vector_buffers/topology/channel/
sender.rs

1use std::{sync::Arc, time::Instant};
2
3use async_recursion::async_recursion;
4use derivative::Derivative;
5use tokio::sync::Mutex;
6use tracing::Span;
7use vector_common::internal_event::{InternalEventHandle, Registered, register};
8
9use super::limited_queue::LimitedSender;
10use crate::{
11    Bufferable, WhenFull,
12    buffer_usage_data::BufferUsageHandle,
13    internal_events::BufferSendDuration,
14    variants::disk_v2::{self, ProductionFilesystem},
15};
16
17/// Adapter for papering over various sender backends.
18#[derive(Clone, Debug)]
19pub enum SenderAdapter<T: Bufferable> {
20    /// The in-memory channel buffer.
21    InMemory(LimitedSender<T>),
22
23    /// The disk v2 buffer.
24    DiskV2(Arc<Mutex<disk_v2::BufferWriter<T, ProductionFilesystem>>>),
25}
26
27impl<T: Bufferable> From<LimitedSender<T>> for SenderAdapter<T> {
28    fn from(v: LimitedSender<T>) -> Self {
29        Self::InMemory(v)
30    }
31}
32
33impl<T: Bufferable> From<disk_v2::BufferWriter<T, ProductionFilesystem>> for SenderAdapter<T> {
34    fn from(v: disk_v2::BufferWriter<T, ProductionFilesystem>) -> Self {
35        Self::DiskV2(Arc::new(Mutex::new(v)))
36    }
37}
38
39impl<T> SenderAdapter<T>
40where
41    T: Bufferable,
42{
43    pub(crate) async fn send(&mut self, item: T) -> crate::Result<()> {
44        match self {
45            Self::InMemory(tx) => tx.send(item).await.map_err(Into::into),
46            Self::DiskV2(writer) => {
47                let mut writer = writer.lock().await;
48
49                writer.write_record(item).await.map(|_| ()).map_err(|e| {
50                    // TODO: Could some errors be handled and not be unrecoverable? Right now,
51                    // encoding should theoretically be recoverable -- encoded value was too big, or
52                    // error during encoding -- but the traits don't allow for recovering the
53                    // original event value because we have to consume it to do the encoding... but
54                    // that might not always be the case.
55                    error!("Disk buffer writer has encountered an unrecoverable error.");
56
57                    e.into()
58                })
59            }
60        }
61    }
62
63    pub(crate) async fn try_send(&mut self, item: T) -> crate::Result<Option<T>> {
64        match self {
65            Self::InMemory(tx) => tx
66                .try_send(item)
67                .map(|()| None)
68                .or_else(|e| Ok(Some(e.into_inner()))),
69            Self::DiskV2(writer) => {
70                let mut writer = writer.lock().await;
71
72                writer.try_write_record(item).await.map_err(|e| {
73                    // TODO: Could some errors be handled and not be unrecoverable? Right now,
74                    // encoding should theoretically be recoverable -- encoded value was too big, or
75                    // error during encoding -- but the traits don't allow for recovering the
76                    // original event value because we have to consume it to do the encoding... but
77                    // that might not always be the case.
78                    error!("Disk buffer writer has encountered an unrecoverable error.");
79
80                    e.into()
81                })
82            }
83        }
84    }
85
86    pub(crate) async fn flush(&mut self) -> crate::Result<()> {
87        match self {
88            Self::InMemory(_) => Ok(()),
89            Self::DiskV2(writer) => {
90                let mut writer = writer.lock().await;
91                writer.flush().await.map_err(|e| {
92                    // Errors on the I/O path, which is all that flushing touches, are never recoverable.
93                    error!("Disk buffer writer has encountered an unrecoverable error.");
94
95                    e.into()
96                })
97            }
98        }
99    }
100
101    pub fn capacity(&self) -> Option<usize> {
102        match self {
103            Self::InMemory(tx) => Some(tx.available_capacity()),
104            Self::DiskV2(_) => None,
105        }
106    }
107}
108
109/// A buffer sender.
110///
111/// The sender handles sending events into the buffer, as well as the behavior around handling
112/// events when the internal channel is full.
113///
114/// When creating a buffer sender/receiver pair, callers can specify the "when full" behavior of the
115/// sender.  This controls how events are handled when the internal channel is full.  Three modes
116/// are possible:
117/// - block
118/// - drop newest
119/// - overflow
120///
121/// In "block" mode, callers are simply forced to wait until the channel has enough capacity to
122/// accept the event.  In "drop newest" mode, any event being sent when the channel is full will be
123/// dropped and proceed no further. In "overflow" mode, events will be sent to another buffer
124/// sender.  Callers can specify the overflow sender to use when constructing their buffers initially.
125///
126/// TODO: We should eventually rework `BufferSender`/`BufferReceiver` so that they contain a vector
127/// of the fields we already have here, but instead of cascading via calling into `overflow`, we'd
128/// linearize the nesting instead, so that `BufferSender` would only ever be calling the underlying
129/// `SenderAdapter` instances instead... which would let us get rid of the boxing and
130/// `#[async_recursion]` stuff.
131#[derive(Clone, Derivative)]
132#[derivative(Debug)]
133pub struct BufferSender<T: Bufferable> {
134    base: SenderAdapter<T>,
135    overflow: Option<Box<BufferSender<T>>>,
136    when_full: WhenFull,
137    instrumentation: Option<BufferUsageHandle>,
138    #[derivative(Debug = "ignore")]
139    send_duration: Option<Registered<BufferSendDuration>>,
140}
141
142impl<T: Bufferable> BufferSender<T> {
143    /// Creates a new [`BufferSender`] wrapping the given channel sender.
144    pub fn new(base: SenderAdapter<T>, when_full: WhenFull) -> Self {
145        Self {
146            base,
147            overflow: None,
148            when_full,
149            instrumentation: None,
150            send_duration: None,
151        }
152    }
153
154    /// Creates a new [`BufferSender`] wrapping the given channel sender and overflow sender.
155    pub fn with_overflow(base: SenderAdapter<T>, overflow: BufferSender<T>) -> Self {
156        Self {
157            base,
158            overflow: Some(Box::new(overflow)),
159            when_full: WhenFull::Overflow,
160            instrumentation: None,
161            send_duration: None,
162        }
163    }
164
165    /// Converts this sender into an overflowing sender using the given `BufferSender<T>`.
166    ///
167    /// Note: this resets the internal state of this sender, and so this should not be called except
168    /// when initially constructing `BufferSender<T>`.
169    #[cfg(test)]
170    pub fn switch_to_overflow(&mut self, overflow: BufferSender<T>) {
171        self.overflow = Some(Box::new(overflow));
172        self.when_full = WhenFull::Overflow;
173    }
174
175    /// Configures this sender to instrument the items passing through it.
176    pub fn with_usage_instrumentation(&mut self, handle: BufferUsageHandle) {
177        self.instrumentation = Some(handle);
178    }
179
180    /// Configures this sender to instrument the send duration.
181    pub fn with_send_duration_instrumentation(&mut self, stage: usize, span: &Span) {
182        let _enter = span.enter();
183        self.send_duration = Some(register(BufferSendDuration { stage }));
184    }
185}
186
187impl<T: Bufferable> BufferSender<T> {
188    #[cfg(test)]
189    pub(crate) fn get_base_ref(&self) -> &SenderAdapter<T> {
190        &self.base
191    }
192
193    #[cfg(test)]
194    pub(crate) fn get_overflow_ref(&self) -> Option<&BufferSender<T>> {
195        self.overflow.as_ref().map(AsRef::as_ref)
196    }
197
198    #[async_recursion]
199    pub async fn send(&mut self, item: T, send_reference: Option<Instant>) -> crate::Result<()> {
200        let item_sizing = self
201            .instrumentation
202            .as_ref()
203            .map(|_| (item.event_count(), item.size_of()));
204
205        let mut was_dropped = false;
206
207        if let Some(instrumentation) = self.instrumentation.as_ref()
208            && let Some((item_count, item_size)) = item_sizing
209        {
210            instrumentation
211                .increment_received_event_count_and_byte_size(item_count as u64, item_size as u64);
212        }
213        match self.when_full {
214            WhenFull::Block => self.base.send(item).await?,
215            WhenFull::DropNewest => {
216                if self.base.try_send(item).await?.is_some() {
217                    was_dropped = true;
218                }
219            }
220            WhenFull::Overflow => {
221                if let Some(item) = self.base.try_send(item).await? {
222                    was_dropped = true;
223                    self.overflow
224                        .as_mut()
225                        .unwrap_or_else(|| unreachable!("overflow must exist"))
226                        .send(item, send_reference)
227                        .await?;
228                }
229            }
230        }
231
232        if let Some(instrumentation) = self.instrumentation.as_ref()
233            && let Some((item_count, item_size)) = item_sizing
234            && was_dropped
235        {
236            instrumentation.increment_dropped_event_count_and_byte_size(
237                item_count as u64,
238                item_size as u64,
239                true,
240            );
241        }
242        if let Some(send_duration) = self.send_duration.as_ref()
243            && let Some(send_reference) = send_reference
244        {
245            send_duration.emit(send_reference.elapsed());
246        }
247
248        Ok(())
249    }
250
251    #[async_recursion]
252    pub async fn flush(&mut self) -> crate::Result<()> {
253        self.base.flush().await?;
254        if let Some(overflow) = self.overflow.as_mut() {
255            overflow.flush().await?;
256        }
257
258        Ok(())
259    }
260}
vector_buffers/topology/channel/sender.rs

vector_buffers/topology/channel/
sender.rs
