tyler/proxmox-backup
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

moving more code to pbs-datastore

Browse Source 
prune and fixed/dynamic index

Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
This commit is contained in:
Wolfgang Bumiller 2021-07-08 10:07:01 +02:00
parent e64f38cb6b
commit 2f02e431b0
7 changed files with 549 additions and 666 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

508
pbs-datastore/src/dynamic_index.rs Normal file
View File

@ -0,0 +1,508 @@
use std::fs::File;
use std::io::{BufWriter, Seek, SeekFrom, Write};
use std::os::unix::io::AsRawFd;
use std::path::{Path, PathBuf};
use std::sync::Arc;
use anyhow::{bail, format_err, Error};
use proxmox::tools::io::ReadExt;
use proxmox::tools::uuid::Uuid;
use proxmox::tools::mmap::Mmap;
use pbs_tools::process_locker::ProcessLockSharedGuard;
use crate::Chunker;
use crate::chunk_stat::ChunkStat;
use crate::chunk_store::ChunkStore;
use crate::data_blob::{DataBlob, DataChunkBuilder};
use crate::file_formats;
use crate::index::{IndexFile, ChunkReadInfo};
/// Header format definition for dynamic index files (`.dixd`)
#[repr(C)]
pub struct DynamicIndexHeader {
pub magic: [u8; 8],
pub uuid: [u8; 16],
pub ctime: i64,
/// Sha256 over the index ``SHA256(offset1||digest1||offset2||digest2||...)``
pub index_csum: [u8; 32],
reserved: [u8; 4032], // overall size is one page (4096 bytes)
}
proxmox::static_assert_size!(DynamicIndexHeader, 4096);
// TODO: Once non-Copy unions are stabilized, use:
// union DynamicIndexHeader {
// reserved: [u8; 4096],
// pub data: DynamicIndexHeaderData,
// }
impl DynamicIndexHeader {
/// Convenience method to allocate a zero-initialized header struct.
pub fn zeroed() -> Box<Self> {
unsafe {
Box::from_raw(std::alloc::alloc_zeroed(std::alloc::Layout::new::<Self>()) as *mut Self)
}
}
pub fn as_bytes(&self) -> &[u8] {
unsafe {
std::slice::from_raw_parts(
self as *const Self as *const u8,
std::mem::size_of::<Self>(),
)
}
}
}
#[derive(Clone, Debug)]
#[repr(C)]
pub struct DynamicEntry {
end_le: u64,
digest: [u8; 32],
}
impl DynamicEntry {
#[inline]
pub fn end(&self) -> u64 {
u64::from_le(self.end_le)
}
}
pub struct DynamicIndexReader {
_file: File,
pub size: usize,
index: Mmap<DynamicEntry>,
pub uuid: [u8; 16],
pub ctime: i64,
pub index_csum: [u8; 32],
}
impl DynamicIndexReader {
pub fn open(path: &Path) -> Result<Self, Error> {
File::open(path)
.map_err(Error::from)
.and_then(Self::new)
.map_err(|err| format_err!("Unable to open dynamic index {:?} - {}", path, err))
}
pub fn index(&self) -> &[DynamicEntry] {
&self.index
}
pub fn new(mut file: std::fs::File) -> Result<Self, Error> {
// FIXME: This is NOT OUR job! Check the callers of this method and remove this!
file.seek(SeekFrom::Start(0))?;
let header_size = std::mem::size_of::<DynamicIndexHeader>();
let rawfd = file.as_raw_fd();
let stat = match nix::sys::stat::fstat(rawfd) {
Ok(stat) => stat,
Err(err) => bail!("fstat failed - {}", err),
};
let size = stat.st_size as usize;
if size < header_size {
bail!("index too small ({})", stat.st_size);
}
let header: Box<DynamicIndexHeader> = unsafe { file.read_host_value_boxed()? };
if header.magic != file_formats::DYNAMIC_SIZED_CHUNK_INDEX_1_0 {
bail!("got unknown magic number");
}
let ctime = proxmox::tools::time::epoch_i64();
let index_size = stat.st_size as usize - header_size;
let index_count = index_size / 40;
if index_count * 40 != index_size {
bail!("got unexpected file size");
}
let index = unsafe {
Mmap::map_fd(
rawfd,
header_size as u64,
index_count,
nix::sys::mman::ProtFlags::PROT_READ,
nix::sys::mman::MapFlags::MAP_PRIVATE,
)?
};
Ok(Self {
_file: file,
size,
index,
ctime,
uuid: header.uuid,
index_csum: header.index_csum,
})
}
#[inline]
#[allow(clippy::cast_ptr_alignment)]
pub fn chunk_end(&self, pos: usize) -> u64 {
if pos >= self.index.len() {
panic!("chunk index out of range");
}
self.index[pos].end()
}
#[inline]
fn chunk_digest(&self, pos: usize) -> &[u8; 32] {
if pos >= self.index.len() {
panic!("chunk index out of range");
}
&self.index[pos].digest
}
pub fn binary_search(
&self,
start_idx: usize,
start: u64,
end_idx: usize,
end: u64,
offset: u64,
) -> Result<usize, Error> {
if (offset >= end) || (offset < start) {
bail!("offset out of range");
}
if end_idx == start_idx {
return Ok(start_idx); // found
}
let middle_idx = (start_idx + end_idx) / 2;
let middle_end = self.chunk_end(middle_idx);
if offset < middle_end {
self.binary_search(start_idx, start, middle_idx, middle_end, offset)
} else {
self.binary_search(middle_idx + 1, middle_end, end_idx, end, offset)
}
}
}
impl IndexFile for DynamicIndexReader {
fn index_count(&self) -> usize {
self.index.len()
}
fn index_digest(&self, pos: usize) -> Option<&[u8; 32]> {
if pos >= self.index.len() {
None
} else {
Some(unsafe { &*(self.chunk_digest(pos).as_ptr() as *const [u8; 32]) })
}
}
fn index_bytes(&self) -> u64 {
if self.index.is_empty() {
0
} else {
self.chunk_end(self.index.len() - 1)
}
}
fn compute_csum(&self) -> ([u8; 32], u64) {
let mut csum = openssl::sha::Sha256::new();
let mut chunk_end = 0;
for pos in 0..self.index_count() {
let info = self.chunk_info(pos).unwrap();
chunk_end = info.range.end;
csum.update(&chunk_end.to_le_bytes());
csum.update(&info.digest);
}
let csum = csum.finish();
(csum, chunk_end)
}
fn chunk_info(&self, pos: usize) -> Option<ChunkReadInfo> {
if pos >= self.index.len() {
return None;
}
let start = if pos == 0 { 0 } else { self.index[pos - 1].end() };
let end = self.index[pos].end();
Some(ChunkReadInfo {
range: start..end,
digest: self.index[pos].digest,
})
}
fn index_ctime(&self) -> i64 {
self.ctime
}
fn index_size(&self) -> usize {
self.size as usize
}
fn chunk_from_offset(&self, offset: u64) -> Option<(usize, u64)> {
let end_idx = self.index.len() - 1;
let end = self.chunk_end(end_idx);
let found_idx = self.binary_search(0, 0, end_idx, end, offset);
let found_idx = match found_idx {
Ok(i) => i,
Err(_) => return None
};
let found_start = if found_idx == 0 {
0
} else {
self.chunk_end(found_idx - 1)
};
Some((found_idx, offset - found_start))
}
}
/// Create dynamic index files (`.dixd`)
pub struct DynamicIndexWriter {
store: Arc<ChunkStore>,
_lock: ProcessLockSharedGuard,
writer: BufWriter<File>,
closed: bool,
filename: PathBuf,
tmp_filename: PathBuf,
csum: Option<openssl::sha::Sha256>,
pub uuid: [u8; 16],
pub ctime: i64,
}
impl Drop for DynamicIndexWriter {
fn drop(&mut self) {
let _ = std::fs::remove_file(&self.tmp_filename); // ignore errors
}
}
impl DynamicIndexWriter {
pub fn create(store: Arc<ChunkStore>, path: &Path) -> Result<Self, Error> {
let shared_lock = store.try_shared_lock()?;
let full_path = store.relative_path(path);
let mut tmp_path = full_path.clone();
tmp_path.set_extension("tmp_didx");
let file = std::fs::OpenOptions::new()
.create(true)
.truncate(true)
.read(true)
.write(true)
.open(&tmp_path)?;
let mut writer = BufWriter::with_capacity(1024 * 1024, file);
let ctime = proxmox::tools::time::epoch_i64();
let uuid = Uuid::generate();
let mut header = DynamicIndexHeader::zeroed();
header.magic = file_formats::DYNAMIC_SIZED_CHUNK_INDEX_1_0;
header.ctime = i64::to_le(ctime);
header.uuid = *uuid.as_bytes();
// header.index_csum = [0u8; 32];
writer.write_all(header.as_bytes())?;
let csum = Some(openssl::sha::Sha256::new());
Ok(Self {
store,
_lock: shared_lock,
writer,
closed: false,
filename: full_path,
tmp_filename: tmp_path,
ctime,
uuid: *uuid.as_bytes(),
csum,
})
}
// fixme: use add_chunk instead?
pub fn insert_chunk(&self, chunk: &DataBlob, digest: &[u8; 32]) -> Result<(bool, u64), Error> {
self.store.insert_chunk(chunk, digest)
}
pub fn close(&mut self) -> Result<[u8; 32], Error> {
if self.closed {
bail!(
"cannot close already closed archive index file {:?}",
self.filename
);
}
self.closed = true;
self.writer.flush()?;
let csum_offset = proxmox::offsetof!(DynamicIndexHeader, index_csum);
self.writer.seek(SeekFrom::Start(csum_offset as u64))?;
let csum = self.csum.take().unwrap();
let index_csum = csum.finish();
self.writer.write_all(&index_csum)?;
self.writer.flush()?;
if let Err(err) = std::fs::rename(&self.tmp_filename, &self.filename) {
bail!("Atomic rename file {:?} failed - {}", self.filename, err);
}
Ok(index_csum)
}
// fixme: rename to add_digest
pub fn add_chunk(&mut self, offset: u64, digest: &[u8; 32]) -> Result<(), Error> {
if self.closed {
bail!(
"cannot write to closed dynamic index file {:?}",
self.filename
);
}
let offset_le: &[u8; 8] = unsafe { &std::mem::transmute::<u64, [u8; 8]>(offset.to_le()) };
if let Some(ref mut csum) = self.csum {
csum.update(offset_le);
csum.update(digest);
}
self.writer.write_all(offset_le)?;
self.writer.write_all(digest)?;
Ok(())
}
}
/// Writer which splits a binary stream into dynamic sized chunks
///
/// And store the resulting chunk list into the index file.
pub struct DynamicChunkWriter {
index: DynamicIndexWriter,
closed: bool,
chunker: Chunker,
stat: ChunkStat,
chunk_offset: usize,
last_chunk: usize,
chunk_buffer: Vec<u8>,
}
impl DynamicChunkWriter {
pub fn new(index: DynamicIndexWriter, chunk_size: usize) -> Self {
Self {
index,
closed: false,
chunker: Chunker::new(chunk_size),
stat: ChunkStat::new(0),
chunk_offset: 0,
last_chunk: 0,
chunk_buffer: Vec::with_capacity(chunk_size * 4),
}
}
pub fn stat(&self) -> &ChunkStat {
&self.stat
}
pub fn close(&mut self) -> Result<(), Error> {
if self.closed {
return Ok(());
}
self.closed = true;
self.write_chunk_buffer()?;
self.index.close()?;
self.stat.size = self.chunk_offset as u64;
// add size of index file
self.stat.size +=
(self.stat.chunk_count * 40 + std::mem::size_of::<DynamicIndexHeader>()) as u64;
Ok(())
}
fn write_chunk_buffer(&mut self) -> Result<(), Error> {
let chunk_size = self.chunk_buffer.len();
if chunk_size == 0 {
return Ok(());
}
let expected_chunk_size = self.chunk_offset - self.last_chunk;
if expected_chunk_size != self.chunk_buffer.len() {
bail!("wrong chunk size {} != {}", expected_chunk_size, chunk_size);
}
self.stat.chunk_count += 1;
self.last_chunk = self.chunk_offset;
let (chunk, digest) = DataChunkBuilder::new(&self.chunk_buffer)
.compress(true)
.build()?;
match self.index.insert_chunk(&chunk, &digest) {
Ok((is_duplicate, compressed_size)) => {
self.stat.compressed_size += compressed_size;
if is_duplicate {
self.stat.duplicate_chunks += 1;
} else {
self.stat.disk_size += compressed_size;
}
println!(
"ADD CHUNK {:016x} {} {}% {} {}",
self.chunk_offset,
chunk_size,
(compressed_size * 100) / (chunk_size as u64),
is_duplicate,
proxmox::tools::digest_to_hex(&digest)
);
self.index.add_chunk(self.chunk_offset as u64, &digest)?;
self.chunk_buffer.truncate(0);
Ok(())
}
Err(err) => {
self.chunk_buffer.truncate(0);
Err(err)
}
}
}
}
impl Write for DynamicChunkWriter {
fn write(&mut self, data: &[u8]) -> std::result::Result<usize, std::io::Error> {
let chunker = &mut self.chunker;
let pos = chunker.scan(data);
if pos > 0 {
self.chunk_buffer.extend_from_slice(&data[0..pos]);
self.chunk_offset += pos;
if let Err(err) = self.write_chunk_buffer() {
return Err(std::io::Error::new(
std::io::ErrorKind::Other,
err.to_string(),
));
}
Ok(pos)
} else {
self.chunk_offset += data.len();
self.chunk_buffer.extend_from_slice(data);
Ok(data.len())
}
}
fn flush(&mut self) -> std::result::Result<(), std::io::Error> {
Err(std::io::Error::new(
std::io::ErrorKind::Other,
"please use close() instead of flush()",
))
}
}

14
src/backup/fixed_index.rs → pbs-datastore/src/fixed_index.rs
View File

@ -7,15 +7,17 @@ use std::io::{Seek, SeekFrom};
use anyhow::{bail, format_err, Error};
use pbs_datastore::chunk_stat::ChunkStat;
use pbs_datastore::chunk_store::ChunkStore;
use pbs_datastore::data_blob::ChunkInfo;
use pbs_datastore::index::{ChunkReadInfo, IndexFile};
use pbs_tools::process_locker::ProcessLockSharedGuard;
use proxmox::tools::io::ReadExt;
use proxmox::tools::Uuid;
use crate::chunk_stat::ChunkStat;
use crate::chunk_store::ChunkStore;
use crate::data_blob::ChunkInfo;
use crate::file_formats;
use crate::index::{ChunkReadInfo, IndexFile};
/// Header format definition for fixed index files (`.fidx`)
#[repr(C)]
pub struct FixedIndexHeader {
@ -81,7 +83,7 @@ impl FixedIndexReader {
let header: Box<FixedIndexHeader> = unsafe { file.read_host_value_boxed()? };
if header.magic != super::FIXED_SIZED_CHUNK_INDEX_1_0 {
if header.magic != file_formats::FIXED_SIZED_CHUNK_INDEX_1_0 {
bail!("got unknown magic number");
}
@ -286,7 +288,7 @@ impl FixedIndexWriter {
let buffer = vec![0u8; header_size];
let header = unsafe { &mut *(buffer.as_ptr() as *mut FixedIndexHeader) };
header.magic = super::FIXED_SIZED_CHUNK_INDEX_1_0;
header.magic = file_formats::FIXED_SIZED_CHUNK_INDEX_1_0;
header.ctime = i64::to_le(ctime);
header.size = u64::to_le(size as u64);
header.chunk_size = u64::to_le(chunk_size as u64);

4
pbs-datastore/src/lib.rs
View File

@ -195,9 +195,13 @@ pub mod file_formats;
pub mod index;
pub mod key_derivation;
pub mod manifest;
pub mod prune;
pub mod read_chunk;
pub mod task;
pub mod dynamic_index;
pub mod fixed_index;
pub use backup_info::{BackupDir, BackupGroup, BackupInfo};
pub use checksum_reader::ChecksumReader;
pub use checksum_writer::ChecksumWriter;

3
src/backup/prune.rs → pbs-datastore/src/prune.rs
View File

@ -1,7 +1,8 @@
use anyhow::{Error};
use std::collections::{HashMap, HashSet};
use std::path::PathBuf;
use anyhow::{Error};
use super::BackupInfo;
enum PruneMark { Keep, KeepPartial, Remove }

22
src/backup/datastore.rs
View File

@ -12,22 +12,26 @@ use lazy_static::lazy_static;
use proxmox::tools::fs::{replace_file, file_read_optional_string, CreateOptions, open_file_locked};
use pbs_api_types::upid::UPID;
use pbs_api_types::{Authid, GarbageCollectionStatus};
use pbs_datastore::{task_log, task_warn};
use pbs_datastore::DataBlob;
use pbs_datastore::backup_info::{BackupGroup, BackupDir};
use pbs_datastore::chunk_store::ChunkStore;
use pbs_datastore::dynamic_index::{DynamicIndexReader, DynamicIndexWriter};
use pbs_datastore::fixed_index::{FixedIndexReader, FixedIndexWriter};
use pbs_datastore::index::IndexFile;
use pbs_datastore::manifest::{
MANIFEST_BLOB_NAME, MANIFEST_LOCK_NAME, CLIENT_LOG_BLOB_NAME,
ArchiveType, BackupManifest,
archive_type,
};
use pbs_datastore::task::TaskState;
use pbs_tools::format::HumanByte;
use pbs_tools::fs::{lock_dir_noblock, DirLockGuard};
use super::backup_info::{BackupGroup, BackupDir};
use super::chunk_store::ChunkStore;
use super::dynamic_index::{DynamicIndexReader, DynamicIndexWriter};
use super::fixed_index::{FixedIndexReader, FixedIndexWriter};
use super::manifest::{MANIFEST_BLOB_NAME, MANIFEST_LOCK_NAME, CLIENT_LOG_BLOB_NAME, BackupManifest};
use super::index::*;
use super::{DataBlob, ArchiveType, archive_type};
use crate::config::datastore::{self, DataStoreConfig};
use crate::tools;
use crate::api2::types::{Authid, GarbageCollectionStatus};
use crate::server::UPID;
lazy_static! {
static ref DATASTORE_MAP: Mutex<HashMap<String, Arc<DataStore>>> = Mutex::new(HashMap::new());

506
src/backup/dynamic_index.rs
View File

@ -1,263 +1,16 @@
use std::fs::File;
use std::io::{self, BufWriter, Seek, SeekFrom, Write};
use std::io::{self, Seek, SeekFrom};
use std::ops::Range;
use std::os::unix::io::AsRawFd;
use std::path::{Path, PathBuf};
use std::sync::{Arc, Mutex};
use std::task::Context;
use std::pin::Pin;
use anyhow::{bail, format_err, Error};
use proxmox::tools::io::ReadExt;
use proxmox::tools::uuid::Uuid;
use proxmox::tools::mmap::Mmap;
use pxar::accessor::{MaybeReady, ReadAt, ReadAtOperation};
use pbs_datastore::Chunker;
use pbs_datastore::index::{IndexFile, ChunkReadInfo};
use pbs_datastore::chunk_stat::ChunkStat;
use pbs_datastore::data_blob::{DataBlob, DataChunkBuilder};
use pbs_datastore::chunk_store::ChunkStore;
use pbs_datastore::dynamic_index::DynamicIndexReader;
use pbs_datastore::read_chunk::ReadChunk;
use pbs_tools::process_locker::ProcessLockSharedGuard;
/// Header format definition for dynamic index files (`.dixd`)
#[repr(C)]
pub struct DynamicIndexHeader {
pub magic: [u8; 8],
pub uuid: [u8; 16],
pub ctime: i64,
/// Sha256 over the index ``SHA256(offset1||digest1||offset2||digest2||...)``
pub index_csum: [u8; 32],
reserved: [u8; 4032], // overall size is one page (4096 bytes)
}
proxmox::static_assert_size!(DynamicIndexHeader, 4096);
// TODO: Once non-Copy unions are stabilized, use:
// union DynamicIndexHeader {
// reserved: [u8; 4096],
// pub data: DynamicIndexHeaderData,
// }
impl DynamicIndexHeader {
/// Convenience method to allocate a zero-initialized header struct.
pub fn zeroed() -> Box<Self> {
unsafe {
Box::from_raw(std::alloc::alloc_zeroed(std::alloc::Layout::new::<Self>()) as *mut Self)
}
}
pub fn as_bytes(&self) -> &[u8] {
unsafe {
std::slice::from_raw_parts(
self as *const Self as *const u8,
std::mem::size_of::<Self>(),
)
}
}
}
#[derive(Clone, Debug)]
#[repr(C)]
pub struct DynamicEntry {
end_le: u64,
digest: [u8; 32],
}
impl DynamicEntry {
#[inline]
pub fn end(&self) -> u64 {
u64::from_le(self.end_le)
}
}
pub struct DynamicIndexReader {
_file: File,
pub size: usize,
index: Mmap<DynamicEntry>,
pub uuid: [u8; 16],
pub ctime: i64,
pub index_csum: [u8; 32],
}
impl DynamicIndexReader {
pub fn open(path: &Path) -> Result<Self, Error> {
File::open(path)
.map_err(Error::from)
.and_then(Self::new)
.map_err(|err| format_err!("Unable to open dynamic index {:?} - {}", path, err))
}
pub fn new(mut file: std::fs::File) -> Result<Self, Error> {
// FIXME: This is NOT OUR job! Check the callers of this method and remove this!
file.seek(SeekFrom::Start(0))?;
let header_size = std::mem::size_of::<DynamicIndexHeader>();
let rawfd = file.as_raw_fd();
let stat = match nix::sys::stat::fstat(rawfd) {
Ok(stat) => stat,
Err(err) => bail!("fstat failed - {}", err),
};
let size = stat.st_size as usize;
if size < header_size {
bail!("index too small ({})", stat.st_size);
}
let header: Box<DynamicIndexHeader> = unsafe { file.read_host_value_boxed()? };
if header.magic != super::DYNAMIC_SIZED_CHUNK_INDEX_1_0 {
bail!("got unknown magic number");
}
let ctime = proxmox::tools::time::epoch_i64();
let index_size = stat.st_size as usize - header_size;
let index_count = index_size / 40;
if index_count * 40 != index_size {
bail!("got unexpected file size");
}
let index = unsafe {
Mmap::map_fd(
rawfd,
header_size as u64,
index_count,
nix::sys::mman::ProtFlags::PROT_READ,
nix::sys::mman::MapFlags::MAP_PRIVATE,
)?
};
Ok(Self {
_file: file,
size,
index,
ctime,
uuid: header.uuid,
index_csum: header.index_csum,
})
}
#[inline]
#[allow(clippy::cast_ptr_alignment)]
fn chunk_end(&self, pos: usize) -> u64 {
if pos >= self.index.len() {
panic!("chunk index out of range");
}
self.index[pos].end()
}
#[inline]
fn chunk_digest(&self, pos: usize) -> &[u8; 32] {
if pos >= self.index.len() {
panic!("chunk index out of range");
}
&self.index[pos].digest
}
// TODO: can we use std::slice::binary_search with Mmap now?
fn binary_search(
&self,
start_idx: usize,
start: u64,
end_idx: usize,
end: u64,
offset: u64,
) -> Result<usize, Error> {
if (offset >= end) || (offset < start) {
bail!("offset out of range");
}
if end_idx == start_idx {
return Ok(start_idx); // found
}
let middle_idx = (start_idx + end_idx) / 2;
let middle_end = self.chunk_end(middle_idx);
if offset < middle_end {
self.binary_search(start_idx, start, middle_idx, middle_end, offset)
} else {
self.binary_search(middle_idx + 1, middle_end, end_idx, end, offset)
}
}
}
impl IndexFile for DynamicIndexReader {
fn index_count(&self) -> usize {
self.index.len()
}
fn index_digest(&self, pos: usize) -> Option<&[u8; 32]> {
if pos >= self.index.len() {
None
} else {
Some(unsafe { &*(self.chunk_digest(pos).as_ptr() as *const [u8; 32]) })
}
}
fn index_bytes(&self) -> u64 {
if self.index.is_empty() {
0
} else {
self.chunk_end(self.index.len() - 1)
}
}
fn compute_csum(&self) -> ([u8; 32], u64) {
let mut csum = openssl::sha::Sha256::new();
let mut chunk_end = 0;
for pos in 0..self.index_count() {
let info = self.chunk_info(pos).unwrap();
chunk_end = info.range.end;
csum.update(&chunk_end.to_le_bytes());
csum.update(&info.digest);
}
let csum = csum.finish();
(csum, chunk_end)
}
fn chunk_info(&self, pos: usize) -> Option<ChunkReadInfo> {
if pos >= self.index.len() {
return None;
}
let start = if pos == 0 { 0 } else { self.index[pos - 1].end() };
let end = self.index[pos].end();
Some(ChunkReadInfo {
range: start..end,
digest: self.index[pos].digest,
})
}
fn index_ctime(&self) -> i64 {
self.ctime
}
fn index_size(&self) -> usize {
self.size as usize
}
fn chunk_from_offset(&self, offset: u64) -> Option<(usize, u64)> {
let end_idx = self.index.len() - 1;
let end = self.chunk_end(end_idx);
let found_idx = self.binary_search(0, 0, end_idx, end, offset);
let found_idx = match found_idx {
Ok(i) => i,
Err(_) => return None
};
let found_start = if found_idx == 0 {
0
} else {
self.chunk_end(found_idx - 1)
};
Some((found_idx, offset - found_start))
}
}
use pbs_datastore::index::IndexFile;
struct CachedChunk {
range: Range<u64>,
@ -358,7 +111,7 @@ impl<S: ReadChunk> crate::tools::BufferedRead for BufferedDynamicReader<S> {
// optimization for sequential read
if buffer_len > 0
&& ((self.buffered_chunk_idx + 1) < index.index.len())
&& ((self.buffered_chunk_idx + 1) < index.index().len())
&& (offset >= (self.buffered_chunk_start + (self.read_buffer.len() as u64)))
{
let next_idx = self.buffered_chunk_idx + 1;
@ -374,7 +127,7 @@ impl<S: ReadChunk> crate::tools::BufferedRead for BufferedDynamicReader<S> {
|| (offset < self.buffered_chunk_start)
|| (offset >= (self.buffered_chunk_start + (self.read_buffer.len() as u64)))
{
let end_idx = index.index.len() - 1;
let end_idx = index.index().len() - 1;
let end = index.chunk_end(end_idx);
let idx = index.binary_search(0, 0, end_idx, end, offset)?;
self.buffer_chunk(idx)?;
@ -474,252 +227,3 @@ impl<R: ReadChunk> ReadAt for LocalDynamicReadAt<R> {
panic!("LocalDynamicReadAt::start_read_at returned Pending");
}
}
/// Create dynamic index files (`.dixd`)
pub struct DynamicIndexWriter {
store: Arc<ChunkStore>,
_lock: ProcessLockSharedGuard,
writer: BufWriter<File>,
closed: bool,
filename: PathBuf,
tmp_filename: PathBuf,
csum: Option<openssl::sha::Sha256>,
pub uuid: [u8; 16],
pub ctime: i64,
}
impl Drop for DynamicIndexWriter {
fn drop(&mut self) {
let _ = std::fs::remove_file(&self.tmp_filename); // ignore errors
}
}
impl DynamicIndexWriter {
pub fn create(store: Arc<ChunkStore>, path: &Path) -> Result<Self, Error> {
let shared_lock = store.try_shared_lock()?;
let full_path = store.relative_path(path);
let mut tmp_path = full_path.clone();
tmp_path.set_extension("tmp_didx");
let file = std::fs::OpenOptions::new()
.create(true)
.truncate(true)
.read(true)
.write(true)
.open(&tmp_path)?;
let mut writer = BufWriter::with_capacity(1024 * 1024, file);
let ctime = proxmox::tools::time::epoch_i64();
let uuid = Uuid::generate();
let mut header = DynamicIndexHeader::zeroed();
header.magic = super::DYNAMIC_SIZED_CHUNK_INDEX_1_0;
header.ctime = i64::to_le(ctime);
header.uuid = *uuid.as_bytes();
// header.index_csum = [0u8; 32];
writer.write_all(header.as_bytes())?;
let csum = Some(openssl::sha::Sha256::new());
Ok(Self {
store,
_lock: shared_lock,
writer,
closed: false,
filename: full_path,
tmp_filename: tmp_path,
ctime,
uuid: *uuid.as_bytes(),
csum,
})
}
// fixme: use add_chunk instead?
pub fn insert_chunk(&self, chunk: &DataBlob, digest: &[u8; 32]) -> Result<(bool, u64), Error> {
self.store.insert_chunk(chunk, digest)
}
pub fn close(&mut self) -> Result<[u8; 32], Error> {
if self.closed {
bail!(
"cannot close already closed archive index file {:?}",
self.filename
);
}
self.closed = true;
self.writer.flush()?;
let csum_offset = proxmox::offsetof!(DynamicIndexHeader, index_csum);
self.writer.seek(SeekFrom::Start(csum_offset as u64))?;
let csum = self.csum.take().unwrap();
let index_csum = csum.finish();
self.writer.write_all(&index_csum)?;
self.writer.flush()?;
if let Err(err) = std::fs::rename(&self.tmp_filename, &self.filename) {
bail!("Atomic rename file {:?} failed - {}", self.filename, err);
}
Ok(index_csum)
}
// fixme: rename to add_digest
pub fn add_chunk(&mut self, offset: u64, digest: &[u8; 32]) -> Result<(), Error> {
if self.closed {
bail!(
"cannot write to closed dynamic index file {:?}",
self.filename
);
}
let offset_le: &[u8; 8] = unsafe { &std::mem::transmute::<u64, [u8; 8]>(offset.to_le()) };
if let Some(ref mut csum) = self.csum {
csum.update(offset_le);
csum.update(digest);
}
self.writer.write_all(offset_le)?;
self.writer.write_all(digest)?;
Ok(())
}
}
/// Writer which splits a binary stream into dynamic sized chunks
///
/// And store the resulting chunk list into the index file.
pub struct DynamicChunkWriter {
index: DynamicIndexWriter,
closed: bool,
chunker: Chunker,
stat: ChunkStat,
chunk_offset: usize,
last_chunk: usize,
chunk_buffer: Vec<u8>,
}
impl DynamicChunkWriter {
pub fn new(index: DynamicIndexWriter, chunk_size: usize) -> Self {
Self {
index,
closed: false,
chunker: Chunker::new(chunk_size),
stat: ChunkStat::new(0),
chunk_offset: 0,
last_chunk: 0,
chunk_buffer: Vec::with_capacity(chunk_size * 4),
}
}
pub fn stat(&self) -> &ChunkStat {
&self.stat
}
pub fn close(&mut self) -> Result<(), Error> {
if self.closed {
return Ok(());
}
self.closed = true;
self.write_chunk_buffer()?;
self.index.close()?;
self.stat.size = self.chunk_offset as u64;
// add size of index file
self.stat.size +=
(self.stat.chunk_count * 40 + std::mem::size_of::<DynamicIndexHeader>()) as u64;
Ok(())
}
fn write_chunk_buffer(&mut self) -> Result<(), Error> {
let chunk_size = self.chunk_buffer.len();
if chunk_size == 0 {
return Ok(());
}
let expected_chunk_size = self.chunk_offset - self.last_chunk;
if expected_chunk_size != self.chunk_buffer.len() {
bail!("wrong chunk size {} != {}", expected_chunk_size, chunk_size);
}
self.stat.chunk_count += 1;
self.last_chunk = self.chunk_offset;
let (chunk, digest) = DataChunkBuilder::new(&self.chunk_buffer)
.compress(true)
.build()?;
match self.index.insert_chunk(&chunk, &digest) {
Ok((is_duplicate, compressed_size)) => {
self.stat.compressed_size += compressed_size;
if is_duplicate {
self.stat.duplicate_chunks += 1;
} else {
self.stat.disk_size += compressed_size;
}
println!(
"ADD CHUNK {:016x} {} {}% {} {}",
self.chunk_offset,
chunk_size,
(compressed_size * 100) / (chunk_size as u64),
is_duplicate,
proxmox::tools::digest_to_hex(&digest)
);
self.index.add_chunk(self.chunk_offset as u64, &digest)?;
self.chunk_buffer.truncate(0);
Ok(())
}
Err(err) => {
self.chunk_buffer.truncate(0);
Err(err)
}
}
}
}
impl Write for DynamicChunkWriter {
fn write(&mut self, data: &[u8]) -> std::result::Result<usize, std::io::Error> {
let chunker = &mut self.chunker;
let pos = chunker.scan(data);
if pos > 0 {
self.chunk_buffer.extend_from_slice(&data[0..pos]);
self.chunk_offset += pos;
if let Err(err) = self.write_chunk_buffer() {
return Err(std::io::Error::new(
std::io::ErrorKind::Other,
err.to_string(),
));
}
Ok(pos)
} else {
self.chunk_offset += data.len();
self.chunk_buffer.extend_from_slice(data);
Ok(data.len())
}
}
fn flush(&mut self) -> std::result::Result<(), std::io::Error> {
Err(std::io::Error::new(
std::io::ErrorKind::Other,
"please use close() instead of flush()",
))
}
}

158
src/backup/mod.rs
View File

@ -1,146 +1,4 @@
//! This module implements the data storage and access layer.
//!
//! # Data formats
//!
//! PBS splits large files into chunks, and stores them deduplicated using
//! a content addressable storage format.
//!
//! Backup snapshots are stored as folders containing a manifest file and
//! potentially one or more index or blob files.
//!
//! The manifest contains hashes of all other files and can be signed by
//! the client.
//!
//! Blob files contain data directly. They are used for config files and
//! the like.
//!
//! Index files are used to reconstruct an original file. They contain a
//! list of SHA256 checksums. The `DynamicIndex*` format is able to deal
//! with dynamic chunk sizes (CT and host backups), whereas the
//! `FixedIndex*` format is an optimization to store a list of equal sized
//! chunks (VMs, whole block devices).
//!
//! A chunk is defined as a binary blob, which is stored inside a
//! [ChunkStore](struct.ChunkStore.html) instead of the backup directory
//! directly, and can be addressed by its SHA256 digest.
//!
//!
//! # Garbage Collection (GC)
//!
//! Deleting backups is as easy as deleting the corresponding .idx files.
//! However, this does not free up any storage, because those files just
//! contain references to chunks.
//!
//! To free up some storage, we run a garbage collection process at
//! regular intervals. The collector uses a mark and sweep approach. In
//! the first phase, it scans all .idx files to mark used chunks. The
//! second phase then removes all unmarked chunks from the store.
//!
//! The locking mechanisms mentioned below make sure that we are the only
//! process running GC. We still want to be able to create backups during
//! GC, so there may be multiple backup threads/tasks running, either
//! started before GC, or while GC is running.
//!
//! ## `atime` based GC
//!
//! The idea here is to mark chunks by updating the `atime` (access
//! timestamp) on the chunk file. This is quite simple and does not need
//! additional RAM.
//!
//! One minor problem is that recent Linux versions use the `relatime`
//! mount flag by default for performance reasons (and we want that). When
//! enabled, `atime` data is written to the disk only if the file has been
//! modified since the `atime` data was last updated (`mtime`), or if the
//! file was last accessed more than a certain amount of time ago (by
//! default 24h). So we may only delete chunks with `atime` older than 24
//! hours.
//!
//! Another problem arises from running backups. The mark phase does not
//! find any chunks from those backups, because there is no .idx file for
//! them (created after the backup). Chunks created or touched by those
//! backups may have an `atime` as old as the start time of those backups.
//! Please note that the backup start time may predate the GC start time.
//! So we may only delete chunks older than the start time of those
//! running backup jobs, which might be more than 24h back (this is the
//! reason why ProcessLocker exclusive locks only have to be exclusive
//! between processes, since within one we can determine the age of the
//! oldest shared lock).
//!
//! ## Store `marks` in RAM using a HASH
//!
//! Might be better. Under investigation.
//!
//!
//! # Locking
//!
//! Since PBS allows multiple potentially interfering operations at the
//! same time (e.g. garbage collect, prune, multiple backup creations
//! (only in separate groups), forget, ...), these need to lock against
//! each other in certain scenarios. There is no overarching global lock
//! though, instead always the finest grained lock possible is used,
//! because running these operations concurrently is treated as a feature
//! on its own.
//!
//! ## Inter-process Locking
//!
//! We need to be able to restart the proxmox-backup service daemons, so
//! that we can update the software without rebooting the host. But such
//! restarts must not abort running backup jobs, so we need to keep the
//! old service running until those jobs are finished. This implies that
//! we need some kind of locking for modifying chunks and indices in the
//! ChunkStore.
//!
//! Please note that it is perfectly valid to have multiple
//! parallel ChunkStore writers, even when they write the same chunk
//! (because the chunk would have the same name and the same data, and
//! writes are completed atomically via a rename). The only problem is
//! garbage collection, because we need to avoid deleting chunks which are
//! still referenced.
//!
//! To do this we use the
//! [ProcessLocker](../tools/struct.ProcessLocker.html).
//!
//! ### ChunkStore-wide
//!
//! * Create Index Files:
//!
//! Acquire shared lock for ChunkStore.
//!
//! Note: When creating .idx files, we create a temporary .tmp file,
//! then do an atomic rename.
//!
//! * Garbage Collect:
//!
//! Acquire exclusive lock for ChunkStore. If we have
//! already a shared lock for the ChunkStore, try to upgrade that
//! lock.
//!
//! Exclusive locks only work _between processes_. It is valid to have an
//! exclusive and one or more shared locks held within one process. Writing
//! chunks within one process is synchronized using the gc_mutex.
//!
//! On server restart, we stop any running GC in the old process to avoid
//! having the exclusive lock held for too long.
//!
//! ## Locking table
//!
//! Below table shows all operations that play a role in locking, and which
//! mechanisms are used to make their concurrent usage safe.
//!
//! | starting ><br>v during | read index file | create index file | GC mark | GC sweep | update manifest | forget | prune | create backup | verify | reader api |
//! |-|-|-|-|-|-|-|-|-|-|-|
//! | **read index file** | / | / | / | / | / | mmap stays valid, oldest_shared_lock prevents GC | see forget column | / | / | / |
//! | **create index file** | / | / | / | / | / | / | / | /, happens at the end, after all chunks are touched | /, only happens without a manifest | / |
//! | **GC mark** | / | Datastore process-lock shared | gc_mutex, exclusive ProcessLocker | gc_mutex | /, GC only cares about index files, not manifests | tells GC about removed chunks | see forget column | /, index files don’t exist yet | / | / |
//! | **GC sweep** | / | Datastore process-lock shared | gc_mutex, exclusive ProcessLocker | gc_mutex | / | /, chunks already marked | see forget column | chunks get touched; chunk_store.mutex; oldest PL lock | / | / |
//! | **update manifest** | / | / | / | / | update_manifest lock | update_manifest lock, remove dir under lock | see forget column | /, “write manifest” happens at the end | /, can call “write manifest”, see that column | / |
//! | **forget** | / | / | removed_during_gc mutex is held during unlink | marking done, doesn’t matter if forgotten now | update_manifest lock, forget waits for lock | /, unlink is atomic | causes forget to fail, but that’s OK | running backup has snapshot flock | /, potentially detects missing folder | shared snap flock |
//! | **prune** | / | / | see forget row | see forget row | see forget row | causes warn in prune, but no error | see forget column | running and last non-running can’t be pruned | see forget row | shared snap flock |
//! | **create backup** | / | only time this happens, thus has snapshot flock | / | chunks get touched; chunk_store.mutex; oldest PL lock | no lock, but cannot exist beforehand | snapshot flock, can’t be forgotten | running and last non-running can’t be pruned | snapshot group flock, only one running per group | /, won’t be verified since manifest missing | / |
//! | **verify** | / | / | / | / | see “update manifest” row | /, potentially detects missing folder | see forget column | / | /, but useless (“update manifest” protects itself) | / |
//! | **reader api** | / | / | / | /, open snap can’t be forgotten, so ref must exist | / | prevented by shared snap flock | prevented by shared snap flock | / | / | /, lock is shared |!
//! * / = no interaction
//! * shared/exclusive from POV of 'starting' process
//! Server/client-specific parts for what's otherwise in pbs-datastore.
use anyhow::{bail, Error};
@ -216,6 +74,13 @@ pub use pbs_datastore::key_derivation;
pub use pbs_datastore::key_derivation::*;
pub use pbs_datastore::manifest;
pub use pbs_datastore::manifest::*;
pub use pbs_datastore::prune;
pub use pbs_datastore::prune::*;
pub use pbs_datastore::dynamic_index::*;
pub use pbs_datastore::fixed_index;
pub use pbs_datastore::fixed_index::*;
pub use pbs_datastore::read_chunk::*;
mod chunk_stream;
@ -225,15 +90,10 @@ pub use chunk_stream::*;
mod read_chunk;
pub use read_chunk::*;
mod fixed_index;
pub use fixed_index::*;
// Split
mod dynamic_index;
pub use dynamic_index::*;
mod prune;
pub use prune::*;
mod datastore;
pub use datastore::*;