Compare commits

...

5 Commits

Author SHA1 Message Date
Kerollmops
353b26091e Remove a log that would log too much 2025-01-28 17:41:37 +01:00
Kerollmops
ece46b1941 Refine the env variable and the max readers 2025-01-28 17:40:50 +01:00
Kerollmops
c64b7c652a Do not create too many rayon tasks 2025-01-28 16:53:34 +01:00
Kerollmops
1c07d92014 Add more logs to see calls to the embedders 2025-01-28 16:20:33 +01:00
Kerollmops
a985b4bc29 Accept the max readers param by env var and increase it 2025-01-28 14:48:01 +01:00
5 changed files with 96 additions and 44 deletions

View File

@@ -1,5 +1,7 @@
use std::collections::BTreeMap;
use std::env::VarError;
use std::path::Path;
use std::str::FromStr;
use std::time::Duration;
use meilisearch_types::heed::{EnvClosingEvent, EnvFlags, EnvOpenOptions};
@@ -302,7 +304,15 @@ fn create_or_open_index(
) -> Result<Index> {
let mut options = EnvOpenOptions::new();
options.map_size(clamp_to_page_size(map_size));
options.max_readers(1024);
let max_readers = match std::env::var("MEILI_EXPERIMENTAL_INDEX_MAX_READERS") {
Ok(value) => u32::from_str(&value).unwrap(),
Err(VarError::NotPresent) => 1024,
Err(VarError::NotUnicode(value)) => panic!(
"Invalid unicode for the `MEILI_EXPERIMENTAL_INDEX_MAX_READERS` env var: {value:?}"
),
};
options.max_readers(max_readers);
if enable_mdb_writemap {
unsafe { options.flags(EnvFlags::WRITE_MAP) };
}

View File

@@ -1,4 +1,4 @@
use std::sync::atomic::{AtomicBool, Ordering};
use std::sync::atomic::{AtomicBool, AtomicUsize, Ordering};
use std::sync::Arc;
use rayon::{ThreadPool, ThreadPoolBuilder};
@@ -9,6 +9,8 @@ use thiserror::Error;
#[derive(Debug)]
pub struct ThreadPoolNoAbort {
thread_pool: ThreadPool,
/// The number of active operations.
active_operations: AtomicUsize,
/// Set to true if the thread pool catched a panic.
pool_catched_panic: Arc<AtomicBool>,
}
@@ -19,7 +21,9 @@ impl ThreadPoolNoAbort {
OP: FnOnce() -> R + Send,
R: Send,
{
self.active_operations.fetch_add(1, Ordering::Relaxed);
let output = self.thread_pool.install(op);
self.active_operations.fetch_sub(1, Ordering::Relaxed);
// While reseting the pool panic catcher we return an error if we catched one.
if self.pool_catched_panic.swap(false, Ordering::SeqCst) {
Err(PanicCatched)
@@ -31,6 +35,11 @@ impl ThreadPoolNoAbort {
pub fn current_num_threads(&self) -> usize {
self.thread_pool.current_num_threads()
}
/// The number of active operations.
pub fn active_operations(&self) -> usize {
self.active_operations.load(Ordering::Relaxed)
}
}
#[derive(Error, Debug)]
@@ -64,6 +73,10 @@ impl ThreadPoolNoAbortBuilder {
let catched_panic = pool_catched_panic.clone();
move |_result| catched_panic.store(true, Ordering::SeqCst)
});
Ok(ThreadPoolNoAbort { thread_pool: self.0.build()?, pool_catched_panic })
Ok(ThreadPoolNoAbort {
thread_pool: self.0.build()?,
active_operations: AtomicUsize::new(0),
pool_catched_panic,
})
}
}

View File

@@ -5,7 +5,7 @@ use rayon::slice::ParallelSlice as _;
use super::error::{EmbedError, EmbedErrorKind, NewEmbedderError, NewEmbedderErrorKind};
use super::rest::{Embedder as RestEmbedder, EmbedderOptions as RestEmbedderOptions};
use super::DistributionShift;
use super::{DistributionShift, REQUEST_PARALLELISM};
use crate::error::FaultSource;
use crate::vector::Embedding;
use crate::ThreadPoolNoAbort;
@@ -113,20 +113,30 @@ impl Embedder {
texts: &[&str],
threads: &ThreadPoolNoAbort,
) -> Result<Vec<Vec<f32>>, EmbedError> {
threads
.install(move || {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.par_chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed(chunk, None))
.collect();
if threads.active_operations() >= REQUEST_PARALLELISM {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed(chunk, None))
.collect();
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
})
.map_err(|error| EmbedError {
kind: EmbedErrorKind::PanicInThreadPool(error),
fault: FaultSource::Bug,
})?
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
} else {
threads
.install(move || {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.par_chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed(chunk, None))
.collect();
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
})
.map_err(|error| EmbedError {
kind: EmbedErrorKind::PanicInThreadPool(error),
fault: FaultSource::Bug,
})?
}
}
pub fn chunk_count_hint(&self) -> usize {

View File

@@ -6,7 +6,7 @@ use rayon::slice::ParallelSlice as _;
use super::error::{EmbedError, NewEmbedderError};
use super::rest::{Embedder as RestEmbedder, EmbedderOptions as RestEmbedderOptions};
use super::DistributionShift;
use super::{DistributionShift, REQUEST_PARALLELISM};
use crate::error::FaultSource;
use crate::vector::error::EmbedErrorKind;
use crate::vector::Embedding;
@@ -270,20 +270,29 @@ impl Embedder {
texts: &[&str],
threads: &ThreadPoolNoAbort,
) -> Result<Vec<Vec<f32>>, EmbedError> {
threads
.install(move || {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.par_chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed(chunk, None))
.collect();
if threads.active_operations() >= REQUEST_PARALLELISM {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed(chunk, None))
.collect();
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
} else {
threads
.install(move || {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.par_chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed(chunk, None))
.collect();
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
})
.map_err(|error| EmbedError {
kind: EmbedErrorKind::PanicInThreadPool(error),
fault: FaultSource::Bug,
})?
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
})
.map_err(|error| EmbedError {
kind: EmbedErrorKind::PanicInThreadPool(error),
fault: FaultSource::Bug,
})?
}
}
pub fn chunk_count_hint(&self) -> usize {

View File

@@ -203,20 +203,30 @@ impl Embedder {
texts: &[&str],
threads: &ThreadPoolNoAbort,
) -> Result<Vec<Embedding>, EmbedError> {
threads
.install(move || {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.par_chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed_ref(chunk, None))
.collect();
if threads.active_operations() >= REQUEST_PARALLELISM {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed_ref(chunk, None))
.collect();
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
})
.map_err(|error| EmbedError {
kind: EmbedErrorKind::PanicInThreadPool(error),
fault: FaultSource::Bug,
})?
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
} else {
threads
.install(move || {
let embeddings: Result<Vec<Vec<Embedding>>, _> = texts
.par_chunks(self.prompt_count_in_chunk_hint())
.map(move |chunk| self.embed_ref(chunk, None))
.collect();
let embeddings = embeddings?;
Ok(embeddings.into_iter().flatten().collect())
})
.map_err(|error| EmbedError {
kind: EmbedErrorKind::PanicInThreadPool(error),
fault: FaultSource::Bug,
})?
}
}
pub fn chunk_count_hint(&self) -> usize {