meilisearch/index-scheduler/src/index_scheduler.rs
2022-10-27 11:33:41 +02:00

520 lines
18 KiB
Rust

use crate::index_mapper::IndexMapper;
use crate::task::{Kind, KindWithContent, Status, Task, TaskView};
use crate::Result;
use file_store::FileStore;
use index::Index;
use milli::update::IndexerConfig;
use synchronoise::SignalEvent;
use std::path::PathBuf;
use std::sync::Arc;
use std::sync::RwLock;
use milli::heed::types::{OwnedType, SerdeBincode, Str};
use milli::heed::{self, Database, Env};
use milli::{RoaringBitmapCodec, BEU32};
use roaring::RoaringBitmap;
use serde::Deserialize;
const DEFAULT_LIMIT: fn() -> u32 = || 20;
#[derive(Debug, Clone, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct Query {
#[serde(default = "DEFAULT_LIMIT")]
limit: u32,
from: Option<u32>,
status: Option<Vec<Status>>,
#[serde(rename = "type")]
kind: Option<Vec<Kind>>,
index_uid: Option<Vec<String>>,
}
pub mod db_name {
pub const ALL_TASKS: &str = "all-tasks";
pub const STATUS: &str = "status";
pub const KIND: &str = "kind";
pub const INDEX_TASKS: &str = "index-tasks";
pub const INDEX_MAPPING: &str = "index-mapping";
}
/// This module is responsible for two things;
/// 1. Resolve the name of the indexes.
/// 2. Schedule the tasks.
#[derive(Clone)]
pub struct IndexScheduler {
/// The list of tasks currently processing.
pub(crate) processing_tasks: Arc<RwLock<RoaringBitmap>>,
pub(crate) file_store: FileStore,
/// The LMDB environment which the DBs are associated with.
pub(crate) env: Env,
// The main database, it contains all the tasks accessible by their Id.
pub(crate) all_tasks: Database<OwnedType<BEU32>, SerdeBincode<Task>>,
/// All the tasks ids grouped by their status.
pub(crate) status: Database<SerdeBincode<Status>, RoaringBitmapCodec>,
/// All the tasks ids grouped by their kind.
pub(crate) kind: Database<SerdeBincode<Kind>, RoaringBitmapCodec>,
/// Store the tasks associated to an index.
pub(crate) index_tasks: Database<Str, RoaringBitmapCodec>,
/// In charge of creating, opening, storing and returning indexes.
pub(crate) index_mapper: IndexMapper,
// set to true when there is work to do.
pub(crate) wake_up: Arc<SignalEvent>,
}
impl IndexScheduler {
pub fn new(
db_path: PathBuf,
update_file_path: PathBuf,
indexes_path: PathBuf,
index_size: usize,
indexer_config: IndexerConfig,
) -> Result<Self> {
std::fs::create_dir_all(&db_path)?;
std::fs::create_dir_all(&update_file_path)?;
std::fs::create_dir_all(&indexes_path)?;
let mut options = heed::EnvOpenOptions::new();
options.max_dbs(6);
let env = options.open(db_path)?;
// we want to start the loop right away in case meilisearch was ctrl+Ced while processing things
let wake_up = SignalEvent::auto(true);
Ok(Self {
// by default there is no processing tasks
processing_tasks: Arc::default(),
file_store: FileStore::new(update_file_path)?,
all_tasks: env.create_database(Some(db_name::ALL_TASKS))?,
status: env.create_database(Some(db_name::STATUS))?,
kind: env.create_database(Some(db_name::KIND))?,
index_tasks: env.create_database(Some(db_name::INDEX_TASKS))?,
index_mapper: IndexMapper::new(&env, indexes_path, index_size, indexer_config)?,
env,
wake_up: Arc::new(wake_up),
})
}
/// Return the index corresponding to the name. If it wasn't opened before
/// it'll be opened. But if it doesn't exist on disk it'll throw an
/// `IndexNotFound` error.
pub fn index(&self, name: &str) -> Result<Index> {
let rtxn = self.env.read_txn()?;
self.index_mapper.index(&rtxn, name)
}
/// Returns the tasks corresponding to the query.
pub fn get_tasks(&self, query: Query) -> Result<Vec<TaskView>> {
let rtxn = self.env.read_txn()?;
let last_task_id = match self.last_task_id(&rtxn)? {
Some(tid) => query.from.map(|from| from.min(tid)).unwrap_or(tid),
None => return Ok(Vec::new()),
};
// This is the list of all the tasks.
let mut tasks = RoaringBitmap::from_iter(0..last_task_id);
if let Some(status) = query.status {
let mut status_tasks = RoaringBitmap::new();
for status in status {
status_tasks |= self.get_status(&rtxn, status)?;
}
tasks &= status_tasks;
}
if let Some(kind) = query.kind {
let mut kind_tasks = RoaringBitmap::new();
for kind in kind {
kind_tasks |= self.get_kind(&rtxn, kind)?;
}
tasks &= kind_tasks;
}
if let Some(index) = query.index_uid {
let mut index_tasks = RoaringBitmap::new();
for index in index {
index_tasks |= self.get_index(&rtxn, &index)?;
}
tasks &= index_tasks;
}
let tasks =
self.get_existing_tasks(&rtxn, tasks.into_iter().rev().take(query.limit as usize))?;
Ok(tasks.into_iter().map(|task| task.as_task_view()).collect())
}
/// Register a new task in the scheduler. If it fails and data was associated with the task
/// it tries to delete the file.
pub fn register(&self, task: KindWithContent) -> Result<TaskView> {
let mut wtxn = self.env.write_txn()?;
let task = Task {
uid: self.next_task_id(&wtxn)?,
enqueued_at: time::OffsetDateTime::now_utc(),
started_at: None,
finished_at: None,
error: None,
details: None,
status: Status::Enqueued,
kind: task,
};
self.all_tasks
.append(&mut wtxn, &BEU32::new(task.uid), &task)?;
if let Some(indexes) = task.indexes() {
for index in indexes {
self.update_index(&mut wtxn, index, |bitmap| drop(bitmap.insert(task.uid)))?;
}
}
self.update_status(&mut wtxn, Status::Enqueued, |bitmap| {
bitmap.insert(task.uid);
})?;
self.update_kind(&mut wtxn, task.kind.as_kind(), |bitmap| {
(bitmap.insert(task.uid));
})?;
// we persist the file in last to be sure everything before was applied successfuly
task.persist()?;
match wtxn.commit() {
Ok(()) => (),
e @ Err(_) => {
task.remove_data()?;
e?;
}
}
self.notify();
Ok(task.as_task_view())
}
/// This worker function must be run in a different thread and must be run only once.
fn run(&self) -> ! {
loop {
self.wake_up.wait();
self.tick()
}
}
/// Create and execute and store the result of one batch of registered tasks.
fn tick(&self) {
let mut wtxn = match self.env.write_txn() {
Ok(wtxn) => wtxn,
Err(e) => {
log::error!("{}", e);
return;
}
};
let batch = match self.create_next_batch(&wtxn) {
Ok(Some(batch)) => batch,
Ok(None) => return,
Err(e) => {
log::error!("{}", e);
return;
}
};
// 1. store the starting date with the bitmap of processing tasks
// 2. update the tasks with a starting date *but* do not write anything on disk
// 3. process the tasks
let _res = self.process_batch(&mut wtxn, batch);
// 4. store the updated tasks on disk
// TODO: TAMO: do this later
// must delete the file on disk
// in case of error, must update the tasks with the error
// in case of « success » we must update all the task on disk
// self.handle_batch_result(res);
match wtxn.commit() {
Ok(()) => log::info!("A batch of tasks was successfully completed."),
Err(e) => {
log::error!("{}", e);
return;
}
}
}
#[cfg(truc)]
fn process_batch(&self, wtxn: &mut RwTxn, batch: &mut Batch) -> Result<()> {
match batch {
Batch::One(task) => match &task.kind {
KindWithContent::ClearAllDocuments { index_name } => {
self.index(&index_name)?.clear_documents()?;
}
KindWithContent::RenameIndex {
index_name: _,
new_name,
} => {
if self.available_index.get(wtxn, &new_name)?.unwrap_or(false) {
return Err(Error::IndexAlreadyExists(new_name.to_string()));
}
todo!("wait for @guigui insight");
}
KindWithContent::CreateIndex {
index_name,
primary_key,
} => {
if self
.available_index
.get(wtxn, &index_name)?
.unwrap_or(false)
{
return Err(Error::IndexAlreadyExists(index_name.to_string()));
}
self.available_index.put(wtxn, &index_name, &true)?;
// TODO: TAMO: give real info to the index
let index = Index::open(
index_name.to_string(),
index_name.to_string(),
100_000_000,
Arc::default(),
)?;
if let Some(primary_key) = primary_key {
index.update_primary_key(primary_key.to_string())?;
}
self.index_map
.write()
.map_err(|_| Error::CorruptedTaskQueue)?
.insert(index_name.to_string(), index.clone());
}
KindWithContent::DeleteIndex { index_name } => {
if !self.available_index.delete(wtxn, &index_name)? {
return Err(Error::IndexNotFound(index_name.to_string()));
}
if let Some(index) = self
.index_map
.write()
.map_err(|_| Error::CorruptedTaskQueue)?
.remove(index_name)
{
index.delete()?;
} else {
// TODO: TAMO: fix the path
std::fs::remove_file(index_name)?;
}
}
KindWithContent::SwapIndex { lhs, rhs } => {
if !self.available_index.get(wtxn, &lhs)?.unwrap_or(false) {
return Err(Error::IndexNotFound(lhs.to_string()));
}
if !self.available_index.get(wtxn, &rhs)?.unwrap_or(false) {
return Err(Error::IndexNotFound(rhs.to_string()));
}
let lhs_bitmap = self.index_tasks.get(wtxn, lhs)?;
let rhs_bitmap = self.index_tasks.get(wtxn, rhs)?;
// the bitmap are lazily created and thus may not exists.
if let Some(bitmap) = rhs_bitmap {
self.index_tasks.put(wtxn, lhs, &bitmap)?;
}
if let Some(bitmap) = lhs_bitmap {
self.index_tasks.put(wtxn, rhs, &bitmap)?;
}
let mut index_map = self
.index_map
.write()
.map_err(|_| Error::CorruptedTaskQueue)?;
let lhs_index = index_map.remove(lhs).unwrap();
let rhs_index = index_map.remove(rhs).unwrap();
index_map.insert(lhs.to_string(), rhs_index);
index_map.insert(rhs.to_string(), lhs_index);
}
_ => unreachable!(),
},
Batch::Cancel(_) => todo!(),
Batch::Snapshot(_) => todo!(),
Batch::Dump(_) => todo!(),
Batch::Contiguous { tasks, kind } => {
// it's safe because you can't batch 0 contiguous tasks.
let first_task = &tasks[0];
// and the two kind of tasks we batch MUST have ONE index name.
let index_name = first_task.indexes().unwrap()[0];
let index = self.index(index_name)?;
match kind {
Kind::DocumentAddition => {
let content_files = tasks.iter().map(|task| match &task.kind {
KindWithContent::DocumentAddition { content_file, .. } => {
content_file.clone()
}
k => unreachable!(
"Internal error, `{:?}` is not supposed to be reachable here",
k.as_kind()
),
});
let results = index.update_documents(
IndexDocumentsMethod::UpdateDocuments,
None,
self.file_store.clone(),
content_files,
)?;
for (task, result) in tasks.iter_mut().zip(results) {
task.finished_at = Some(OffsetDateTime::now_utc());
match result {
Ok(_) => task.status = Status::Succeeded,
Err(_) => task.status = Status::Succeeded,
}
}
}
Kind::DocumentDeletion => {
let ids: Vec<_> = tasks
.iter()
.flat_map(|task| match &task.kind {
KindWithContent::DocumentDeletion { documents_ids, .. } => {
documents_ids.clone()
}
k => unreachable!(
"Internal error, `{:?}` is not supposed to be reachable here",
k.as_kind()
),
})
.collect();
let result = index.delete_documents(&ids);
for task in tasks.iter_mut() {
task.finished_at = Some(OffsetDateTime::now_utc());
match result {
Ok(_) => task.status = Status::Succeeded,
Err(_) => task.status = Status::Succeeded,
}
}
}
_ => unreachable!(),
}
}
Batch::Empty => todo!(),
}
Ok(())
}
/// Notify the scheduler there is or may be work to do.
pub fn notify(&self) {
self.wake_up.signal()
}
}
#[cfg(test)]
mod tests {
use big_s::S;
use insta::assert_debug_snapshot;
use tempfile::TempDir;
use uuid::Uuid;
use crate::assert_smol_debug_snapshot;
use super::*;
fn new() -> IndexScheduler {
let dir = TempDir::new().unwrap();
IndexScheduler::new(
dir.path().join("db_path"),
dir.path().join("file_store"),
dir.path().join("indexes"),
100_000_000,
IndexerConfig::default(),
)
.unwrap()
}
#[test]
fn simple_new() {
new();
}
#[test]
fn register() {
let index_scheduler = new();
let kinds = [
KindWithContent::IndexCreation {
index_uid: S("catto"),
primary_key: Some(S("mouse")),
},
KindWithContent::DocumentAddition {
index_uid: S("catto"),
primary_key: None,
content_file: Uuid::new_v4(),
documents_count: 12,
allow_index_creation: true,
},
KindWithContent::CancelTask { tasks: vec![0, 1] },
KindWithContent::DocumentAddition {
index_uid: S("catto"),
primary_key: None,
content_file: Uuid::new_v4(),
documents_count: 50,
allow_index_creation: true,
},
KindWithContent::DocumentAddition {
index_uid: S("doggo"),
primary_key: Some(S("bone")),
content_file: Uuid::new_v4(),
documents_count: 5000,
allow_index_creation: true,
},
];
let mut inserted_tasks = Vec::new();
for (idx, kind) in kinds.into_iter().enumerate() {
let k = kind.as_kind();
let task = index_scheduler.register(kind).unwrap();
assert_eq!(task.uid, idx as u32);
assert_eq!(task.status, Status::Enqueued);
assert_eq!(task.kind, k);
inserted_tasks.push(task);
}
let rtxn = index_scheduler.env.read_txn().unwrap();
let mut all_tasks = Vec::new();
for ret in index_scheduler.all_tasks.iter(&rtxn).unwrap() {
all_tasks.push(ret.unwrap().0);
}
// we can't assert on the content of the tasks because there is the date and uuid that changes everytime.
assert_smol_debug_snapshot!(all_tasks, @"[U32(0), U32(1), U32(2), U32(3), U32(4)]");
let mut status = Vec::new();
for ret in index_scheduler.status.iter(&rtxn).unwrap() {
status.push(ret.unwrap());
}
assert_smol_debug_snapshot!(status, @"[(Enqueued, RoaringBitmap<[0, 1, 2, 3, 4]>)]");
let mut kind = Vec::new();
for ret in index_scheduler.kind.iter(&rtxn).unwrap() {
kind.push(ret.unwrap());
}
assert_smol_debug_snapshot!(kind, @"[(DocumentAddition, RoaringBitmap<[1, 3, 4]>), (IndexCreation, RoaringBitmap<[0]>), (CancelTask, RoaringBitmap<[2]>)]");
let mut index_tasks = Vec::new();
for ret in index_scheduler.index_tasks.iter(&rtxn).unwrap() {
index_tasks.push(ret.unwrap());
}
assert_smol_debug_snapshot!(index_tasks, @r###"[("catto", RoaringBitmap<[0, 1, 3]>), ("doggo", RoaringBitmap<[4]>)]"###);
}
}