[refactor-block] refactor DataBlock and DataBlockPath

This commit is contained in:
Alex Auvolat 2024-02-23 12:11:14 +01:00
parent 07c7895948
commit cd1069c1d4
No known key found for this signature in database
GPG key ID: 0E496D15096376BE
3 changed files with 88 additions and 78 deletions

View file

@ -13,77 +13,99 @@ pub enum DataBlockHeader {
Compressed, Compressed,
} }
/// A possibly compressed block of data #[derive(Debug)]
pub enum DataBlock { pub struct DataBlockElem<T> {
/// Uncompressed data header: DataBlockHeader,
Plain(Bytes), elem: T,
/// Data compressed with zstd
Compressed(Bytes),
} }
#[derive(Debug)] /// A possibly compressed block of data
pub enum DataBlockPath { pub type DataBlock = DataBlockElem<Bytes>;
/// Uncompressed data fail
Plain(PathBuf), /// A path to a possibly compressed block of data
/// Compressed data fail pub type DataBlockPath = DataBlockElem<PathBuf>;
Compressed(PathBuf),
impl DataBlockHeader {
pub fn is_compressed(&self) -> bool {
matches!(self, DataBlockHeader::Compressed)
}
}
impl<T> DataBlockElem<T> {
pub fn from_parts(header: DataBlockHeader, elem: T) -> Self {
Self { header, elem }
}
pub fn plain(elem: T) -> Self {
Self {
header: DataBlockHeader::Plain,
elem,
}
}
pub fn compressed(elem: T) -> Self {
Self {
header: DataBlockHeader::Compressed,
elem,
}
}
pub fn into_parts(self) -> (DataBlockHeader, T) {
(self.header, self.elem)
}
pub fn as_parts_ref(&self) -> (DataBlockHeader, &T) {
(self.header, &self.elem)
}
/// Query whether this block is compressed
pub fn is_compressed(&self) -> bool {
self.header.is_compressed()
}
} }
impl DataBlock { impl DataBlock {
/// Query whether this block is compressed
pub fn is_compressed(&self) -> bool {
matches!(self, DataBlock::Compressed(_))
}
/// Get the inner, possibly compressed buffer. You should probably use [`DataBlock::verify_get`] /// Get the inner, possibly compressed buffer. You should probably use [`DataBlock::verify_get`]
/// instead /// instead
pub fn inner_buffer(&self) -> &[u8] { pub fn inner_buffer(&self) -> &[u8] {
use DataBlock::*; &self.elem
let (Plain(ref res) | Compressed(ref res)) = self;
res
} }
/// Verify data integrity. Does not return the buffer content. /// Verify data integrity. Does not return the buffer content.
pub fn verify(&self, hash: Hash) -> Result<(), Error> { pub fn verify(&self, hash: Hash) -> Result<(), Error> {
match self { match self.header {
DataBlock::Plain(data) => { DataBlockHeader::Plain => {
if blake2sum(data) == hash { if blake2sum(&self.elem) == hash {
Ok(()) Ok(())
} else { } else {
Err(Error::CorruptData(hash)) Err(Error::CorruptData(hash))
} }
} }
DataBlock::Compressed(data) => zstd::stream::copy_decode(&data[..], std::io::sink()) DataBlockHeader::Compressed => {
.map_err(|_| Error::CorruptData(hash)), zstd::stream::copy_decode(&self.elem[..], std::io::sink())
.map_err(|_| Error::CorruptData(hash))
}
} }
} }
pub async fn from_buffer(data: Bytes, level: Option<i32>) -> DataBlock { pub async fn from_buffer(data: Bytes, level: Option<i32>) -> DataBlock {
tokio::task::spawn_blocking(move || { tokio::task::spawn_blocking(move || {
if let Some(level) = level { if let Some(level) = level {
if let Ok(data) = zstd_encode(&data[..], level) { if let Ok(data_compressed) = zstd_encode(&data[..], level) {
return DataBlock::Compressed(data.into()); return DataBlock {
header: DataBlockHeader::Compressed,
elem: data_compressed.into(),
};
} }
} }
DataBlock::Plain(data) DataBlock {
header: DataBlockHeader::Plain,
elem: data.into(),
}
}) })
.await .await
.unwrap() .unwrap()
} }
pub fn into_parts(self) -> (DataBlockHeader, Bytes) {
match self {
DataBlock::Plain(data) => (DataBlockHeader::Plain, data),
DataBlock::Compressed(data) => (DataBlockHeader::Compressed, data),
}
}
pub fn from_parts(h: DataBlockHeader, bytes: Bytes) -> Self {
match h {
DataBlockHeader::Plain => DataBlock::Plain(bytes),
DataBlockHeader::Compressed => DataBlock::Compressed(bytes),
}
}
} }
fn zstd_encode<R: std::io::Read>(mut source: R, level: i32) -> std::io::Result<Vec<u8>> { fn zstd_encode<R: std::io::Read>(mut source: R, level: i32) -> std::io::Result<Vec<u8>> {

View file

@ -547,10 +547,7 @@ impl BlockManager {
hash: &Hash, hash: &Hash,
block_path: &DataBlockPath, block_path: &DataBlockPath,
) -> Result<DataBlock, Error> { ) -> Result<DataBlock, Error> {
let (path, compressed) = match block_path { let (header, path) = block_path.as_parts_ref();
DataBlockPath::Plain(p) => (p, false),
DataBlockPath::Compressed(p) => (p, true),
};
let mut f = fs::File::open(&path).await?; let mut f = fs::File::open(&path).await?;
let mut data = vec![]; let mut data = vec![];
@ -558,11 +555,7 @@ impl BlockManager {
self.metrics.bytes_read.add(data.len() as u64); self.metrics.bytes_read.add(data.len() as u64);
drop(f); drop(f);
let data = if compressed { let data = DataBlock::from_parts(header, data.into());
DataBlock::Compressed(data.into())
} else {
DataBlock::Plain(data.into())
};
if data.verify(*hash).is_err() { if data.verify(*hash).is_err() {
self.metrics.corruption_counter.add(1); self.metrics.corruption_counter.add(1);
@ -615,20 +608,20 @@ impl BlockManager {
// first and then a compressed one (as compression may have been // first and then a compressed one (as compression may have been
// previously enabled). // previously enabled).
if fs::metadata(&path).await.is_ok() { if fs::metadata(&path).await.is_ok() {
return Some(DataBlockPath::Plain(path)); return Some(DataBlockPath::plain(path));
} }
path.set_extension("zst"); path.set_extension("zst");
if fs::metadata(&path).await.is_ok() { if fs::metadata(&path).await.is_ok() {
return Some(DataBlockPath::Compressed(path)); return Some(DataBlockPath::compressed(path));
} }
} else { } else {
path.set_extension("zst"); path.set_extension("zst");
if fs::metadata(&path).await.is_ok() { if fs::metadata(&path).await.is_ok() {
return Some(DataBlockPath::Compressed(path)); return Some(DataBlockPath::compressed(path));
} }
path.set_extension(""); path.set_extension("");
if fs::metadata(&path).await.is_ok() { if fs::metadata(&path).await.is_ok() {
return Some(DataBlockPath::Plain(path)); return Some(DataBlockPath::plain(path));
} }
} }
} }
@ -709,24 +702,25 @@ impl BlockManagerLocked {
tgt_path.set_extension("zst"); tgt_path.set_extension("zst");
} }
let to_delete = match (existing_path, compressed) { let existing_info = existing_path.map(|x| x.into_parts());
let to_delete = match (existing_info, compressed) {
// If the block is stored in the wrong directory, // If the block is stored in the wrong directory,
// write it again at the correct path and delete the old path // write it again at the correct path and delete the old path
(Some(DataBlockPath::Plain(p)), false) if p != tgt_path => Some(p), (Some((DataBlockHeader::Plain, p)), false) if p != tgt_path => Some(p),
(Some(DataBlockPath::Compressed(p)), true) if p != tgt_path => Some(p), (Some((DataBlockHeader::Compressed, p)), true) if p != tgt_path => Some(p),
// If the block is already stored not compressed but we have a compressed // If the block is already stored not compressed but we have a compressed
// copy, write the compressed copy and delete the uncompressed one // copy, write the compressed copy and delete the uncompressed one
(Some(DataBlockPath::Plain(plain_path)), true) => Some(plain_path), (Some((DataBlockHeader::Plain, plain_path)), true) => Some(plain_path),
// If the block is already stored compressed, // If the block is already stored compressed,
// keep the stored copy, we have nothing to do // keep the stored copy, we have nothing to do
(Some(DataBlockPath::Compressed(_)), _) => return Ok(()), (Some((DataBlockHeader::Compressed, _)), _) => return Ok(()),
// If the block is already stored not compressed, // If the block is already stored not compressed,
// and we don't have a compressed copy either, // and we don't have a compressed copy either,
// keep the stored copy, we have nothing to do // keep the stored copy, we have nothing to do
(Some(DataBlockPath::Plain(_)), false) => return Ok(()), (Some((DataBlockHeader::Plain, _)), false) => return Ok(()),
// If the block isn't stored already, just store what is given to us // If the block isn't stored already, just store what is given to us
(None, _) => None, (None, _) => None,
@ -778,18 +772,14 @@ impl BlockManagerLocked {
} }
async fn move_block_to_corrupted(&self, block_path: &DataBlockPath) -> Result<(), Error> { async fn move_block_to_corrupted(&self, block_path: &DataBlockPath) -> Result<(), Error> {
let (path, path2) = match block_path { let (header, path) = block_path.as_parts_ref();
DataBlockPath::Plain(p) => {
let mut p2 = p.clone(); let mut path2 = path.clone();
p2.set_extension("corrupted"); if header.is_compressed() {
(p, p2) path2.set_extension("zst.corrupted");
} } else {
DataBlockPath::Compressed(p) => { path2.set_extension("corrupted");
let mut p2 = p.clone(); }
p2.set_extension("zst.corrupted");
(p, p2)
}
};
fs::rename(path, path2).await?; fs::rename(path, path2).await?;
Ok(()) Ok(())
@ -799,9 +789,7 @@ impl BlockManagerLocked {
let rc = mgr.rc.get_block_rc(hash)?; let rc = mgr.rc.get_block_rc(hash)?;
if rc.is_deletable() { if rc.is_deletable() {
while let Some(path) = mgr.find_block(hash).await { while let Some(path) = mgr.find_block(hash).await {
let path = match path { let (_header, path) = path.as_parts_ref();
DataBlockPath::Plain(p) | DataBlockPath::Compressed(p) => p,
};
fs::remove_file(path).await?; fs::remove_file(path).await?;
mgr.metrics.delete_counter.add(1); mgr.metrics.delete_counter.add(1);
} }

View file

@ -584,8 +584,8 @@ impl Worker for RebalanceWorker {
let prim_loc = self.manager.data_layout.load().primary_block_dir(&hash); let prim_loc = self.manager.data_layout.load().primary_block_dir(&hash);
if path.ancestors().all(|x| x != prim_loc) { if path.ancestors().all(|x| x != prim_loc) {
let block_path = match path.extension() { let block_path = match path.extension() {
None => DataBlockPath::Plain(path.clone()), None => DataBlockPath::plain(path.clone()),
Some(x) if x.to_str() == Some("zst") => DataBlockPath::Compressed(path.clone()), Some(x) if x.to_str() == Some("zst") => DataBlockPath::compressed(path.clone()),
_ => { _ => {
warn!("not rebalancing file: {}", path.to_string_lossy()); warn!("not rebalancing file: {}", path.to_string_lossy());
return Ok(WorkerState::Busy); return Ok(WorkerState::Busy);