200 0 : config: &'a DiskUsageEvictionTaskConfig,
1201 0 : ) -> anyhow::Result<Usage<'a>> {
1202 0 : let mock_config = {
1203 0 : #[cfg(feature = "testing")]
1204 0 : {
1205 0 : config.mock_statvfs.as_ref()
1206 : }
1207 : #[cfg(not(feature = "testing"))]
1208 : {
1209 : None
1210 : }
1211 : };
1212 :
1213 0 : let stat = Statvfs::get(tenants_dir, mock_config)
1214 0 : .context("statvfs failed, presumably directory got unlinked")?;
1215 :
1216 : // https://unix.stackexchange.com/a/703650
1217 0 : let blocksize = if stat.fragment_size() > 0 {
1218 0 : stat.fragment_size()
1219 : } else {
1220 0 : stat.block_size()
1221 : };
1222 :
1223 : // use blocks_available (b_avail) since, pageserver runs as unprivileged user
1224 0 : let avail_bytes = stat.blocks_available() * blocksize;
1225 0 : let total_bytes = stat.blocks() * blocksize;
1226 0 :
1227 0 : Ok(Usage {
1228 0 : config,
1229 0 : total_bytes,
1230 0 : avail_bytes,
1231 0 : })
1232 0 : }
1233 :
1234 : #[test]
1235 2 : fn max_usage_pct_pressure() {
1236 2 : use super::EvictionOrder;
1237 2 : use super::Usage as _;
1238 2 : use std::time::Duration;
1239 2 : use utils::serde_percent::Percent;
1240 2 :
1241 2 : let mut usage = Usage {
1242 2 : config: &DiskUsageEvictionTaskConfig {
1243 2 : max_usage_pct: Percent::new(85).unwrap(),
1244 2 : min_avail_bytes: 0,
1245 2 : period: Duration::MAX,
1246 2 : #[cfg(feature = "testing")]
1247 2 : mock_statvfs: None,
1248 2 : eviction_order: EvictionOrder::default(),
1249 2 : },
1250 2 : total_bytes: 100_000,
1251 2 : avail_bytes: 0,
1252 2 : };
1253 2 :
1254 2 : assert!(usage.has_pressure(), "expected pressure at 100%");
1255 :
1256 2 : usage.add_available_bytes(14_000);
1257 2 : assert!(usage.has_pressure(), "expected pressure at 86%");
1258 :
1259 2 : usage.add_available_bytes(999);
1260 2 : assert!(usage.has_pressure(), "expected pressure at 85.001%");
1261 :
1262 2 : usage.add_available_bytes(1);
1263 2 : assert!(usage.has_pressure(), "expected pressure at precisely 85%");
1264 :
1265 2 : usage.add_available_bytes(1);
1266 2 : assert!(!usage.has_pressure(), "no pressure at 84.999%");
1267 :
1268 2 : usage.add_available_bytes(999);
1269 2 : assert!(!usage.has_pressure(), "no pressure at 84%");
1270 :
1271 2 : usage.add_available_bytes(16_000);
1272 2 : assert!(!usage.has_pressure());
1273 2 : }
1274 : }
1275 :
1276 : #[cfg(test)]
1277 : mod tests {
1278 : use super::*;
1279 :
1280 : #[test]
1281 2 : fn relative_equal_bounds() {
1282 2 : let order = EvictionOrder::RelativeAccessed {
1283 2 : highest_layer_count_loses_first: false,
1284 2 : };
1285 2 :
1286 2 : let len = 10;
1287 2 : let v = (0..len)
1288 20 : .map(|i| order.relative_last_activity(len, i).into_inner())
1289 2 : .collect::<Vec<_>>();
1290 2 :
1291 2 : assert_eq!(v.first(), Some(&1.0));
1292 2 : assert_eq!(v.last(), Some(&0.0));
1293 18 : assert!(v.windows(2).all(|slice| slice[0] > slice[1]));
1294 2 : }
1295 :
1296 : #[test]
1297 2 : fn relative_spare_bounds() {
1298 2 : let order = EvictionOrder::RelativeAccessed {
1299 2 : highest_layer_count_loses_first: true,
1300 2 : };
1301 2 :
1302 2 : let len = 10;
1303 2 : let v = (0..len)
1304 20 : .map(|i| order.relative_last_activity(len, i).into_inner())
1305 2 : .collect::<Vec<_>>();
1306 2 :
1307 2 : assert_eq!(v.first(), Some(&1.0));
1308 2 : assert_eq!(v.last(), Some(&0.1));
1309 18 : assert!(v.windows(2).all(|slice| slice[0] > slice[1]));
1310 2 : }
1311 : }
|
