4315007357

Build Type

push

github

Committed by GitHub

Commit Message

fix: Sink should only be built after all source checkpoints are checked (#1112)

Run Details

280 of 280 new or added lines in 24 files covered. (100.0%)

28292 of 38914 relevant lines covered (72.7%)

64132.7 hits per line

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

77.54

/dozer-core/src/executor.rs

use crate::builder_dag::{BuilderDag, NodeKind};
use crate::dag_metadata::DagMetadata;
use crate::dag_schemas::DagSchemas;
use crate::errors::ExecutionError;
use crate::Dag;

use daggy::petgraph::visit::IntoNodeIdentifiers;
use dozer_types::node::NodeHandle;
use dozer_types::types::Operation;

use crate::epoch::Epoch;
use std::collections::hash_map::Entry;
use std::collections::HashMap;
use std::fmt::Debug;
use std::panic::panic_any;
use std::path::PathBuf;
use std::sync::atomic::AtomicBool;
use std::sync::Arc;
use std::thread::JoinHandle;
use std::thread::{self, Builder};
use std::time::Duration;

#[derive(Clone)]
pub struct ExecutorOptions {
    pub commit_sz: u32,
    pub channel_buffer_sz: usize,
    pub commit_time_threshold: Duration,

    pub max_map_size: usize,
}

impl Default for ExecutorOptions {
    fn default() -> Self {
        Self {
            commit_sz: 10_000,
            channel_buffer_sz: 20_000,
            commit_time_threshold: Duration::from_millis(50),
            max_map_size: 1024 * 1024 * 1024 * 1024,
        }
    }
}

#[derive(Clone, Debug, PartialEq, Eq)]
pub(crate) enum InputPortState {
    Open,
    Terminated,
}

#[derive(Clone, Debug, PartialEq, Eq)]
pub enum ExecutorOperation {
    Op { op: Operation },
    Commit { epoch: Epoch },
    Terminate,
    SnapshottingDone {},
}

mod execution_dag;
mod name;
mod node;
mod processor_node;
mod receiver_loop;
mod sink_node;
mod source_node;

use node::Node;
use processor_node::ProcessorNode;
use sink_node::SinkNode;

use self::execution_dag::ExecutionDag;
use self::source_node::{create_source_nodes, SourceListenerNode, SourceSenderNode};

pub struct DagExecutor {
    builder_dag: BuilderDag,
    options: ExecutorOptions,
}

pub struct DagExecutorJoinHandle {
    join_handles: HashMap<NodeHandle, JoinHandle<()>>,
}

impl DagExecutor {
    pub fn new<T: Clone + Debug>(
        dag: Dag<T>,
        path: PathBuf,
        options: ExecutorOptions,
    ) -> Result<Self, ExecutionError> {
        let dag_schemas = DagSchemas::new(dag)?;
        let builder_dag = BuilderDag::new(dag_schemas, path, options.max_map_size)?;

        Ok(Self {
            builder_dag,
            options,
        })
    }

    pub fn validate<T: Clone + Debug>(dag: Dag<T>, path: PathBuf) -> Result<(), ExecutionError> {
        let dag_schemas = DagSchemas::new(dag)?;
        DagMetadata::new(dag_schemas, path)?;
        Ok(())
    }

    pub fn start(self, running: Arc<AtomicBool>) -> Result<DagExecutorJoinHandle, ExecutionError> {
        // Construct execution dag.
        let mut execution_dag =
            ExecutionDag::new(self.builder_dag, self.options.channel_buffer_sz)?;
        let node_indexes = execution_dag.graph().node_identifiers().collect::<Vec<_>>();

        // Start the threads.
        let mut join_handles = HashMap::new();
        for node_index in node_indexes {
            let node = execution_dag.graph()[node_index]
                .as_ref()
                .expect("We created all nodes");
            let node_handle = node.handle.clone();
            match &node.kind {
                NodeKind::Source(_, _) => {
                    let (source_sender_node, source_listener_node) = create_source_nodes(
                        &mut execution_dag,
                        node_index,
                        &self.options,
                        running.clone(),
                    );
                    join_handles.insert(
                        node_handle,
                        start_source(source_sender_node, source_listener_node)?,
                    );
                }
                NodeKind::Processor(_) => {
                    let processor_node = ProcessorNode::new(&mut execution_dag, node_index);
                    join_handles.insert(node_handle, start_processor(processor_node)?);
                }
                NodeKind::Sink(_) => {
                    let sink_node = SinkNode::new(&mut execution_dag, node_index);
                    join_handles.insert(node_handle, start_sink(sink_node)?);
                }
            }
        }

        Ok(DagExecutorJoinHandle { join_handles })
    }
}

impl DagExecutorJoinHandle {
    pub fn join(mut self) -> Result<(), ExecutionError> {
        let handles: Vec<NodeHandle> = self.join_handles.iter().map(|e| e.0.clone()).collect();

        loop {
            for handle in &handles {
                if let Entry::Occupied(entry) = self.join_handles.entry(handle.clone()) {
                    if entry.get().is_finished() {
                        if let Err(e) = entry.remove().join() {
                            panic_any(e);
                        }
                    }
                }
            }

            if self.join_handles.is_empty() {
                return Ok(());
            }

            thread::sleep(Duration::from_millis(250));
        }
    }
}

fn start_source(
    source_sender: SourceSenderNode,
    source_listener: SourceListenerNode,
) -> Result<JoinHandle<()>, ExecutionError> {
    let handle = source_sender.handle().clone();

    let _st_handle = Builder::new()
        .name(format!("{handle}-sender"))
        .spawn(move || match source_sender.run() {
            Ok(_) => {}
            // Channel disconnection means the source listener has quit.
            // Maybe it quit gracefully so we don't need to panic.
            Err(ExecutionError::CannotSendToChannel) => {}
            // Other errors result in panic.
            Err(e) => std::panic::panic_any(e),
        })?;

    Ok(Builder::new()
        .name(format!("{handle}-listener"))
        .spawn(move || {
            if let Err(e) = source_listener.run() {
                std::panic::panic_any(e);
            }
        })?)
}

fn start_processor(processor: ProcessorNode) -> Result<JoinHandle<()>, ExecutionError> {
    Ok(Builder::new()
        .name(processor.handle().to_string())
        .spawn(move || {
            if let Err(e) = processor.run() {
                std::panic::panic_any(e);
            }
        })?)
}

fn start_sink(sink: SinkNode) -> Result<JoinHandle<()>, ExecutionError> {
    Ok(Builder::new().name(sink.handle().to_string()).spawn(|| {
        if let Err(e) = sink.run() {
            std::panic::panic_any(e);
        }
    })?)
}

1	use crate::builder_dag::{BuilderDag, NodeKind};
2	use crate::dag_metadata::DagMetadata;
3	use crate::dag_schemas::DagSchemas;
4	use crate::errors::ExecutionError;
5	use crate::Dag;
6
7	use daggy::petgraph::visit::IntoNodeIdentifiers;
8	use dozer_types::node::NodeHandle;
9	use dozer_types::types::Operation;
10
11	use crate::epoch::Epoch;
12	use std::collections::hash_map::Entry;
13	use std::collections::HashMap;
14	use std::fmt::Debug;
15	use std::panic::panic_any;
16	use std::path::PathBuf;
17	use std::sync::atomic::AtomicBool;
18	use std::sync::Arc;
19	use std::thread::JoinHandle;
20	use std::thread::{self, Builder};
21	use std::time::Duration;
22
23	#[derive(Clone)]	×
24	pub struct ExecutorOptions {
25	pub commit_sz: u32,
26	pub channel_buffer_sz: usize,
27	pub commit_time_threshold: Duration,
28
29	pub max_map_size: usize,
30	}
31
32	impl Default for ExecutorOptions {
33	fn default() -> Self {	330✔
34	Self {	330✔
35	commit_sz: 10_000,	330✔
36	channel_buffer_sz: 20_000,	330✔
37	commit_time_threshold: Duration::from_millis(50),	330✔
38	max_map_size: 1024 * 1024 * 1024 * 1024,	330✔
39	}	330✔
40	}	330✔
41	}
42
43	#[derive(Clone, Debug, PartialEq, Eq)]	1,661✔
44	pub(crate) enum InputPortState {
45	Open,
46	Terminated,
47	}
48
49	#[derive(Clone, Debug, PartialEq, Eq)]	83,248✔
50	pub enum ExecutorOperation {
51	Op { op: Operation },
52	Commit { epoch: Epoch },
53	Terminate,
54	SnapshottingDone {},
55	}
56
57	mod execution_dag;
58	mod name;
59	mod node;
60	mod processor_node;
61	mod receiver_loop;
62	mod sink_node;
63	mod source_node;
64
65	use node::Node;
66	use processor_node::ProcessorNode;
67	use sink_node::SinkNode;
68
69	use self::execution_dag::ExecutionDag;
70	use self::source_node::{create_source_nodes, SourceListenerNode, SourceSenderNode};
71
72	pub struct DagExecutor {
73	builder_dag: BuilderDag,
74	options: ExecutorOptions,
75	}
76
77	pub struct DagExecutorJoinHandle {
78	join_handles: HashMap<NodeHandle, JoinHandle<()>>,
79	}
80
81	impl DagExecutor {	×
82	pub fn new<T: Clone + Debug>(	223✔
83	dag: Dag<T>,	223✔
84	path: PathBuf,	223✔
85	options: ExecutorOptions,	223✔
86	) -> Result<Self, ExecutionError> {	223✔
87	let dag_schemas = DagSchemas::new(dag)?;	223✔
88	let builder_dag = BuilderDag::new(dag_schemas, path, options.max_map_size)?;	223✔
89		×
90	Ok(Self {	219✔
91	builder_dag,	219✔
92	options,	219✔
93	})	219✔
94	}	221✔
95		×
96	pub fn validate<T: Clone + Debug>(dag: Dag<T>, path: PathBuf) -> Result<(), ExecutionError> {	8✔
97	let dag_schemas = DagSchemas::new(dag)?;	8✔
98	DagMetadata::new(dag_schemas, path)?;	8✔
99	Ok(())	8✔
100	}	8✔
101		×
102	pub fn start(self, running: Arc<AtomicBool>) -> Result<DagExecutorJoinHandle, ExecutionError> {	332✔
103	// Construct execution dag.	×
104	let mut execution_dag =	332✔
105	ExecutionDag::new(self.builder_dag, self.options.channel_buffer_sz)?;	332✔
106	let node_indexes = execution_dag.graph().node_identifiers().collect::<Vec<_>>();	332✔
107		332✔
108	// Start the threads.	332✔
109	let mut join_handles = HashMap::new();	332✔
110	for node_index in node_indexes {	2,027✔
111	let node = execution_dag.graph()[node_index]	1,695✔
112	.as_ref()	1,695✔
113	.expect("We created all nodes");	1,695✔
114	let node_handle = node.handle.clone();	1,695✔
115	match &node.kind {	1,695✔
116	NodeKind::Source(_, _) => {	×
117	let (source_sender_node, source_listener_node) = create_source_nodes(	342✔
118	&mut execution_dag,	342✔
119	node_index,	342✔
120	&self.options,	342✔
121	running.clone(),	342✔
122	);	342✔
123	join_handles.insert(	342✔
124	node_handle,	342✔
125	start_source(source_sender_node, source_listener_node)?,	342✔
126	);	×
127	}	×
128	NodeKind::Processor(_) => {	×
129	let processor_node = ProcessorNode::new(&mut execution_dag, node_index);	1,017✔
130	join_handles.insert(node_handle, start_processor(processor_node)?);	1,017✔
131	}	×
132	NodeKind::Sink(_) => {	×
133	let sink_node = SinkNode::new(&mut execution_dag, node_index);	336✔
134	join_handles.insert(node_handle, start_sink(sink_node)?);	336✔
135	}
136	}	×
137	}	×
138		×
139	Ok(DagExecutorJoinHandle { join_handles })	332✔
140	}	332✔
141	}	×
142		×
143	impl DagExecutorJoinHandle {	×
144	pub fn join(mut self) -> Result<(), ExecutionError> {	331✔
145	let handles: Vec<NodeHandle> = self.join_handles.iter().map(\|e\| e.0.clone()).collect();	1,692✔
146		×
147	loop {	×
148	for handle in &handles {	8,870✔
149	if let Entry::Occupied(entry) = self.join_handles.entry(handle.clone()) {	7,373✔
150	if entry.get().is_finished() {	7,157✔
151	if let Err(e) = entry.remove().join() {	1,680✔
152	panic_any(e);	5✔
153	}	1,675✔
154	}	5,477✔
155	}	216✔
156	}	×
157		×
158	if self.join_handles.is_empty() {	1,497✔
159	return Ok(());	326✔
160	}	1,171✔
161		1,171✔
162	thread::sleep(Duration::from_millis(250));	1,171✔
163	}
164	}	326✔
165	}	×
166		×
167	fn start_source(	342✔
168	source_sender: SourceSenderNode,	342✔
169	source_listener: SourceListenerNode,	342✔
170	) -> Result<JoinHandle<()>, ExecutionError> {	342✔
171	let handle = source_sender.handle().clone();	342✔
172		×
173	let _st_handle = Builder::new()	342✔
174	.name(format!("{handle}-sender"))	342✔
175	.spawn(move \|\| match source_sender.run() {	342✔
176	Ok(_) => {}	335✔
177	// Channel disconnection means the source listener has quit.	×
178	// Maybe it quit gracefully so we don't need to panic.	×
179	Err(ExecutionError::CannotSendToChannel) => {}	6✔
180	// Other errors result in panic.	×
181	Err(e) => std::panic::panic_any(e),	1✔
182	})?;	342✔
183		×
184	Ok(Builder::new()	342✔
185	.name(format!("{handle}-listener"))	342✔
186	.spawn(move \|\| {	342✔
187	if let Err(e) = source_listener.run() {	342✔
188	std::panic::panic_any(e);	5✔
189	}	337✔
190	})?)	342✔
191	}	342✔
192		×
193	fn start_processor(processor: ProcessorNode) -> Result<JoinHandle<()>, ExecutionError> {	1,017✔
194	Ok(Builder::new()	1,017✔
195	.name(processor.handle().to_string())	1,017✔
196	.spawn(move \|\| {	1,017✔
197	if let Err(e) = processor.run() {	1,017✔
198	std::panic::panic_any(e);	7✔
199	}	1,010✔
200	})?)	1,017✔
201	}	1,017✔
202		×
203	fn start_sink(sink: SinkNode) -> Result<JoinHandle<()>, ExecutionError> {	336✔
204	Ok(Builder::new().name(sink.handle().to_string()).spawn(\|\| {	336✔
205	if let Err(e) = sink.run() {	336✔
206	std::panic::panic_any(e);	5✔
207	}	331✔
208	})?)	336✔
209	}	336✔

getdozer / dozer / 4315007357

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