Line data Source code
1 : use std::collections::HashMap;
2 : use std::fmt::Display;
3 : use std::sync::Arc;
4 : use std::time::Duration;
5 :
6 : use anyhow::{Context, anyhow};
7 : use futures::StreamExt;
8 : use futures::future::Either;
9 : use pageserver_api::shard::ShardIdentity;
10 : use postgres_backend::{CopyStreamHandlerEnd, PostgresBackend};
11 : use postgres_ffi::get_current_timestamp;
12 : use postgres_ffi::waldecoder::{WalDecodeError, WalStreamDecoder};
13 : use pq_proto::{BeMessage, InterpretedWalRecordsBody, WalSndKeepAlive};
14 : use tokio::io::{AsyncRead, AsyncWrite};
15 : use tokio::sync::mpsc::error::SendError;
16 : use tokio::task::JoinHandle;
17 : use tokio::time::MissedTickBehavior;
18 : use tracing::{Instrument, error, info, info_span};
19 : use utils::critical;
20 : use utils::lsn::Lsn;
21 : use utils::postgres_client::{Compression, InterpretedFormat};
22 : use wal_decoder::models::{InterpretedWalRecord, InterpretedWalRecords};
23 : use wal_decoder::wire_format::ToWireFormat;
24 :
25 : use crate::metrics::WAL_READERS;
26 : use crate::send_wal::{EndWatchView, WalSenderGuard};
27 : use crate::timeline::WalResidentTimeline;
28 : use crate::wal_reader_stream::{StreamingWalReader, WalBytes};
29 :
30 : /// Identifier used to differentiate between senders of the same
31 : /// shard.
32 : ///
33 : /// In the steady state there's only one, but two pageservers may
34 : /// temporarily have the same shard attached and attempt to ingest
35 : /// WAL for it. See also [`ShardSenderId`].
36 : #[derive(Hash, Eq, PartialEq, Copy, Clone)]
37 : struct SenderId(u8);
38 :
39 : impl SenderId {
40 6 : fn first() -> Self {
41 6 : SenderId(0)
42 6 : }
43 :
44 2 : fn next(&self) -> Self {
45 2 : SenderId(self.0.checked_add(1).expect("few senders"))
46 2 : }
47 : }
48 :
49 : #[derive(Hash, Eq, PartialEq)]
50 : struct ShardSenderId {
51 : shard: ShardIdentity,
52 : sender_id: SenderId,
53 : }
54 :
55 : impl Display for ShardSenderId {
56 0 : fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
57 0 : write!(f, "{}{}", self.sender_id.0, self.shard.shard_slug())
58 0 : }
59 : }
60 :
61 : impl ShardSenderId {
62 877 : fn new(shard: ShardIdentity, sender_id: SenderId) -> Self {
63 877 : ShardSenderId { shard, sender_id }
64 877 : }
65 :
66 0 : fn shard(&self) -> ShardIdentity {
67 0 : self.shard
68 0 : }
69 : }
70 :
71 : /// Shard-aware fan-out interpreted record reader.
72 : /// Reads WAL from disk, decodes it, intepretets it, and sends
73 : /// it to any [`InterpretedWalSender`] connected to it.
74 : /// Each [`InterpretedWalSender`] corresponds to one shard
75 : /// and gets interpreted records concerning that shard only.
76 : pub(crate) struct InterpretedWalReader {
77 : wal_stream: StreamingWalReader,
78 : shard_senders: HashMap<ShardIdentity, smallvec::SmallVec<[ShardSenderState; 1]>>,
79 : shard_notification_rx: Option<tokio::sync::mpsc::UnboundedReceiver<AttachShardNotification>>,
80 : state: Arc<std::sync::RwLock<InterpretedWalReaderState>>,
81 : pg_version: u32,
82 : }
83 :
84 : /// A handle for [`InterpretedWalReader`] which allows for interacting with it
85 : /// when it runs as a separate tokio task.
86 : #[derive(Debug)]
87 : pub(crate) struct InterpretedWalReaderHandle {
88 : join_handle: JoinHandle<Result<(), InterpretedWalReaderError>>,
89 : state: Arc<std::sync::RwLock<InterpretedWalReaderState>>,
90 : shard_notification_tx: tokio::sync::mpsc::UnboundedSender<AttachShardNotification>,
91 : }
92 :
93 : struct ShardSenderState {
94 : sender_id: SenderId,
95 : tx: tokio::sync::mpsc::Sender<Batch>,
96 : next_record_lsn: Lsn,
97 : }
98 :
99 : /// State of [`InterpretedWalReader`] visible outside of the task running it.
100 : #[derive(Debug)]
101 : pub(crate) enum InterpretedWalReaderState {
102 : Running {
103 : current_position: Lsn,
104 : /// Tracks the start of the PG WAL LSN from which the current batch of
105 : /// interpreted records originated.
106 : current_batch_wal_start: Option<Lsn>,
107 : },
108 : Done,
109 : }
110 :
111 : pub(crate) struct Batch {
112 : wal_end_lsn: Lsn,
113 : available_wal_end_lsn: Lsn,
114 : records: InterpretedWalRecords,
115 : }
116 :
117 : #[derive(thiserror::Error, Debug)]
118 : pub enum InterpretedWalReaderError {
119 : /// Handler initiates the end of streaming.
120 : #[error("decode error: {0}")]
121 : Decode(#[from] WalDecodeError),
122 : #[error("read or interpret error: {0}")]
123 : ReadOrInterpret(#[from] anyhow::Error),
124 : #[error("wal stream closed")]
125 : WalStreamClosed,
126 : }
127 :
128 : enum CurrentPositionUpdate {
129 : Reset { from: Lsn, to: Lsn },
130 : NotReset(Lsn),
131 : }
132 :
133 : impl CurrentPositionUpdate {
134 0 : fn current_position(&self) -> Lsn {
135 0 : match self {
136 0 : CurrentPositionUpdate::Reset { from: _, to } => *to,
137 0 : CurrentPositionUpdate::NotReset(lsn) => *lsn,
138 : }
139 0 : }
140 :
141 0 : fn previous_position(&self) -> Lsn {
142 0 : match self {
143 0 : CurrentPositionUpdate::Reset { from, to: _ } => *from,
144 0 : CurrentPositionUpdate::NotReset(lsn) => *lsn,
145 : }
146 0 : }
147 : }
148 :
149 : impl InterpretedWalReaderState {
150 6 : fn current_position(&self) -> Option<Lsn> {
151 6 : match self {
152 : InterpretedWalReaderState::Running {
153 3 : current_position, ..
154 3 : } => Some(*current_position),
155 3 : InterpretedWalReaderState::Done => None,
156 : }
157 6 : }
158 :
159 : #[cfg(test)]
160 57 : fn current_batch_wal_start(&self) -> Option<Lsn> {
161 57 : match self {
162 : InterpretedWalReaderState::Running {
163 57 : current_batch_wal_start,
164 57 : ..
165 57 : } => *current_batch_wal_start,
166 0 : InterpretedWalReaderState::Done => None,
167 : }
168 57 : }
169 :
170 : // Reset the current position of the WAL reader if the requested starting position
171 : // of the new shard is smaller than the current value.
172 4 : fn maybe_reset(&mut self, new_shard_start_pos: Lsn) -> CurrentPositionUpdate {
173 4 : match self {
174 : InterpretedWalReaderState::Running {
175 4 : current_position,
176 4 : current_batch_wal_start,
177 4 : } => {
178 4 : if new_shard_start_pos < *current_position {
179 3 : let from = *current_position;
180 3 : *current_position = new_shard_start_pos;
181 3 : *current_batch_wal_start = None;
182 3 : CurrentPositionUpdate::Reset {
183 3 : from,
184 3 : to: *current_position,
185 3 : }
186 : } else {
187 : // Edge case: The new shard is at the same current position as
188 : // the reader. Note that the current position is WAL record aligned,
189 : // so the reader might have done some partial reads and updated the
190 : // batch start. If that's the case, adjust the batch start to match
191 : // starting position of the new shard. It can lead to some shards
192 : // seeing overlaps, but in that case the actual record LSNs are checked
193 : // which should be fine based on the filtering logic.
194 1 : if let Some(start) = current_batch_wal_start {
195 0 : *start = std::cmp::min(*start, new_shard_start_pos);
196 1 : }
197 1 : CurrentPositionUpdate::NotReset(*current_position)
198 : }
199 : }
200 : InterpretedWalReaderState::Done => {
201 0 : panic!("maybe_reset called on finished reader")
202 : }
203 : }
204 4 : }
205 :
206 50 : fn update_current_batch_wal_start(&mut self, lsn: Lsn) {
207 50 : match self {
208 : InterpretedWalReaderState::Running {
209 50 : current_batch_wal_start,
210 50 : ..
211 50 : } => {
212 50 : if current_batch_wal_start.is_none() {
213 6 : *current_batch_wal_start = Some(lsn);
214 44 : }
215 : }
216 : InterpretedWalReaderState::Done => {
217 0 : panic!("update_current_batch_wal_start called on finished reader")
218 : }
219 : }
220 50 : }
221 :
222 41 : fn replace_current_batch_wal_start(&mut self, with: Lsn) -> Lsn {
223 41 : match self {
224 : InterpretedWalReaderState::Running {
225 41 : current_batch_wal_start,
226 41 : ..
227 41 : } => current_batch_wal_start.replace(with).unwrap(),
228 : InterpretedWalReaderState::Done => {
229 0 : panic!("take_current_batch_wal_start called on finished reader")
230 : }
231 : }
232 41 : }
233 :
234 41 : fn update_current_position(&mut self, lsn: Lsn) {
235 41 : match self {
236 : InterpretedWalReaderState::Running {
237 41 : current_position, ..
238 41 : } => {
239 41 : *current_position = lsn;
240 41 : }
241 : InterpretedWalReaderState::Done => {
242 0 : panic!("update_current_position called on finished reader")
243 : }
244 : }
245 41 : }
246 : }
247 :
248 : pub(crate) struct AttachShardNotification {
249 : shard_id: ShardIdentity,
250 : sender: tokio::sync::mpsc::Sender<Batch>,
251 : start_pos: Lsn,
252 : }
253 :
254 : impl InterpretedWalReader {
255 : /// Spawn the reader in a separate tokio task and return a handle
256 3 : pub(crate) fn spawn(
257 3 : wal_stream: StreamingWalReader,
258 3 : start_pos: Lsn,
259 3 : tx: tokio::sync::mpsc::Sender<Batch>,
260 3 : shard: ShardIdentity,
261 3 : pg_version: u32,
262 3 : appname: &Option<String>,
263 3 : ) -> InterpretedWalReaderHandle {
264 3 : let state = Arc::new(std::sync::RwLock::new(InterpretedWalReaderState::Running {
265 3 : current_position: start_pos,
266 3 : current_batch_wal_start: None,
267 3 : }));
268 3 :
269 3 : let (shard_notification_tx, shard_notification_rx) = tokio::sync::mpsc::unbounded_channel();
270 :
271 3 : let reader = InterpretedWalReader {
272 3 : wal_stream,
273 3 : shard_senders: HashMap::from([(
274 3 : shard,
275 3 : smallvec::smallvec![ShardSenderState {
276 0 : sender_id: SenderId::first(),
277 0 : tx,
278 0 : next_record_lsn: start_pos,
279 0 : }],
280 : )]),
281 3 : shard_notification_rx: Some(shard_notification_rx),
282 3 : state: state.clone(),
283 3 : pg_version,
284 3 : };
285 3 :
286 3 : let metric = WAL_READERS
287 3 : .get_metric_with_label_values(&["task", appname.as_deref().unwrap_or("safekeeper")])
288 3 : .unwrap();
289 :
290 3 : let join_handle = tokio::task::spawn(
291 3 : async move {
292 3 : metric.inc();
293 3 : scopeguard::defer! {
294 3 : metric.dec();
295 3 : }
296 3 :
297 3 : reader
298 3 : .run_impl(start_pos)
299 3 : .await
300 0 : .inspect_err(|err| match err {
301 : // TODO: we may want to differentiate these errors further.
302 : InterpretedWalReaderError::Decode(_) => {
303 0 : critical!("failed to decode WAL record: {err:?}");
304 : }
305 0 : err => error!("failed to read WAL record: {err}"),
306 0 : })
307 0 : }
308 3 : .instrument(info_span!("interpreted wal reader")),
309 : );
310 :
311 3 : InterpretedWalReaderHandle {
312 3 : join_handle,
313 3 : state,
314 3 : shard_notification_tx,
315 3 : }
316 3 : }
317 :
318 : /// Construct the reader without spawning anything
319 : /// Callers should drive the future returned by [`Self::run`].
320 1 : pub(crate) fn new(
321 1 : wal_stream: StreamingWalReader,
322 1 : start_pos: Lsn,
323 1 : tx: tokio::sync::mpsc::Sender<Batch>,
324 1 : shard: ShardIdentity,
325 1 : pg_version: u32,
326 1 : shard_notification_rx: Option<
327 1 : tokio::sync::mpsc::UnboundedReceiver<AttachShardNotification>,
328 1 : >,
329 1 : ) -> InterpretedWalReader {
330 1 : let state = Arc::new(std::sync::RwLock::new(InterpretedWalReaderState::Running {
331 1 : current_position: start_pos,
332 1 : current_batch_wal_start: None,
333 1 : }));
334 1 :
335 1 : InterpretedWalReader {
336 1 : wal_stream,
337 1 : shard_senders: HashMap::from([(
338 1 : shard,
339 1 : smallvec::smallvec![ShardSenderState {
340 0 : sender_id: SenderId::first(),
341 0 : tx,
342 0 : next_record_lsn: start_pos,
343 0 : }],
344 : )]),
345 1 : shard_notification_rx,
346 1 : state: state.clone(),
347 1 : pg_version,
348 1 : }
349 1 : }
350 :
351 : /// Entry point for future (polling) based wal reader.
352 1 : pub(crate) async fn run(
353 1 : self,
354 1 : start_pos: Lsn,
355 1 : appname: &Option<String>,
356 1 : ) -> Result<(), CopyStreamHandlerEnd> {
357 1 : let metric = WAL_READERS
358 1 : .get_metric_with_label_values(&["future", appname.as_deref().unwrap_or("safekeeper")])
359 1 : .unwrap();
360 1 :
361 1 : metric.inc();
362 1 : scopeguard::defer! {
363 1 : metric.dec();
364 1 : }
365 1 :
366 1 : match self.run_impl(start_pos).await {
367 0 : Err(err @ InterpretedWalReaderError::Decode(_)) => {
368 0 : critical!("failed to decode WAL record: {err:?}");
369 : }
370 0 : Err(err) => error!("failed to read WAL record: {err}"),
371 0 : Ok(()) => info!("interpreted wal reader exiting"),
372 : }
373 :
374 0 : Err(CopyStreamHandlerEnd::Other(anyhow!(
375 0 : "interpreted wal reader finished"
376 0 : )))
377 0 : }
378 :
379 : /// Send interpreted WAL to one or more [`InterpretedWalSender`]s
380 : /// Stops when an error is encountered or when the [`InterpretedWalReaderHandle`]
381 : /// goes out of scope.
382 4 : async fn run_impl(mut self, start_pos: Lsn) -> Result<(), InterpretedWalReaderError> {
383 4 : let defer_state = self.state.clone();
384 4 : scopeguard::defer! {
385 4 : *defer_state.write().unwrap() = InterpretedWalReaderState::Done;
386 4 : }
387 4 :
388 4 : let mut wal_decoder = WalStreamDecoder::new(start_pos, self.pg_version);
389 :
390 : loop {
391 58 : tokio::select! {
392 : // Main branch for reading WAL and forwarding it
393 58 : wal_or_reset = self.wal_stream.next() => {
394 50 : let wal = wal_or_reset.map(|wor| wor.get_wal().expect("reset handled in select branch below"));
395 : let WalBytes {
396 50 : wal,
397 50 : wal_start_lsn,
398 50 : wal_end_lsn,
399 50 : available_wal_end_lsn,
400 50 : } = match wal {
401 50 : Some(some) => some.map_err(InterpretedWalReaderError::ReadOrInterpret)?,
402 : None => {
403 : // [`StreamingWalReader::next`] is an endless stream of WAL.
404 : // It shouldn't ever finish unless it panicked or became internally
405 : // inconsistent.
406 0 : return Result::Err(InterpretedWalReaderError::WalStreamClosed);
407 : }
408 : };
409 :
410 50 : self.state.write().unwrap().update_current_batch_wal_start(wal_start_lsn);
411 50 :
412 50 : wal_decoder.feed_bytes(&wal);
413 50 :
414 50 : // Deserialize and interpret WAL records from this batch of WAL.
415 50 : // Interpreted records for each shard are collected separately.
416 50 : let shard_ids = self.shard_senders.keys().copied().collect::<Vec<_>>();
417 50 : let mut records_by_sender: HashMap<ShardSenderId, Vec<InterpretedWalRecord>> = HashMap::new();
418 50 : let mut max_next_record_lsn = None;
419 50 : let mut max_end_record_lsn = None;
420 656 : while let Some((next_record_lsn, recdata)) = wal_decoder.poll_decode()?
421 : {
422 606 : assert!(next_record_lsn.is_aligned());
423 606 : max_next_record_lsn = Some(next_record_lsn);
424 606 : max_end_record_lsn = Some(wal_decoder.lsn());
425 :
426 606 : let interpreted = InterpretedWalRecord::from_bytes_filtered(
427 606 : recdata,
428 606 : &shard_ids,
429 606 : next_record_lsn,
430 606 : self.pg_version,
431 606 : )
432 606 : .with_context(|| "Failed to interpret WAL")?;
433 :
434 1412 : for (shard, record) in interpreted {
435 : // Shard zero needs to track the start LSN of the latest record
436 : // in adition to the LSN of the next record to ingest. The former
437 : // is included in basebackup persisted by the compute in WAL.
438 806 : if !shard.is_shard_zero() && record.is_empty() {
439 200 : continue;
440 606 : }
441 606 :
442 606 : let mut states_iter = self.shard_senders
443 606 : .get(&shard)
444 606 : .expect("keys collected above")
445 606 : .iter()
446 1002 : .filter(|state| record.next_record_lsn > state.next_record_lsn)
447 606 : .peekable();
448 996 : while let Some(state) = states_iter.next() {
449 796 : let shard_sender_id = ShardSenderId::new(shard, state.sender_id);
450 796 :
451 796 : // The most commont case is one sender per shard. Peek and break to avoid the
452 796 : // clone in that situation.
453 796 : if states_iter.peek().is_none() {
454 406 : records_by_sender.entry(shard_sender_id).or_default().push(record);
455 406 : break;
456 390 : } else {
457 390 : records_by_sender.entry(shard_sender_id).or_default().push(record.clone());
458 390 : }
459 : }
460 : }
461 : }
462 :
463 50 : let max_next_record_lsn = match max_next_record_lsn {
464 41 : Some(lsn) => lsn,
465 : None => {
466 9 : continue;
467 : }
468 : };
469 :
470 : // Update the current position such that new receivers can decide
471 : // whether to attach to us or spawn a new WAL reader.
472 41 : let batch_wal_start_lsn = {
473 41 : let mut guard = self.state.write().unwrap();
474 41 : guard.update_current_position(max_next_record_lsn);
475 41 : guard.replace_current_batch_wal_start(max_end_record_lsn.unwrap())
476 41 : };
477 41 :
478 41 : // Send interpreted records downstream. Anything that has already been seen
479 41 : // by a shard is filtered out.
480 41 : let mut shard_senders_to_remove = Vec::new();
481 96 : for (shard, states) in &mut self.shard_senders {
482 136 : for state in states {
483 81 : let shard_sender_id = ShardSenderId::new(*shard, state.sender_id);
484 :
485 81 : let batch = if max_next_record_lsn > state.next_record_lsn {
486 : // This batch contains at least one record that this shard has not
487 : // seen yet.
488 67 : let records = records_by_sender.remove(&shard_sender_id).unwrap_or_default();
489 67 :
490 67 : InterpretedWalRecords {
491 67 : records,
492 67 : next_record_lsn: max_next_record_lsn,
493 67 : raw_wal_start_lsn: Some(batch_wal_start_lsn),
494 67 : }
495 14 : } else if wal_end_lsn > state.next_record_lsn {
496 : // All the records in this batch were seen by the shard
497 : // However, the batch maps to a chunk of WAL that the
498 : // shard has not yet seen. Notify it of the start LSN
499 : // of the PG WAL chunk such that it doesn't look like a gap.
500 0 : InterpretedWalRecords {
501 0 : records: Vec::default(),
502 0 : next_record_lsn: state.next_record_lsn,
503 0 : raw_wal_start_lsn: Some(batch_wal_start_lsn),
504 0 : }
505 : } else {
506 : // The shard has seen this chunk of WAL before. Skip it.
507 14 : continue;
508 : };
509 :
510 67 : let res = state.tx.send(Batch {
511 67 : wal_end_lsn,
512 67 : available_wal_end_lsn,
513 67 : records: batch,
514 67 : }).await;
515 :
516 67 : if res.is_err() {
517 0 : shard_senders_to_remove.push(shard_sender_id);
518 67 : } else {
519 67 : state.next_record_lsn = std::cmp::max(state.next_record_lsn, max_next_record_lsn);
520 67 : }
521 : }
522 : }
523 :
524 : // Clean up any shard senders that have dropped out.
525 : // This is inefficient, but such events are rare (connection to PS termination)
526 : // and the number of subscriptions on the same shards very small (only one
527 : // for the steady state).
528 41 : for to_remove in shard_senders_to_remove {
529 0 : let shard_senders = self.shard_senders.get_mut(&to_remove.shard()).expect("saw it above");
530 0 : if let Some(idx) = shard_senders.iter().position(|s| s.sender_id == to_remove.sender_id) {
531 0 : shard_senders.remove(idx);
532 0 : tracing::info!("Removed shard sender {}", to_remove);
533 0 : }
534 :
535 0 : if shard_senders.is_empty() {
536 0 : self.shard_senders.remove(&to_remove.shard());
537 0 : }
538 : }
539 : },
540 : // Listen for new shards that want to attach to this reader.
541 : // If the reader is not running as a task, then this is not supported
542 : // (see the pending branch below).
543 58 : notification = match self.shard_notification_rx.as_mut() {
544 58 : Some(rx) => Either::Left(rx.recv()),
545 0 : None => Either::Right(std::future::pending())
546 : } => {
547 4 : if let Some(n) = notification {
548 4 : let AttachShardNotification { shard_id, sender, start_pos } = n;
549 4 :
550 4 : // Update internal and external state, then reset the WAL stream
551 4 : // if required.
552 4 : let senders = self.shard_senders.entry(shard_id).or_default();
553 4 : let new_sender_id = match senders.last() {
554 2 : Some(sender) => sender.sender_id.next(),
555 2 : None => SenderId::first()
556 : };
557 :
558 4 : senders.push(ShardSenderState { sender_id: new_sender_id, tx: sender, next_record_lsn: start_pos});
559 4 :
560 4 : // If the shard is subscribing below the current position the we need
561 4 : // to update the cursor that tracks where we are at in the WAL
562 4 : // ([`Self::state`]) and reset the WAL stream itself
563 4 : // (`[Self::wal_stream`]). This must be done atomically from the POV of
564 4 : // anything outside the select statement.
565 4 : let position_reset = self.state.write().unwrap().maybe_reset(start_pos);
566 4 : match position_reset {
567 3 : CurrentPositionUpdate::Reset { from: _, to } => {
568 3 : self.wal_stream.reset(to).await;
569 3 : wal_decoder = WalStreamDecoder::new(to, self.pg_version);
570 : },
571 1 : CurrentPositionUpdate::NotReset(_) => {}
572 : };
573 :
574 4 : tracing::info!(
575 0 : "Added shard sender {} with start_pos={} previous_pos={} current_pos={}",
576 0 : ShardSenderId::new(shard_id, new_sender_id),
577 0 : start_pos,
578 0 : position_reset.previous_position(),
579 0 : position_reset.current_position(),
580 : );
581 0 : }
582 : }
583 : }
584 : }
585 0 : }
586 :
587 : #[cfg(test)]
588 1 : fn state(&self) -> Arc<std::sync::RwLock<InterpretedWalReaderState>> {
589 1 : self.state.clone()
590 1 : }
591 : }
592 :
593 : impl InterpretedWalReaderHandle {
594 : /// Fan-out the reader by attaching a new shard to it
595 3 : pub(crate) fn fanout(
596 3 : &self,
597 3 : shard_id: ShardIdentity,
598 3 : sender: tokio::sync::mpsc::Sender<Batch>,
599 3 : start_pos: Lsn,
600 3 : ) -> Result<(), SendError<AttachShardNotification>> {
601 3 : self.shard_notification_tx.send(AttachShardNotification {
602 3 : shard_id,
603 3 : sender,
604 3 : start_pos,
605 3 : })
606 3 : }
607 :
608 : /// Get the current WAL position of the reader
609 6 : pub(crate) fn current_position(&self) -> Option<Lsn> {
610 6 : self.state.read().unwrap().current_position()
611 6 : }
612 :
613 6 : pub(crate) fn abort(&self) {
614 6 : self.join_handle.abort()
615 6 : }
616 : }
617 :
618 : impl Drop for InterpretedWalReaderHandle {
619 3 : fn drop(&mut self) {
620 3 : tracing::info!("Aborting interpreted wal reader");
621 3 : self.abort()
622 3 : }
623 : }
624 :
625 : pub(crate) struct InterpretedWalSender<'a, IO> {
626 : pub(crate) format: InterpretedFormat,
627 : pub(crate) compression: Option<Compression>,
628 : pub(crate) appname: Option<String>,
629 :
630 : pub(crate) tli: WalResidentTimeline,
631 : pub(crate) start_lsn: Lsn,
632 :
633 : pub(crate) pgb: &'a mut PostgresBackend<IO>,
634 : pub(crate) end_watch_view: EndWatchView,
635 : pub(crate) wal_sender_guard: Arc<WalSenderGuard>,
636 : pub(crate) rx: tokio::sync::mpsc::Receiver<Batch>,
637 : }
638 :
639 : impl<IO: AsyncRead + AsyncWrite + Unpin> InterpretedWalSender<'_, IO> {
640 : /// Send interpreted WAL records over the network.
641 : /// Also manages keep-alives if nothing was sent for a while.
642 0 : pub(crate) async fn run(mut self) -> Result<(), CopyStreamHandlerEnd> {
643 0 : let mut keepalive_ticker = tokio::time::interval(Duration::from_secs(1));
644 0 : keepalive_ticker.set_missed_tick_behavior(MissedTickBehavior::Skip);
645 0 : keepalive_ticker.reset();
646 0 :
647 0 : let mut wal_position = self.start_lsn;
648 :
649 : loop {
650 0 : tokio::select! {
651 0 : batch = self.rx.recv() => {
652 0 : let batch = match batch {
653 0 : Some(b) => b,
654 : None => {
655 0 : return Result::Err(
656 0 : CopyStreamHandlerEnd::Other(anyhow!("Interpreted WAL reader exited early"))
657 0 : );
658 : }
659 : };
660 :
661 0 : wal_position = batch.wal_end_lsn;
662 :
663 0 : let buf = batch
664 0 : .records
665 0 : .to_wire(self.format, self.compression)
666 0 : .await
667 0 : .with_context(|| "Failed to serialize interpreted WAL")
668 0 : .map_err(CopyStreamHandlerEnd::from)?;
669 :
670 : // Reset the keep alive ticker since we are sending something
671 : // over the wire now.
672 0 : keepalive_ticker.reset();
673 0 :
674 0 : self.pgb
675 0 : .write_message(&BeMessage::InterpretedWalRecords(InterpretedWalRecordsBody {
676 0 : streaming_lsn: batch.wal_end_lsn.0,
677 0 : commit_lsn: batch.available_wal_end_lsn.0,
678 0 : data: &buf,
679 0 : })).await?;
680 : }
681 : // Send a periodic keep alive when the connection has been idle for a while.
682 : // Since we've been idle, also check if we can stop streaming.
683 0 : _ = keepalive_ticker.tick() => {
684 0 : if let Some(remote_consistent_lsn) = self.wal_sender_guard
685 0 : .walsenders()
686 0 : .get_ws_remote_consistent_lsn(self.wal_sender_guard.id())
687 : {
688 0 : if self.tli.should_walsender_stop(remote_consistent_lsn).await {
689 : // Stop streaming if the receivers are caught up and
690 : // there's no active compute. This causes the loop in
691 : // [`crate::send_interpreted_wal::InterpretedWalSender::run`]
692 : // to exit and terminate the WAL stream.
693 0 : break;
694 0 : }
695 0 : }
696 :
697 0 : self.pgb
698 0 : .write_message(&BeMessage::KeepAlive(WalSndKeepAlive {
699 0 : wal_end: self.end_watch_view.get().0,
700 0 : timestamp: get_current_timestamp(),
701 0 : request_reply: true,
702 0 : }))
703 0 : .await?;
704 : },
705 : }
706 : }
707 :
708 0 : Err(CopyStreamHandlerEnd::ServerInitiated(format!(
709 0 : "ending streaming to {:?} at {}, receiver is caughtup and there is no computes",
710 0 : self.appname, wal_position,
711 0 : )))
712 0 : }
713 : }
714 : #[cfg(test)]
715 : mod tests {
716 : use std::collections::HashMap;
717 : use std::str::FromStr;
718 : use std::time::Duration;
719 :
720 : use pageserver_api::shard::{ShardIdentity, ShardStripeSize};
721 : use postgres_ffi::MAX_SEND_SIZE;
722 : use tokio::sync::mpsc::error::TryRecvError;
723 : use utils::id::{NodeId, TenantTimelineId};
724 : use utils::lsn::Lsn;
725 : use utils::shard::{ShardCount, ShardNumber};
726 :
727 : use crate::send_interpreted_wal::{AttachShardNotification, Batch, InterpretedWalReader};
728 : use crate::test_utils::Env;
729 : use crate::wal_reader_stream::StreamingWalReader;
730 :
731 : #[tokio::test]
732 1 : async fn test_interpreted_wal_reader_fanout() {
733 1 : let _ = env_logger::builder().is_test(true).try_init();
734 1 :
735 1 : const SIZE: usize = 8 * 1024;
736 1 : const MSG_COUNT: usize = 200;
737 1 : const PG_VERSION: u32 = 17;
738 1 : const SHARD_COUNT: u8 = 2;
739 1 :
740 1 : let start_lsn = Lsn::from_str("0/149FD18").unwrap();
741 1 : let env = Env::new(true).unwrap();
742 1 : let tli = env
743 1 : .make_timeline(NodeId(1), TenantTimelineId::generate(), start_lsn)
744 1 : .await
745 1 : .unwrap();
746 1 :
747 1 : let resident_tli = tli.wal_residence_guard().await.unwrap();
748 1 : let end_watch = Env::write_wal(tli, start_lsn, SIZE, MSG_COUNT, c"neon-file:", None)
749 1 : .await
750 1 : .unwrap();
751 1 : let end_pos = end_watch.get();
752 1 :
753 1 : tracing::info!("Doing first round of reads ...");
754 1 :
755 1 : let streaming_wal_reader = StreamingWalReader::new(
756 1 : resident_tli,
757 1 : None,
758 1 : start_lsn,
759 1 : end_pos,
760 1 : end_watch,
761 1 : MAX_SEND_SIZE,
762 1 : );
763 1 :
764 1 : let shard_0 = ShardIdentity::new(
765 1 : ShardNumber(0),
766 1 : ShardCount(SHARD_COUNT),
767 1 : ShardStripeSize::default(),
768 1 : )
769 1 : .unwrap();
770 1 :
771 1 : let shard_1 = ShardIdentity::new(
772 1 : ShardNumber(1),
773 1 : ShardCount(SHARD_COUNT),
774 1 : ShardStripeSize::default(),
775 1 : )
776 1 : .unwrap();
777 1 :
778 1 : let mut shards = HashMap::new();
779 1 :
780 3 : for shard_number in 0..SHARD_COUNT {
781 2 : let shard_id = ShardIdentity::new(
782 2 : ShardNumber(shard_number),
783 2 : ShardCount(SHARD_COUNT),
784 2 : ShardStripeSize::default(),
785 2 : )
786 2 : .unwrap();
787 2 : let (tx, rx) = tokio::sync::mpsc::channel::<Batch>(MSG_COUNT * 2);
788 2 : shards.insert(shard_id, (Some(tx), Some(rx)));
789 2 : }
790 1 :
791 1 : let shard_0_tx = shards.get_mut(&shard_0).unwrap().0.take().unwrap();
792 1 : let mut shard_0_rx = shards.get_mut(&shard_0).unwrap().1.take().unwrap();
793 1 :
794 1 : let handle = InterpretedWalReader::spawn(
795 1 : streaming_wal_reader,
796 1 : start_lsn,
797 1 : shard_0_tx,
798 1 : shard_0,
799 1 : PG_VERSION,
800 1 : &Some("pageserver".to_string()),
801 1 : );
802 1 :
803 1 : tracing::info!("Reading all WAL with only shard 0 attached ...");
804 1 :
805 1 : let mut shard_0_interpreted_records = Vec::new();
806 13 : while let Some(batch) = shard_0_rx.recv().await {
807 13 : shard_0_interpreted_records.push(batch.records);
808 13 : if batch.wal_end_lsn == batch.available_wal_end_lsn {
809 1 : break;
810 12 : }
811 1 : }
812 1 :
813 1 : let shard_1_tx = shards.get_mut(&shard_1).unwrap().0.take().unwrap();
814 1 : let mut shard_1_rx = shards.get_mut(&shard_1).unwrap().1.take().unwrap();
815 1 :
816 1 : tracing::info!("Attaching shard 1 to the reader at start of WAL");
817 1 : handle.fanout(shard_1, shard_1_tx, start_lsn).unwrap();
818 1 :
819 1 : tracing::info!("Reading all WAL with shard 0 and shard 1 attached ...");
820 1 :
821 1 : let mut shard_1_interpreted_records = Vec::new();
822 13 : while let Some(batch) = shard_1_rx.recv().await {
823 13 : shard_1_interpreted_records.push(batch.records);
824 13 : if batch.wal_end_lsn == batch.available_wal_end_lsn {
825 1 : break;
826 12 : }
827 1 : }
828 1 :
829 1 : // This test uses logical messages. Those only go to shard 0. Check that the
830 1 : // filtering worked and shard 1 did not get any.
831 1 : assert!(
832 1 : shard_1_interpreted_records
833 1 : .iter()
834 13 : .all(|recs| recs.records.is_empty())
835 1 : );
836 1 :
837 1 : // Shard 0 should not receive anything more since the reader is
838 1 : // going through wal that it has already processed.
839 1 : let res = shard_0_rx.try_recv();
840 1 : if let Ok(ref ok) = res {
841 1 : tracing::error!(
842 1 : "Shard 0 received batch: wal_end_lsn={} available_wal_end_lsn={}",
843 1 : ok.wal_end_lsn,
844 1 : ok.available_wal_end_lsn
845 1 : );
846 1 : }
847 1 : assert!(matches!(res, Err(TryRecvError::Empty)));
848 1 :
849 1 : // Check that the next records lsns received by the two shards match up.
850 1 : let shard_0_next_lsns = shard_0_interpreted_records
851 1 : .iter()
852 13 : .map(|recs| recs.next_record_lsn)
853 1 : .collect::<Vec<_>>();
854 1 : let shard_1_next_lsns = shard_1_interpreted_records
855 1 : .iter()
856 13 : .map(|recs| recs.next_record_lsn)
857 1 : .collect::<Vec<_>>();
858 1 : assert_eq!(shard_0_next_lsns, shard_1_next_lsns);
859 1 :
860 1 : handle.abort();
861 1 : let mut done = false;
862 2 : for _ in 0..5 {
863 2 : if handle.current_position().is_none() {
864 1 : done = true;
865 1 : break;
866 1 : }
867 1 : tokio::time::sleep(Duration::from_millis(1)).await;
868 1 : }
869 1 :
870 1 : assert!(done);
871 1 : }
872 :
873 : #[tokio::test]
874 1 : async fn test_interpreted_wal_reader_same_shard_fanout() {
875 1 : let _ = env_logger::builder().is_test(true).try_init();
876 1 :
877 1 : const SIZE: usize = 8 * 1024;
878 1 : const MSG_COUNT: usize = 200;
879 1 : const PG_VERSION: u32 = 17;
880 1 : const SHARD_COUNT: u8 = 2;
881 1 :
882 1 : let start_lsn = Lsn::from_str("0/149FD18").unwrap();
883 1 : let env = Env::new(true).unwrap();
884 1 : let tli = env
885 1 : .make_timeline(NodeId(1), TenantTimelineId::generate(), start_lsn)
886 1 : .await
887 1 : .unwrap();
888 1 :
889 1 : let resident_tli = tli.wal_residence_guard().await.unwrap();
890 1 : let mut next_record_lsns = Vec::default();
891 1 : let end_watch = Env::write_wal(
892 1 : tli,
893 1 : start_lsn,
894 1 : SIZE,
895 1 : MSG_COUNT,
896 1 : c"neon-file:",
897 1 : Some(&mut next_record_lsns),
898 1 : )
899 1 : .await
900 1 : .unwrap();
901 1 : let end_pos = end_watch.get();
902 1 :
903 1 : let streaming_wal_reader = StreamingWalReader::new(
904 1 : resident_tli,
905 1 : None,
906 1 : start_lsn,
907 1 : end_pos,
908 1 : end_watch,
909 1 : MAX_SEND_SIZE,
910 1 : );
911 1 :
912 1 : let shard_0 = ShardIdentity::new(
913 1 : ShardNumber(0),
914 1 : ShardCount(SHARD_COUNT),
915 1 : ShardStripeSize::default(),
916 1 : )
917 1 : .unwrap();
918 1 :
919 1 : struct Sender {
920 1 : tx: Option<tokio::sync::mpsc::Sender<Batch>>,
921 1 : rx: tokio::sync::mpsc::Receiver<Batch>,
922 1 : shard: ShardIdentity,
923 1 : start_lsn: Lsn,
924 1 : received_next_record_lsns: Vec<Lsn>,
925 1 : }
926 1 :
927 1 : impl Sender {
928 3 : fn new(start_lsn: Lsn, shard: ShardIdentity) -> Self {
929 3 : let (tx, rx) = tokio::sync::mpsc::channel::<Batch>(MSG_COUNT * 2);
930 3 : Self {
931 3 : tx: Some(tx),
932 3 : rx,
933 3 : shard,
934 3 : start_lsn,
935 3 : received_next_record_lsns: Vec::default(),
936 3 : }
937 3 : }
938 1 : }
939 1 :
940 1 : assert!(next_record_lsns.len() > 7);
941 1 : let start_lsns = vec![
942 1 : next_record_lsns[5],
943 1 : next_record_lsns[1],
944 1 : next_record_lsns[3],
945 1 : ];
946 1 : let mut senders = start_lsns
947 1 : .into_iter()
948 3 : .map(|lsn| Sender::new(lsn, shard_0))
949 1 : .collect::<Vec<_>>();
950 1 :
951 1 : let first_sender = senders.first_mut().unwrap();
952 1 : let handle = InterpretedWalReader::spawn(
953 1 : streaming_wal_reader,
954 1 : first_sender.start_lsn,
955 1 : first_sender.tx.take().unwrap(),
956 1 : first_sender.shard,
957 1 : PG_VERSION,
958 1 : &Some("pageserver".to_string()),
959 1 : );
960 1 :
961 2 : for sender in senders.iter_mut().skip(1) {
962 2 : handle
963 2 : .fanout(sender.shard, sender.tx.take().unwrap(), sender.start_lsn)
964 2 : .unwrap();
965 2 : }
966 1 :
967 3 : for sender in senders.iter_mut() {
968 1 : loop {
969 39 : let batch = sender.rx.recv().await.unwrap();
970 39 : tracing::info!(
971 1 : "Sender with start_lsn={} received batch ending at {} with {} records",
972 0 : sender.start_lsn,
973 0 : batch.wal_end_lsn,
974 0 : batch.records.records.len()
975 1 : );
976 1 :
977 627 : for rec in batch.records.records {
978 588 : sender.received_next_record_lsns.push(rec.next_record_lsn);
979 588 : }
980 1 :
981 39 : if batch.wal_end_lsn == batch.available_wal_end_lsn {
982 3 : break;
983 36 : }
984 1 : }
985 1 : }
986 1 :
987 1 : handle.abort();
988 1 : let mut done = false;
989 2 : for _ in 0..5 {
990 2 : if handle.current_position().is_none() {
991 1 : done = true;
992 1 : break;
993 1 : }
994 1 : tokio::time::sleep(Duration::from_millis(1)).await;
995 1 : }
996 1 :
997 1 : assert!(done);
998 1 :
999 4 : for sender in senders {
1000 3 : tracing::info!(
1001 1 : "Validating records received by sender with start_lsn={}",
1002 1 : sender.start_lsn
1003 1 : );
1004 1 :
1005 3 : assert!(sender.received_next_record_lsns.is_sorted());
1006 3 : let expected = next_record_lsns
1007 3 : .iter()
1008 600 : .filter(|lsn| **lsn > sender.start_lsn)
1009 3 : .copied()
1010 3 : .collect::<Vec<_>>();
1011 3 : assert_eq!(sender.received_next_record_lsns, expected);
1012 1 : }
1013 1 : }
1014 :
1015 : #[tokio::test]
1016 1 : async fn test_batch_start_tracking_on_reset() {
1017 1 : // When the WAL stream is reset to an older LSN,
1018 1 : // the current batch start LSN should be invalidated.
1019 1 : // This test constructs such a scenario:
1020 1 : // 1. Shard 0 is reading somewhere ahead
1021 1 : // 2. Reader reads some WAL, but does not decode a full record (partial read)
1022 1 : // 3. Shard 1 attaches to the reader and resets it to an older LSN
1023 1 : // 4. Shard 1 should get the correct batch WAL start LSN
1024 1 : let _ = env_logger::builder().is_test(true).try_init();
1025 1 :
1026 1 : const SIZE: usize = 64 * 1024;
1027 1 : const MSG_COUNT: usize = 10;
1028 1 : const PG_VERSION: u32 = 17;
1029 1 : const SHARD_COUNT: u8 = 2;
1030 1 : const WAL_READER_BATCH_SIZE: usize = 8192;
1031 1 :
1032 1 : let start_lsn = Lsn::from_str("0/149FD18").unwrap();
1033 1 : let env = Env::new(true).unwrap();
1034 1 : let mut next_record_lsns = Vec::default();
1035 1 : let tli = env
1036 1 : .make_timeline(NodeId(1), TenantTimelineId::generate(), start_lsn)
1037 1 : .await
1038 1 : .unwrap();
1039 1 :
1040 1 : let resident_tli = tli.wal_residence_guard().await.unwrap();
1041 1 : let end_watch = Env::write_wal(
1042 1 : tli,
1043 1 : start_lsn,
1044 1 : SIZE,
1045 1 : MSG_COUNT,
1046 1 : c"neon-file:",
1047 1 : Some(&mut next_record_lsns),
1048 1 : )
1049 1 : .await
1050 1 : .unwrap();
1051 1 :
1052 1 : assert!(next_record_lsns.len() > 3);
1053 1 : let shard_0_start_lsn = next_record_lsns[3];
1054 1 :
1055 1 : let end_pos = end_watch.get();
1056 1 :
1057 1 : let streaming_wal_reader = StreamingWalReader::new(
1058 1 : resident_tli,
1059 1 : None,
1060 1 : shard_0_start_lsn,
1061 1 : end_pos,
1062 1 : end_watch,
1063 1 : WAL_READER_BATCH_SIZE,
1064 1 : );
1065 1 :
1066 1 : let shard_0 = ShardIdentity::new(
1067 1 : ShardNumber(0),
1068 1 : ShardCount(SHARD_COUNT),
1069 1 : ShardStripeSize::default(),
1070 1 : )
1071 1 : .unwrap();
1072 1 :
1073 1 : let shard_1 = ShardIdentity::new(
1074 1 : ShardNumber(1),
1075 1 : ShardCount(SHARD_COUNT),
1076 1 : ShardStripeSize::default(),
1077 1 : )
1078 1 : .unwrap();
1079 1 :
1080 1 : let mut shards = HashMap::new();
1081 1 :
1082 3 : for shard_number in 0..SHARD_COUNT {
1083 2 : let shard_id = ShardIdentity::new(
1084 2 : ShardNumber(shard_number),
1085 2 : ShardCount(SHARD_COUNT),
1086 2 : ShardStripeSize::default(),
1087 2 : )
1088 2 : .unwrap();
1089 2 : let (tx, rx) = tokio::sync::mpsc::channel::<Batch>(MSG_COUNT * 2);
1090 2 : shards.insert(shard_id, (Some(tx), Some(rx)));
1091 2 : }
1092 1 :
1093 1 : let shard_0_tx = shards.get_mut(&shard_0).unwrap().0.take().unwrap();
1094 1 :
1095 1 : let (shard_notification_tx, shard_notification_rx) = tokio::sync::mpsc::unbounded_channel();
1096 1 :
1097 1 : let reader = InterpretedWalReader::new(
1098 1 : streaming_wal_reader,
1099 1 : shard_0_start_lsn,
1100 1 : shard_0_tx,
1101 1 : shard_0,
1102 1 : PG_VERSION,
1103 1 : Some(shard_notification_rx),
1104 1 : );
1105 1 :
1106 1 : let reader_state = reader.state();
1107 1 : let mut reader_fut = std::pin::pin!(reader.run(shard_0_start_lsn, &None));
1108 1 : loop {
1109 57 : let poll = futures::poll!(reader_fut.as_mut());
1110 57 : assert!(poll.is_pending());
1111 1 :
1112 57 : let guard = reader_state.read().unwrap();
1113 57 : if guard.current_batch_wal_start().is_some() {
1114 1 : break;
1115 56 : }
1116 1 : }
1117 1 :
1118 1 : shard_notification_tx
1119 1 : .send(AttachShardNotification {
1120 1 : shard_id: shard_1,
1121 1 : sender: shards.get_mut(&shard_1).unwrap().0.take().unwrap(),
1122 1 : start_pos: start_lsn,
1123 1 : })
1124 1 : .unwrap();
1125 1 :
1126 1 : let mut shard_1_rx = shards.get_mut(&shard_1).unwrap().1.take().unwrap();
1127 1 : loop {
1128 64 : let poll = futures::poll!(reader_fut.as_mut());
1129 64 : assert!(poll.is_pending());
1130 1 :
1131 64 : let try_recv_res = shard_1_rx.try_recv();
1132 63 : match try_recv_res {
1133 1 : Ok(batch) => {
1134 1 : assert_eq!(batch.records.raw_wal_start_lsn.unwrap(), start_lsn);
1135 1 : break;
1136 1 : }
1137 63 : Err(tokio::sync::mpsc::error::TryRecvError::Empty) => {}
1138 1 : Err(tokio::sync::mpsc::error::TryRecvError::Disconnected) => {
1139 1 : unreachable!();
1140 1 : }
1141 1 : }
1142 1 : }
1143 1 : }
1144 :
1145 : #[tokio::test]
1146 1 : async fn test_shard_zero_does_not_skip_empty_records() {
1147 1 : let _ = env_logger::builder().is_test(true).try_init();
1148 1 :
1149 1 : const SIZE: usize = 8 * 1024;
1150 1 : const MSG_COUNT: usize = 10;
1151 1 : const PG_VERSION: u32 = 17;
1152 1 :
1153 1 : let start_lsn = Lsn::from_str("0/149FD18").unwrap();
1154 1 : let env = Env::new(true).unwrap();
1155 1 : let tli = env
1156 1 : .make_timeline(NodeId(1), TenantTimelineId::generate(), start_lsn)
1157 1 : .await
1158 1 : .unwrap();
1159 1 :
1160 1 : let resident_tli = tli.wal_residence_guard().await.unwrap();
1161 1 : let mut next_record_lsns = Vec::new();
1162 1 : let end_watch = Env::write_wal(
1163 1 : tli,
1164 1 : start_lsn,
1165 1 : SIZE,
1166 1 : MSG_COUNT,
1167 1 : // This is a logical message prefix that is not persisted to key value storage.
1168 1 : // We use it in order to validate that shard zero receives emtpy interpreted records.
1169 1 : c"test:",
1170 1 : Some(&mut next_record_lsns),
1171 1 : )
1172 1 : .await
1173 1 : .unwrap();
1174 1 : let end_pos = end_watch.get();
1175 1 :
1176 1 : let streaming_wal_reader = StreamingWalReader::new(
1177 1 : resident_tli,
1178 1 : None,
1179 1 : start_lsn,
1180 1 : end_pos,
1181 1 : end_watch,
1182 1 : MAX_SEND_SIZE,
1183 1 : );
1184 1 :
1185 1 : let shard = ShardIdentity::unsharded();
1186 1 : let (records_tx, mut records_rx) = tokio::sync::mpsc::channel::<Batch>(MSG_COUNT * 2);
1187 1 :
1188 1 : let handle = InterpretedWalReader::spawn(
1189 1 : streaming_wal_reader,
1190 1 : start_lsn,
1191 1 : records_tx,
1192 1 : shard,
1193 1 : PG_VERSION,
1194 1 : &Some("pageserver".to_string()),
1195 1 : );
1196 1 :
1197 1 : let mut interpreted_records = Vec::new();
1198 1 : while let Some(batch) = records_rx.recv().await {
1199 1 : interpreted_records.push(batch.records);
1200 1 : if batch.wal_end_lsn == batch.available_wal_end_lsn {
1201 1 : break;
1202 1 : }
1203 1 : }
1204 1 :
1205 1 : let received_next_record_lsns = interpreted_records
1206 1 : .into_iter()
1207 1 : .flat_map(|b| b.records)
1208 9 : .map(|rec| rec.next_record_lsn)
1209 1 : .collect::<Vec<_>>();
1210 1 :
1211 1 : // By default this also includes the start LSN. Trim it since it shouldn't be received.
1212 1 : let next_record_lsns = next_record_lsns.into_iter().skip(1).collect::<Vec<_>>();
1213 1 :
1214 1 : assert_eq!(received_next_record_lsns, next_record_lsns);
1215 1 :
1216 1 : handle.abort();
1217 1 : let mut done = false;
1218 2 : for _ in 0..5 {
1219 2 : if handle.current_position().is_none() {
1220 1 : done = true;
1221 1 : break;
1222 1 : }
1223 1 : tokio::time::sleep(Duration::from_millis(1)).await;
1224 1 : }
1225 1 :
1226 1 : assert!(done);
1227 1 : }
1228 : }
|
