nominal_streaming/
stream.rs

1use std::collections::HashMap;
2use std::fmt::Debug;
3use std::path::PathBuf;
4use std::sync::atomic::AtomicBool;
5use std::sync::atomic::AtomicU64;
6use std::sync::atomic::AtomicUsize;
7use std::sync::atomic::Ordering;
8use std::sync::Arc;
9use std::thread;
10use std::time::Duration;
11use std::time::Instant;
12use std::time::UNIX_EPOCH;
13
14use conjure_object::BearerToken;
15use conjure_object::ResourceIdentifier;
16use nominal_api::tonic::io::nominal::scout::api::proto::array_points::ArrayType;
17use nominal_api::tonic::io::nominal::scout::api::proto::points::PointsType;
18use nominal_api::tonic::io::nominal::scout::api::proto::ArrayPoints;
19use nominal_api::tonic::io::nominal::scout::api::proto::Channel;
20use nominal_api::tonic::io::nominal::scout::api::proto::DoublePoint;
21use nominal_api::tonic::io::nominal::scout::api::proto::IntegerPoint;
22use nominal_api::tonic::io::nominal::scout::api::proto::Points;
23use nominal_api::tonic::io::nominal::scout::api::proto::Series;
24use nominal_api::tonic::io::nominal::scout::api::proto::StringPoint;
25use nominal_api::tonic::io::nominal::scout::api::proto::WriteRequestNominal;
26use parking_lot::Condvar;
27use parking_lot::Mutex;
28use parking_lot::MutexGuard;
29use tracing::debug;
30use tracing::error;
31use tracing::info;
32
33use crate::client::NominalApiClients;
34use crate::client::PRODUCTION_API_URL;
35use crate::consumer::AvroFileConsumer;
36use crate::consumer::DualWriteRequestConsumer;
37use crate::consumer::ListeningWriteRequestConsumer;
38use crate::consumer::NominalCoreConsumer;
39use crate::consumer::RequestConsumerWithFallback;
40use crate::consumer::WriteRequestConsumer;
41use crate::listener::LoggingListener;
42use crate::types::ChannelDescriptor;
43use crate::types::IntoPoints;
44use crate::types::IntoTimestamp;
45
46#[derive(Debug, Clone)]
47pub struct NominalStreamOpts {
48    pub max_points_per_record: usize,
49    pub max_request_delay: Duration,
50    pub max_buffered_requests: usize,
51    pub request_dispatcher_tasks: usize,
52    pub base_api_url: String,
53}
54
55impl Default for NominalStreamOpts {
56    fn default() -> Self {
57        Self {
58            max_points_per_record: 250_000,
59            max_request_delay: Duration::from_millis(100),
60            max_buffered_requests: 4,
61            request_dispatcher_tasks: 8,
62            base_api_url: PRODUCTION_API_URL.to_string(),
63        }
64    }
65}
66
67#[derive(Default)]
68pub struct NominalDatasetStreamBuilder {
69    stream_to_core: Option<(BearerToken, ResourceIdentifier, tokio::runtime::Handle)>,
70    stream_to_file: Option<PathBuf>,
71    file_fallback: Option<PathBuf>,
72    listeners: Vec<Arc<dyn crate::listener::NominalStreamListener>>,
73    opts: NominalStreamOpts,
74}
75
76impl Debug for NominalDatasetStreamBuilder {
77    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
78        f.debug_struct("NominalDatasetStreamBuilder")
79            .field("stream_to_core", &self.stream_to_core.is_some())
80            .field("stream_to_file", &self.stream_to_file)
81            .field("file_fallback", &self.file_fallback)
82            .field("listeners", &self.listeners.len())
83            .finish()
84    }
85}
86
87impl NominalDatasetStreamBuilder {
88    pub fn new() -> Self {
89        Self::default()
90    }
91}
92
93impl NominalDatasetStreamBuilder {
94    pub fn stream_to_core(
95        self,
96        bearer_token: BearerToken,
97        dataset: ResourceIdentifier,
98        handle: tokio::runtime::Handle,
99    ) -> NominalDatasetStreamBuilder {
100        NominalDatasetStreamBuilder {
101            stream_to_core: Some((bearer_token, dataset, handle)),
102            stream_to_file: self.stream_to_file,
103            file_fallback: self.file_fallback,
104            listeners: self.listeners,
105            opts: self.opts,
106        }
107    }
108
109    pub fn stream_to_file(mut self, file_path: impl Into<PathBuf>) -> Self {
110        self.stream_to_file = Some(file_path.into());
111        self
112    }
113
114    pub fn with_file_fallback(mut self, file_path: impl Into<PathBuf>) -> Self {
115        self.file_fallback = Some(file_path.into());
116        self
117    }
118
119    pub fn add_listener(
120        mut self,
121        listener: Arc<dyn crate::listener::NominalStreamListener>,
122    ) -> Self {
123        self.listeners.push(listener);
124        self
125    }
126
127    pub fn with_listeners(
128        mut self,
129        listeners: Vec<Arc<dyn crate::listener::NominalStreamListener>>,
130    ) -> Self {
131        self.listeners = listeners;
132        self
133    }
134
135    pub fn with_options(mut self, opts: NominalStreamOpts) -> Self {
136        self.opts = opts;
137        self
138    }
139
140    #[cfg(feature = "logging")]
141    fn init_logging(self, directive: Option<&str>) -> Self {
142        use tracing_subscriber::layer::SubscriberExt;
143        use tracing_subscriber::util::SubscriberInitExt;
144
145        // Build the filter, either from an explicit directive or the environment.
146        let base = tracing_subscriber::EnvFilter::builder()
147            .with_default_directive(tracing_subscriber::filter::LevelFilter::DEBUG.into());
148        let env_filter = match directive {
149            Some(d) => base.parse_lossy(d),
150            None => base.from_env_lossy(),
151        };
152
153        let subscriber = tracing_subscriber::registry()
154            .with(
155                tracing_subscriber::fmt::layer()
156                    .with_thread_ids(true)
157                    .with_thread_names(true)
158                    .with_line_number(true),
159            )
160            .with(env_filter);
161
162        if let Err(error) = subscriber.try_init() {
163            eprintln!("nominal streaming failed to enable logging: {error}");
164        }
165
166        self
167    }
168
169    #[cfg(feature = "logging")]
170    pub fn enable_logging(self) -> Self {
171        self.init_logging(None)
172    }
173
174    #[cfg(feature = "logging")]
175    pub fn enable_logging_with_directive(self, log_directive: &str) -> Self {
176        self.init_logging(Some(log_directive))
177    }
178
179    pub fn build(self) -> NominalDatasetStream {
180        let core_consumer = self.core_consumer();
181        let file_consumer = self.file_consumer();
182        let fallback_consumer = self.fallback_consumer();
183
184        match (core_consumer, file_consumer, fallback_consumer) {
185            (None, None, _) => panic!("nominal dataset stream must either stream to file or core"),
186            (Some(_), Some(_), Some(_)) => {
187                panic!("must choose one of stream_to_file and file_fallback when streaming to core")
188            }
189            (Some(core), None, None) => self.into_stream(core),
190            (Some(core), None, Some(fallback)) => {
191                self.into_stream(RequestConsumerWithFallback::new(core, fallback))
192            }
193            (None, Some(file), None) => self.into_stream(file),
194            (None, Some(file), Some(fallback)) => {
195                // todo: should this even be supported?
196                self.into_stream(RequestConsumerWithFallback::new(file, fallback))
197            }
198            (Some(core), Some(file), None) => {
199                self.into_stream(DualWriteRequestConsumer::new(core, file))
200            }
201        }
202    }
203
204    fn core_consumer(&self) -> Option<NominalCoreConsumer<BearerToken>> {
205        self.stream_to_core
206            .as_ref()
207            .map(|(auth_provider, dataset, handle)| {
208                NominalCoreConsumer::new(
209                    NominalApiClients::from_uri(self.opts.base_api_url.as_str()),
210                    handle.clone(),
211                    auth_provider.clone(),
212                    dataset.clone(),
213                )
214            })
215    }
216
217    fn file_consumer(&self) -> Option<AvroFileConsumer> {
218        self.stream_to_file
219            .as_ref()
220            .map(|path| AvroFileConsumer::new_with_full_path(path).unwrap())
221    }
222
223    fn fallback_consumer(&self) -> Option<AvroFileConsumer> {
224        self.file_fallback
225            .as_ref()
226            .map(|path| AvroFileConsumer::new_with_full_path(path).unwrap())
227    }
228
229    fn into_stream<C: WriteRequestConsumer + 'static>(self, consumer: C) -> NominalDatasetStream {
230        let mut listeners = self.listeners;
231        listeners.push(Arc::new(LoggingListener));
232        let listening_consumer = ListeningWriteRequestConsumer::new(consumer, listeners);
233        NominalDatasetStream::new_with_consumer(listening_consumer, self.opts)
234    }
235}
236
237// for backcompat, new code should use NominalDatasetStream
238#[deprecated]
239pub type NominalDatasourceStream = NominalDatasetStream;
240
241pub struct NominalDatasetStream {
242    opts: NominalStreamOpts,
243    running: Arc<AtomicBool>,
244    unflushed_points: Arc<AtomicUsize>,
245    primary_buffer: Arc<SeriesBuffer>,
246    secondary_buffer: Arc<SeriesBuffer>,
247    primary_handle: thread::JoinHandle<()>,
248    secondary_handle: thread::JoinHandle<()>,
249}
250
251impl NominalDatasetStream {
252    pub fn builder() -> NominalDatasetStreamBuilder {
253        NominalDatasetStreamBuilder::new()
254    }
255
256    pub fn new_with_consumer<C: WriteRequestConsumer + 'static>(
257        consumer: C,
258        opts: NominalStreamOpts,
259    ) -> Self {
260        let primary_buffer = Arc::new(SeriesBuffer::new(opts.max_points_per_record));
261        let secondary_buffer = Arc::new(SeriesBuffer::new(opts.max_points_per_record));
262
263        let (request_tx, request_rx) =
264            crossbeam_channel::bounded::<(WriteRequestNominal, usize)>(opts.max_buffered_requests);
265
266        let running = Arc::new(AtomicBool::new(true));
267        let unflushed_points = Arc::new(AtomicUsize::new(0));
268
269        let primary_handle = thread::Builder::new()
270            .name("nmstream_primary".to_string())
271            .spawn({
272                let points_buffer = Arc::clone(&primary_buffer);
273                let running = running.clone();
274                let tx = request_tx.clone();
275                move || {
276                    batch_processor(running, points_buffer, tx, opts.max_request_delay);
277                }
278            })
279            .unwrap();
280
281        let secondary_handle = thread::Builder::new()
282            .name("nmstream_secondary".to_string())
283            .spawn({
284                let secondary_buffer = Arc::clone(&secondary_buffer);
285                let running = running.clone();
286                move || {
287                    batch_processor(
288                        running,
289                        secondary_buffer,
290                        request_tx,
291                        opts.max_request_delay,
292                    );
293                }
294            })
295            .unwrap();
296
297        let consumer = Arc::new(consumer);
298
299        for i in 0..opts.request_dispatcher_tasks {
300            thread::Builder::new()
301                .name(format!("nmstream_dispatch_{i}"))
302                .spawn({
303                    let running = Arc::clone(&running);
304                    let unflushed_points = Arc::clone(&unflushed_points);
305                    let rx = request_rx.clone();
306                    let consumer = consumer.clone();
307                    move || {
308                        debug!("starting request dispatcher");
309                        request_dispatcher(running, unflushed_points, rx, consumer);
310                    }
311                })
312                .unwrap();
313        }
314
315        NominalDatasetStream {
316            opts,
317            running,
318            unflushed_points,
319            primary_buffer,
320            secondary_buffer,
321            primary_handle,
322            secondary_handle,
323        }
324    }
325
326    pub fn double_writer<'a>(
327        &'a self,
328        channel_descriptor: &'a ChannelDescriptor,
329    ) -> NominalDoubleWriter<'a> {
330        NominalDoubleWriter {
331            writer: NominalChannelWriter::new(self, channel_descriptor),
332        }
333    }
334
335    pub fn string_writer<'a>(
336        &'a self,
337        channel_descriptor: &'a ChannelDescriptor,
338    ) -> NominalStringWriter<'a> {
339        NominalStringWriter {
340            writer: NominalChannelWriter::new(self, channel_descriptor),
341        }
342    }
343
344    pub fn integer_writer<'a>(
345        &'a self,
346        channel_descriptor: &'a ChannelDescriptor,
347    ) -> NominalIntegerWriter<'a> {
348        NominalIntegerWriter {
349            writer: NominalChannelWriter::new(self, channel_descriptor),
350        }
351    }
352
353    pub fn enqueue(&self, channel_descriptor: &ChannelDescriptor, new_points: impl IntoPoints) {
354        let new_points = new_points.into_points();
355        let new_count = points_len(&new_points);
356
357        self.when_capacity(new_count, |mut sb| {
358            sb.extend(channel_descriptor, new_points)
359        });
360    }
361
362    fn when_capacity(&self, new_count: usize, callback: impl FnOnce(SeriesBufferGuard)) {
363        self.unflushed_points
364            .fetch_add(new_count, Ordering::Release);
365
366        if self.primary_buffer.has_capacity(new_count) {
367            debug!("adding {} points to primary buffer", new_count);
368            callback(self.primary_buffer.lock());
369        } else if self.secondary_buffer.has_capacity(new_count) {
370            // primary buffer is definitely full
371            self.primary_handle.thread().unpark();
372            debug!("adding {} points to secondary buffer", new_count);
373            callback(self.secondary_buffer.lock());
374        } else {
375            let buf = if self.primary_buffer < self.secondary_buffer {
376                info!("waiting for primary buffer to flush to append {new_count} points...");
377                self.primary_handle.thread().unpark();
378                &self.primary_buffer
379            } else {
380                info!("waiting for secondary buffer to flush to append {new_count} points...");
381                self.secondary_handle.thread().unpark();
382                &self.secondary_buffer
383            };
384
385            buf.on_notify(callback);
386        }
387    }
388}
389
390pub struct NominalChannelWriter<'ds, T>
391where
392    Vec<T>: IntoPoints,
393{
394    channel: &'ds ChannelDescriptor,
395    stream: &'ds NominalDatasetStream,
396    last_flushed_at: Instant,
397    unflushed: Vec<T>,
398}
399
400impl<T> NominalChannelWriter<'_, T>
401where
402    Vec<T>: IntoPoints,
403{
404    fn new<'ds>(
405        stream: &'ds NominalDatasetStream,
406        channel: &'ds ChannelDescriptor,
407    ) -> NominalChannelWriter<'ds, T> {
408        NominalChannelWriter {
409            channel,
410            stream,
411            last_flushed_at: Instant::now(),
412            unflushed: vec![],
413        }
414    }
415
416    fn push_point(&mut self, point: T) {
417        self.unflushed.push(point);
418        if self.unflushed.len() >= self.stream.opts.max_points_per_record
419            || self.last_flushed_at.elapsed() > self.stream.opts.max_request_delay
420        {
421            debug!(
422                "conditionally flushing {:?}, ({} points, {:?} since last)",
423                self.channel,
424                self.unflushed.len(),
425                self.last_flushed_at.elapsed()
426            );
427            self.flush();
428        }
429    }
430
431    fn flush(&mut self) {
432        if self.unflushed.is_empty() {
433            return;
434        }
435        info!(
436            "flushing writer for {:?} with {} points",
437            self.channel,
438            self.unflushed.len()
439        );
440        self.stream.when_capacity(self.unflushed.len(), |mut buf| {
441            let to_flush: Vec<T> = self.unflushed.drain(..).collect();
442            buf.extend(self.channel, to_flush);
443            self.last_flushed_at = Instant::now();
444        })
445    }
446}
447
448impl<T> Drop for NominalChannelWriter<'_, T>
449where
450    Vec<T>: IntoPoints,
451{
452    fn drop(&mut self) {
453        info!("flushing then dropping writer for: {:?}", self.channel);
454        self.flush();
455    }
456}
457
458pub struct NominalDoubleWriter<'ds> {
459    writer: NominalChannelWriter<'ds, DoublePoint>,
460}
461
462impl NominalDoubleWriter<'_> {
463    pub fn push(&mut self, timestamp: impl IntoTimestamp, value: f64) {
464        self.writer.push_point(DoublePoint {
465            timestamp: Some(timestamp.into_timestamp()),
466            value,
467        });
468    }
469}
470
471pub struct NominalIntegerWriter<'ds> {
472    writer: NominalChannelWriter<'ds, IntegerPoint>,
473}
474
475impl NominalIntegerWriter<'_> {
476    pub fn push(&mut self, timestamp: impl IntoTimestamp, value: i64) {
477        self.writer.push_point(IntegerPoint {
478            timestamp: Some(timestamp.into_timestamp()),
479            value,
480        });
481    }
482}
483
484pub struct NominalStringWriter<'ds> {
485    writer: NominalChannelWriter<'ds, StringPoint>,
486}
487
488impl NominalStringWriter<'_> {
489    pub fn push(&mut self, timestamp: impl IntoTimestamp, value: impl Into<String>) {
490        self.writer.push_point(StringPoint {
491            timestamp: Some(timestamp.into_timestamp()),
492            value: value.into(),
493        });
494    }
495}
496
497struct SeriesBuffer {
498    points: Mutex<HashMap<ChannelDescriptor, PointsType>>,
499    count: AtomicUsize,
500    flush_time: AtomicU64,
501    condvar: Condvar,
502    max_capacity: usize,
503}
504
505struct SeriesBufferGuard<'sb> {
506    sb: MutexGuard<'sb, HashMap<ChannelDescriptor, PointsType>>,
507    count: &'sb AtomicUsize,
508}
509
510impl SeriesBufferGuard<'_> {
511    fn extend(&mut self, channel_descriptor: &ChannelDescriptor, points: impl IntoPoints) {
512        let points = points.into_points();
513        let new_point_count = points_len(&points);
514
515        if !self.sb.contains_key(channel_descriptor) {
516            self.sb.insert(channel_descriptor.clone(), points);
517        } else {
518            match (self.sb.get_mut(channel_descriptor).unwrap(), points) {
519                (PointsType::DoublePoints(existing), PointsType::DoublePoints(new)) => {
520                    existing.points.extend(new.points)
521                }
522                (PointsType::StringPoints(existing), PointsType::StringPoints(new)) => {
523                    existing.points.extend(new.points)
524                }
525                (PointsType::IntegerPoints(existing), PointsType::IntegerPoints(new)) => {
526                    existing.points.extend(new.points)
527                }
528                (
529                    PointsType::ArrayPoints(ArrayPoints {
530                        array_type: Some(ArrayType::DoubleArrayPoints(existing)),
531                    }),
532                    PointsType::ArrayPoints(ArrayPoints {
533                        array_type: Some(ArrayType::DoubleArrayPoints(new)),
534                    }),
535                ) => existing.points.extend(new.points),
536                (
537                    PointsType::ArrayPoints(ArrayPoints {
538                        array_type: Some(ArrayType::StringArrayPoints(existing)),
539                    }),
540                    PointsType::ArrayPoints(ArrayPoints {
541                        array_type: Some(ArrayType::StringArrayPoints(new)),
542                    }),
543                ) => existing.points.extend(new.points),
544                (
545                    PointsType::ArrayPoints(ArrayPoints { array_type: None }),
546                    PointsType::ArrayPoints(ArrayPoints { array_type: None }),
547                ) => {}
548                // this is hideous, but exhaustive matching is good to avoid future errors
549                (
550                    PointsType::DoublePoints(_),
551                    PointsType::IntegerPoints(_)
552                    | PointsType::StringPoints(_)
553                    | PointsType::ArrayPoints(_),
554                )
555                | (
556                    PointsType::StringPoints(_),
557                    PointsType::DoublePoints(_)
558                    | PointsType::IntegerPoints(_)
559                    | PointsType::ArrayPoints(_),
560                )
561                | (
562                    PointsType::IntegerPoints(_),
563                    PointsType::DoublePoints(_)
564                    | PointsType::StringPoints(_)
565                    | PointsType::ArrayPoints(_),
566                )
567                | (
568                    PointsType::ArrayPoints(_),
569                    PointsType::DoublePoints(_)
570                    | PointsType::StringPoints(_)
571                    | PointsType::IntegerPoints(_),
572                )
573                | (
574                    PointsType::ArrayPoints(ArrayPoints {
575                        array_type: Some(_),
576                    }),
577                    PointsType::ArrayPoints(ArrayPoints { array_type: None }),
578                )
579                | (
580                    PointsType::ArrayPoints(ArrayPoints { array_type: None }),
581                    PointsType::ArrayPoints(ArrayPoints {
582                        array_type: Some(_),
583                    }),
584                )
585                | (
586                    PointsType::ArrayPoints(ArrayPoints {
587                        array_type: Some(ArrayType::DoubleArrayPoints(_)),
588                    }),
589                    PointsType::ArrayPoints(ArrayPoints {
590                        array_type: Some(ArrayType::StringArrayPoints(_)),
591                    }),
592                )
593                | (
594                    PointsType::ArrayPoints(ArrayPoints {
595                        array_type: Some(ArrayType::StringArrayPoints(_)),
596                    }),
597                    PointsType::ArrayPoints(ArrayPoints {
598                        array_type: Some(ArrayType::DoubleArrayPoints(_)),
599                    }),
600                ) => {
601                    // todo: improve error
602                    panic!("mismatched types");
603                }
604            }
605        }
606
607        self.count.fetch_add(new_point_count, Ordering::Release);
608    }
609}
610
611impl PartialEq for SeriesBuffer {
612    fn eq(&self, other: &Self) -> bool {
613        self.flush_time.load(Ordering::Acquire) == other.flush_time.load(Ordering::Acquire)
614    }
615}
616
617impl PartialOrd for SeriesBuffer {
618    fn partial_cmp(&self, other: &Self) -> Option<std::cmp::Ordering> {
619        let flush_time = self.flush_time.load(Ordering::Acquire);
620        let other_flush_time = other.flush_time.load(Ordering::Acquire);
621        flush_time.partial_cmp(&other_flush_time)
622    }
623}
624
625impl SeriesBuffer {
626    fn new(capacity: usize) -> Self {
627        Self {
628            points: Mutex::new(HashMap::new()),
629            count: AtomicUsize::new(0),
630            flush_time: AtomicU64::new(0),
631            condvar: Condvar::new(),
632            max_capacity: capacity,
633        }
634    }
635
636    /// Checks if the buffer has enough capacity to add new points.
637    /// Note that the buffer can be larger than MAX_POINTS_PER_RECORD if a single batch of points
638    /// larger than MAX_POINTS_PER_RECORD is inserted while the buffer is empty. This avoids needing
639    /// to handle splitting batches of points across multiple requests.
640    fn has_capacity(&self, new_points_count: usize) -> bool {
641        let count = self.count.load(Ordering::Acquire);
642        count == 0 || count + new_points_count <= self.max_capacity
643    }
644
645    fn lock(&self) -> SeriesBufferGuard<'_> {
646        SeriesBufferGuard {
647            sb: self.points.lock(),
648            count: &self.count,
649        }
650    }
651
652    fn take(&self) -> (usize, Vec<Series>) {
653        let mut points = self.lock();
654        self.flush_time.store(
655            UNIX_EPOCH.elapsed().unwrap().as_nanos() as u64,
656            Ordering::Release,
657        );
658        let result = points
659            .sb
660            .drain()
661            .map(|(ChannelDescriptor { name, tags }, points)| {
662                let channel = Channel { name };
663                let points_obj = Points {
664                    points_type: Some(points),
665                };
666                Series {
667                    channel: Some(channel),
668                    tags: tags
669                        .map(|tags| tags.into_iter().collect())
670                        .unwrap_or_default(),
671                    points: Some(points_obj),
672                }
673            })
674            .collect();
675        let result_count = self
676            .count
677            .fetch_update(Ordering::Release, Ordering::Acquire, |_| Some(0))
678            .unwrap();
679        (result_count, result)
680    }
681
682    fn is_empty(&self) -> bool {
683        self.count() == 0
684    }
685
686    fn count(&self) -> usize {
687        self.count.load(Ordering::Acquire)
688    }
689
690    fn on_notify(&self, on_notify: impl FnOnce(SeriesBufferGuard)) {
691        let mut points_lock = self.points.lock();
692        // concurrency bug without this - the buffer could have been emptied since we
693        // checked the count, so this will wait forever & block any new points from entering
694        if !points_lock.is_empty() {
695            self.condvar.wait(&mut points_lock);
696        } else {
697            debug!("buffer emptied since last check, skipping condvar wait");
698        }
699        on_notify(SeriesBufferGuard {
700            sb: points_lock,
701            count: &self.count,
702        });
703    }
704
705    fn notify(&self) -> bool {
706        self.condvar.notify_one()
707    }
708}
709
710fn batch_processor(
711    running: Arc<AtomicBool>,
712    points_buffer: Arc<SeriesBuffer>,
713    request_chan: crossbeam_channel::Sender<(WriteRequestNominal, usize)>,
714    max_request_delay: Duration,
715) {
716    loop {
717        debug!("starting processor loop");
718        if points_buffer.is_empty() {
719            if !running.load(Ordering::Acquire) {
720                debug!("batch processor thread exiting due to running flag");
721                drop(request_chan);
722                break;
723            } else {
724                debug!("empty points buffer, waiting");
725                thread::park_timeout(max_request_delay);
726            }
727            continue;
728        }
729        let (point_count, series) = points_buffer.take();
730
731        if points_buffer.notify() {
732            debug!("notified one waiting thread after clearing points buffer");
733        }
734
735        let write_request = WriteRequestNominal { series };
736
737        if request_chan.is_full() {
738            debug!("ready to queue request but request channel is full");
739        }
740        let rep = request_chan.send((write_request, point_count));
741        debug!("queued request for processing");
742        if rep.is_err() {
743            error!("failed to send request to dispatcher");
744        } else {
745            debug!("finished submitting request");
746        }
747
748        thread::park_timeout(max_request_delay);
749    }
750    debug!("batch processor thread exiting");
751}
752
753impl Drop for NominalDatasetStream {
754    fn drop(&mut self) {
755        debug!("starting drop for NominalDatasetStream");
756        self.running.store(false, Ordering::Release);
757        while self.unflushed_points.load(Ordering::Acquire) > 0 {
758            debug!(
759                "waiting for all points to be flushed before dropping stream, {} points remaining",
760                self.unflushed_points.load(Ordering::Acquire)
761            );
762            // todo: reduce this + give up after some maximum timeout is reached
763            thread::sleep(Duration::from_millis(50));
764        }
765    }
766}
767
768fn request_dispatcher<C: WriteRequestConsumer + 'static>(
769    running: Arc<AtomicBool>,
770    unflushed_points: Arc<AtomicUsize>,
771    request_rx: crossbeam_channel::Receiver<(WriteRequestNominal, usize)>,
772    consumer: Arc<C>,
773) {
774    let mut total_request_time = 0;
775    loop {
776        match request_rx.recv() {
777            Ok((request, point_count)) => {
778                debug!("received writerequest from channel");
779                let req_start = Instant::now();
780                match consumer.consume(&request) {
781                    Ok(_) => {
782                        let time = req_start.elapsed().as_millis();
783                        debug!("request of {} points sent in {} ms", point_count, time);
784                        total_request_time += time as u64;
785                    }
786                    Err(e) => {
787                        error!("Failed to send request: {e:?}");
788                    }
789                }
790                unflushed_points.fetch_sub(point_count, Ordering::Release);
791
792                if unflushed_points.load(Ordering::Acquire) == 0 && !running.load(Ordering::Acquire)
793                {
794                    info!("all points flushed, closing dispatcher thread");
795                    // notify the processor thread that all points have been flushed
796                    drop(request_rx);
797                    break;
798                }
799            }
800            Err(e) => {
801                debug!("request channel closed, exiting dispatcher thread. info: '{e}'");
802                break;
803            }
804        }
805    }
806    debug!(
807        "request dispatcher thread exiting. total request time: {}",
808        total_request_time
809    );
810}
811
812fn points_len(points_type: &PointsType) -> usize {
813    match points_type {
814        PointsType::DoublePoints(points) => points.points.len(),
815        PointsType::StringPoints(points) => points.points.len(),
816        PointsType::IntegerPoints(points) => points.points.len(),
817        // is this correct?
818        PointsType::ArrayPoints(points) => match &points.array_type {
819            Some(ArrayType::DoubleArrayPoints(points)) => points.points.len(),
820            Some(ArrayType::StringArrayPoints(points)) => points.points.len(),
821            None => 0,
822        },
823    }
824}
nominal_streaming/stream.rs

nominal_streaming/
stream.rs
