4698125137

Build Type

github

Committed by GitHub

Commit Message

Merge ef21858a9 into ca6ea5f58

Pull Request Pull Request #1432: chore: Add deb generation to release process

Run Details

34321 of 44357 relevant lines covered (77.37%)

16844.6 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

94.74

/dozer-core/src/executor/source_node.rs

use std::{
    sync::{
        atomic::{AtomicBool, Ordering},
        Arc,
    },
    time::Duration,
};

use crossbeam::channel::{bounded, Receiver, RecvTimeoutError, Sender};
use dozer_types::ingestion_types::IngestionMessage;
use dozer_types::{
    log::debug,
    node::{NodeHandle, OpIdentifier},
};

use crate::{
    builder_dag::NodeKind,
    channels::SourceChannelForwarder,
    errors::ExecutionError,
    forwarder::{SourceChannelManager, StateWriter},
    node::{PortHandle, Source},
};

use super::{execution_dag::ExecutionDag, node::Node, ExecutorOptions};

impl SourceChannelForwarder for InternalChannelSourceForwarder {
    fn send(&mut self, message: IngestionMessage, port: PortHandle) -> Result<(), ExecutionError> {
        Ok(self.sender.send((port, message))?)
    }
}

/// The sender half of a source in the execution DAG.
#[derive(Debug)]
pub struct SourceSenderNode {
    /// Node handle in description DAG.
    node_handle: NodeHandle,
    /// The source.
    source: Box<dyn Source>,
    /// Last checkpointed output data sequence number.
    last_checkpoint: Option<OpIdentifier>,
    /// The forwarder that will be passed to the source for outputting data.
    forwarder: InternalChannelSourceForwarder,
}

impl SourceSenderNode {
    pub fn handle(&self) -> &NodeHandle {
        &self.node_handle
    }
}

impl Node for SourceSenderNode {
    fn run(mut self) -> Result<(), ExecutionError> {
        let result = self.source.start(
            &mut self.forwarder,
            self.last_checkpoint
                .map(|op_id| (op_id.txid, op_id.seq_in_tx)),
        );
        debug!("[{}-sender] Quit", self.node_handle);
        result
    }
}

/// The listener part of a source in the execution DAG.
#[derive(Debug)]
pub struct SourceListenerNode {
    /// Node handle in description DAG.
    node_handle: NodeHandle,
    /// Output from corresponding source sender.
    receiver: Receiver<(PortHandle, IngestionMessage)>,
    /// Receiving timeout.
    timeout: Duration,
    /// If the execution DAG should be running. Used for determining if a `terminate` message should be sent.
    running: Arc<AtomicBool>,
    /// This node's output channel manager, for communicating to other sources to coordinate terminate and commit, forwarding data, writing metadata and writing port state.
    channel_manager: SourceChannelManager,
}

#[derive(Debug, Clone, PartialEq)]
enum DataKind {
    Data((PortHandle, IngestionMessage)),
    NoDataBecauseOfTimeout,
    NoDataBecauseOfChannelDisconnection,
}

impl SourceListenerNode {
    /// Returns if the node should terminate.
    fn send_and_trigger_commit_if_needed(
        &mut self,
        data: DataKind,
    ) -> Result<bool, ExecutionError> {
        // If termination was requested the or source quit, we try to terminate.
        let terminating = data == DataKind::NoDataBecauseOfChannelDisconnection
            || !self.running.load(Ordering::SeqCst);
        // If this commit was not requested with termination at the start, we shouldn't terminate either.
        let terminating = match data {
            DataKind::Data((port, message)) => self
                .channel_manager
                .send_and_trigger_commit_if_needed(message, port, terminating)?,
            DataKind::NoDataBecauseOfTimeout | DataKind::NoDataBecauseOfChannelDisconnection => {
                self.channel_manager.trigger_commit_if_needed(terminating)?
            }
        };
        if terminating {
            self.channel_manager.terminate()?;
            debug!("[{}-listener] Quitting", &self.node_handle);
        }
        Ok(terminating)
    }
}

impl Node for SourceListenerNode {
    fn run(mut self) -> Result<(), ExecutionError> {
        loop {
            let terminating = match self.receiver.recv_timeout(self.timeout) {
                Ok(data) => self.send_and_trigger_commit_if_needed(DataKind::Data(data))?,
                Err(RecvTimeoutError::Timeout) => {
                    self.send_and_trigger_commit_if_needed(DataKind::NoDataBecauseOfTimeout)?
                }
                Err(RecvTimeoutError::Disconnected) => self.send_and_trigger_commit_if_needed(
                    DataKind::NoDataBecauseOfChannelDisconnection,
                )?,
            };
            if terminating {
                return Ok(());
            }
        }
    }
}

#[derive(Debug)]
struct InternalChannelSourceForwarder {
    sender: Sender<(PortHandle, IngestionMessage)>,
}

impl InternalChannelSourceForwarder {
    pub fn new(sender: Sender<(PortHandle, IngestionMessage)>) -> Self {
        Self { sender }
    }
}

pub fn create_source_nodes(
    dag: &mut ExecutionDag,
    node_index: daggy::NodeIndex,
    options: &ExecutorOptions,
    running: Arc<AtomicBool>,
) -> (SourceSenderNode, SourceListenerNode) {
    // Get the source node.
    let Some(node) = dag.node_weight_mut(node_index).take() else {
        panic!("Must pass in a node")
    };
    let node_handle = node.handle;
    let NodeKind::Source(source, last_checkpoint) = node.kind else {
        panic!("Must pass in a source node");
    };

    // Create channel between source sender and source listener.
    let (source_sender, source_receiver) = bounded(options.channel_buffer_sz);
    // let (source_sender, source_receiver) = bounded(1);

    // Create source listener.
    let forwarder = InternalChannelSourceForwarder::new(source_sender);
    let source_sender_node = SourceSenderNode {
        node_handle: node_handle.clone(),
        source,
        last_checkpoint,
        forwarder,
    };

    // Create source sender node.
    let (senders, record_writers) = dag.collect_senders_and_record_writers(node_index);
    let state_writer = StateWriter::new(record_writers);
    let channel_manager = SourceChannelManager::new(
        node_handle.clone(),
        senders,
        state_writer,
        true,
        options.commit_sz,
        options.commit_time_threshold,
        dag.epoch_manager().clone(),
    );
    let source_listener_node = SourceListenerNode {
        node_handle,
        receiver: source_receiver,
        timeout: options.commit_time_threshold,
        running,
        channel_manager,
    };

    (source_sender_node, source_listener_node)
}

1	use std::{
2	sync::{
3	atomic::{AtomicBool, Ordering},
4	Arc,
5	},
6	time::Duration,
7	};
8
9	use crossbeam::channel::{bounded, Receiver, RecvTimeoutError, Sender};
10	use dozer_types::ingestion_types::IngestionMessage;
11	use dozer_types::{
12	log::debug,
13	node::{NodeHandle, OpIdentifier},
14	};
15
16	use crate::{
17	builder_dag::NodeKind,
18	channels::SourceChannelForwarder,
19	errors::ExecutionError,
20	forwarder::{SourceChannelManager, StateWriter},
21	node::{PortHandle, Source},
22	};
23
24	use super::{execution_dag::ExecutionDag, node::Node, ExecutorOptions};
25
26	impl SourceChannelForwarder for InternalChannelSourceForwarder {
27	fn send(&mut self, message: IngestionMessage, port: PortHandle) -> Result<(), ExecutionError> {	2,145,283✔
28	Ok(self.sender.send((port, message))?)	2,145,283✔
29	}	2,145,283✔
30	}
31
32	/// The sender half of a source in the execution DAG.
33	#[derive(Debug)]	×
34	pub struct SourceSenderNode {
35	/// Node handle in description DAG.
36	node_handle: NodeHandle,
37	/// The source.
38	source: Box<dyn Source>,
39	/// Last checkpointed output data sequence number.
40	last_checkpoint: Option<OpIdentifier>,
41	/// The forwarder that will be passed to the source for outputting data.
42	forwarder: InternalChannelSourceForwarder,
43	}
44
45	impl SourceSenderNode {
46	pub fn handle(&self) -> &NodeHandle {	675✔
47	&self.node_handle	675✔
48	}	675✔
49	}
50
51	impl Node for SourceSenderNode {
52	fn run(mut self) -> Result<(), ExecutionError> {	675✔
53	let result = self.source.start(	675✔
54	&mut self.forwarder,	675✔
55	self.last_checkpoint	675✔
56	.map(\|op_id\| (op_id.txid, op_id.seq_in_tx)),	675✔
57	);	675✔
58	debug!("[{}-sender] Quit", self.node_handle);	675✔
59	result	648✔
60	}	648✔
61	}
62
63	/// The listener part of a source in the execution DAG.
64	#[derive(Debug)]	×
65	pub struct SourceListenerNode {
66	/// Node handle in description DAG.
67	node_handle: NodeHandle,
68	/// Output from corresponding source sender.
69	receiver: Receiver<(PortHandle, IngestionMessage)>,
70	/// Receiving timeout.
71	timeout: Duration,
72	/// If the execution DAG should be running. Used for determining if a `terminate` message should be sent.
73	running: Arc<AtomicBool>,
74	/// This node's output channel manager, for communicating to other sources to coordinate terminate and commit, forwarding data, writing metadata and writing port state.
75	channel_manager: SourceChannelManager,
76	}
77
78	#[derive(Debug, Clone, PartialEq)]	2,049,741✔
79	enum DataKind {
80	Data((PortHandle, IngestionMessage)),
81	NoDataBecauseOfTimeout,
82	NoDataBecauseOfChannelDisconnection,
83	}
84
85	impl SourceListenerNode {
86	/// Returns if the node should terminate.
87	fn send_and_trigger_commit_if_needed(	2,035,511✔
88	&mut self,	2,035,511✔
89	data: DataKind,	2,035,511✔
90	) -> Result<bool, ExecutionError> {	2,035,511✔
91	// If termination was requested the or source quit, we try to terminate.
92	let terminating = data == DataKind::NoDataBecauseOfChannelDisconnection	2,035,511✔
93	\|\| !self.running.load(Ordering::SeqCst);	2,034,846✔
94	// If this commit was not requested with termination at the start, we shouldn't terminate either.
95	let terminating = match data {	2,035,511✔
96	DataKind::Data((port, message)) => self	2,032,422✔
97	.channel_manager	2,032,422✔
98	.send_and_trigger_commit_if_needed(message, port, terminating)?,	2,032,422✔
99	DataKind::NoDataBecauseOfTimeout \| DataKind::NoDataBecauseOfChannelDisconnection => {
100	self.channel_manager.trigger_commit_if_needed(terminating)?	3,089✔
101	}
102	};
103	if terminating {	2,035,508✔
104	self.channel_manager.terminate()?;	6,272✔
105	debug!("[{}-listener] Quitting", &self.node_handle);	6,272✔
106	}	2,029,236✔
107	Ok(terminating)	2,029,908✔
108	}	2,029,911✔
109	}
110
111	impl Node for SourceListenerNode {
112	fn run(mut self) -> Result<(), ExecutionError> {	675✔
113	loop {
114	let terminating = match self.receiver.recv_timeout(self.timeout) {	2,045,401✔
115	Ok(data) => self.send_and_trigger_commit_if_needed(DataKind::Data(data))?,	2,042,312✔
116	Err(RecvTimeoutError::Timeout) => {
117	self.send_and_trigger_commit_if_needed(DataKind::NoDataBecauseOfTimeout)?	2,424✔
118	}
119	Err(RecvTimeoutError::Disconnected) => self.send_and_trigger_commit_if_needed(	665✔
120	DataKind::NoDataBecauseOfChannelDisconnection,	665✔
121	)?,	665✔
122	};
123	if terminating {	2,045,398✔
124	return Ok(());	672✔
125	}	2,044,726✔
126	}
127	}	675✔
128	}
129
130	#[derive(Debug)]	×
131	struct InternalChannelSourceForwarder {
132	sender: Sender<(PortHandle, IngestionMessage)>,
133	}
134
135	impl InternalChannelSourceForwarder {
136	pub fn new(sender: Sender<(PortHandle, IngestionMessage)>) -> Self {	675✔
137	Self { sender }	675✔
138	}	675✔
139	}
140
141	pub fn create_source_nodes(	675✔
142	dag: &mut ExecutionDag,	675✔
143	node_index: daggy::NodeIndex,	675✔
144	options: &ExecutorOptions,	675✔
145	running: Arc<AtomicBool>,	675✔
146	) -> (SourceSenderNode, SourceListenerNode) {	675✔
147	// Get the source node.
148	let Some(node) = dag.node_weight_mut(node_index).take() else {	675✔
149	panic!("Must pass in a node")	×
150	};
151	let node_handle = node.handle;	675✔
152	let NodeKind::Source(source, last_checkpoint) = node.kind else {	675✔
153	panic!("Must pass in a source node");	×
154	};
155
156	// Create channel between source sender and source listener.
157	let (source_sender, source_receiver) = bounded(options.channel_buffer_sz);	675✔
158	// let (source_sender, source_receiver) = bounded(1);	675✔
159		675✔
160	// Create source listener.	675✔
161	let forwarder = InternalChannelSourceForwarder::new(source_sender);	675✔
162	let source_sender_node = SourceSenderNode {	675✔
163	node_handle: node_handle.clone(),	675✔
164	source,	675✔
165	last_checkpoint,	675✔
166	forwarder,	675✔
167	};	675✔
168		675✔
169	// Create source sender node.	675✔
170	let (senders, record_writers) = dag.collect_senders_and_record_writers(node_index);	675✔
171	let state_writer = StateWriter::new(record_writers);	675✔
172	let channel_manager = SourceChannelManager::new(	675✔
173	node_handle.clone(),	675✔
174	senders,	675✔
175	state_writer,	675✔
176	true,	675✔
177	options.commit_sz,	675✔
178	options.commit_time_threshold,	675✔
179	dag.epoch_manager().clone(),	675✔
180	);	675✔
181	let source_listener_node = SourceListenerNode {	675✔
182	node_handle,	675✔
183	receiver: source_receiver,	675✔
184	timeout: options.commit_time_threshold,	675✔
185	running,	675✔
186	channel_manager,	675✔
187	};	675✔
188		675✔
189	(source_sender_node, source_listener_node)	675✔
190	}	675✔

getdozer / dozer / 4698125137

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous