src/backup/chunk_store.rs

   1 use failure::*;
   2
   3 use std::path::{Path, PathBuf};
   4 use std::io::Write;
   5 use std::sync::{Arc, Mutex};
   6 use std::os::unix::io::AsRawFd;
   7 use serde::Serialize;
   8
   9 use proxmox::tools::fs::{CreateOptions, create_path, create_dir};
  10
  11 use crate::tools;
  12 use super::DataBlob;
  13 use crate::server::WorkerTask;
  14
  15 #[derive(Clone, Serialize)]
  16 pub struct GarbageCollectionStatus {
  17     pub upid: Option<String>,
  18     pub index_file_count: usize,
  19     pub index_data_bytes: u64,
  20     pub disk_bytes: u64,
  21     pub disk_chunks: usize,
  22     pub removed_bytes: u64,
  23     pub removed_chunks: usize,
  24 }
  25
  26 impl Default for GarbageCollectionStatus {
  27     fn default() -> Self {
  28         GarbageCollectionStatus {
  29             upid: None,
  30             index_file_count: 0,
  31             index_data_bytes: 0,
  32             disk_bytes: 0,
  33             disk_chunks: 0,
  34             removed_bytes: 0,
  35             removed_chunks: 0,
  36         }
  37     }
  38 }
  39
  40 /// File system based chunk store
  41 pub struct ChunkStore {
  42     name: String, // used for error reporting
  43     pub (crate) base: PathBuf,
  44     chunk_dir: PathBuf,
  45     mutex: Mutex<bool>,
  46     locker: Arc<Mutex<tools::ProcessLocker>>,
  47 }
  48
  49 // TODO: what about sysctl setting vm.vfs_cache_pressure (0 - 100) ?
  50
  51 pub fn verify_chunk_size(size: usize) -> Result<(), Error> {
  52
  53     static SIZES: [usize; 7] = [64*1024, 128*1024, 256*1024, 512*1024, 1024*1024, 2048*1024, 4096*1024];
  54
  55     if !SIZES.contains(&size) {
  56         bail!("Got unsupported chunk size '{}'", size);
  57     }
  58     Ok(())
  59 }
  60
  61 fn digest_to_prefix(digest: &[u8]) -> PathBuf {
  62
  63     let mut buf = Vec::<u8>::with_capacity(2+1+2+1);
  64
  65     const HEX_CHARS: &[u8; 16] = b"0123456789abcdef";
  66
  67     buf.push(HEX_CHARS[(digest[0] as usize) >> 4]);
  68     buf.push(HEX_CHARS[(digest[0] as usize) &0xf]);
  69     buf.push(HEX_CHARS[(digest[1] as usize) >> 4]);
  70     buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]);
  71     buf.push('/' as u8);
  72
  73     let path = unsafe { String::from_utf8_unchecked(buf)};
  74
  75     path.into()
  76 }
  77
  78 impl ChunkStore {
  79
  80     fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf {
  81
  82         let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref());
  83         chunk_dir.push(".chunks");
  84
  85         chunk_dir
  86     }
  87
  88     pub fn create<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
  89
  90         let base: PathBuf = path.into();
  91
  92         if !base.is_absolute() {
  93             bail!("expected absolute path - got {:?}", base);
  94         }
  95
  96         let chunk_dir = Self::chunk_dir(&base);
  97
  98         let backup_user = crate::backup::backup_user()?;
  99
 100         let options = CreateOptions::new()
 101             .owner(backup_user.uid)
 102             .group(backup_user.gid);
 103
 104         let default_options = CreateOptions::new();
 105
 106         if let Err(err) = create_path(&base, Some(default_options.clone()), Some(options.clone())) {
 107             bail!("unable to create chunk store '{}' at {:?} - {}", name, base, err);
 108         }
 109
 110         if let Err(err) = create_dir(&chunk_dir, options.clone()) {
 111             bail!("unable to create chunk store '{}' subdir {:?} - {}", name, chunk_dir, err);
 112         }
 113
 114         // create lock file with correct owner/group
 115         let lockfile_path = Self::lockfile_path(&base);
 116         proxmox::tools::fs::replace_file(lockfile_path, b"", options.clone())?;
 117
 118         // create 64*1024 subdirs
 119         let mut last_percentage = 0;
 120
 121         for i in 0..64*1024 {
 122             let mut l1path = chunk_dir.clone();
 123             l1path.push(format!("{:04x}", i));
 124             if let Err(err) = create_dir(&l1path, options.clone()) {
 125                 bail!("unable to create chunk store '{}' subdir {:?} - {}", name, l1path, err);
 126             }
 127             let percentage = (i*100)/(64*1024);
 128             if percentage != last_percentage {
 129                 eprintln!("Percentage done: {}", percentage);
 130                 last_percentage = percentage;
 131             }
 132         }
 133
 134
 135         Self::open(name, base)
 136     }
 137
 138     fn lockfile_path<P: Into<PathBuf>>(base: P) -> PathBuf {
 139         let base: PathBuf = base.into();
 140
 141         let mut lockfile_path = base.clone();
 142         lockfile_path.push(".lock");
 143
 144         lockfile_path
 145     }
 146
 147     pub fn open<P: Into<PathBuf>>(name: &str, base: P) -> Result<Self, Error> {
 148
 149         let base: PathBuf = base.into();
 150
 151         if !base.is_absolute() {
 152             bail!("expected absolute path - got {:?}", base);
 153         }
 154
 155         let chunk_dir = Self::chunk_dir(&base);
 156
 157         if let Err(err) = std::fs::metadata(&chunk_dir) {
 158             bail!("unable to open chunk store '{}' at {:?} - {}", name, chunk_dir, err);
 159         }
 160
 161         let lockfile_path = Self::lockfile_path(&base);
 162
 163         let locker = tools::ProcessLocker::new(&lockfile_path)?;
 164
 165         Ok(ChunkStore {
 166             name: name.to_owned(),
 167             base,
 168             chunk_dir,
 169             locker,
 170             mutex: Mutex::new(false)
 171         })
 172     }
 173
 174     pub fn touch_chunk(&self, digest: &[u8; 32]) -> Result<(), Error> {
 175
 176         let (chunk_path, _digest_str) = self.chunk_path(digest);
 177
 178         const UTIME_NOW: i64 = ((1 << 30) - 1);
 179         const UTIME_OMIT: i64 = ((1 << 30) - 2);
 180
 181         let times: [libc::timespec; 2] = [
 182             libc::timespec { tv_sec: 0, tv_nsec: UTIME_NOW },
 183             libc::timespec { tv_sec: 0, tv_nsec: UTIME_OMIT }
 184         ];
 185
 186         use nix::NixPath;
 187
 188         let res = chunk_path.with_nix_path(|cstr| unsafe {
 189             libc::utimensat(-1, cstr.as_ptr(), &times[0], libc::AT_SYMLINK_NOFOLLOW)
 190         })?;
 191
 192         if let Err(err) = nix::errno::Errno::result(res) {
 193             bail!("updata atime failed for chunk {:?} - {}", chunk_path, err);
 194         }
 195
 196         Ok(())
 197     }
 198
 199     pub fn read_chunk(&self, digest: &[u8; 32]) -> Result<DataBlob, Error> {
 200
 201         let (chunk_path, digest_str) = self.chunk_path(digest);
 202         let mut file = std::fs::File::open(&chunk_path)
 203             .map_err(|err| {
 204                 format_err!(
 205                     "store '{}', unable to read chunk '{}' - {}",
 206                     self.name,
 207                     digest_str,
 208                     err,
 209                 )
 210             })?;
 211
 212         DataBlob::load(&mut file)
 213     }
 214
 215     pub fn get_chunk_iterator(
 216         &self,
 217     ) -> Result<
 218         impl Iterator<Item = (Result<tools::fs::ReadDirEntry, Error>, usize)> + std::iter::FusedIterator,
 219         Error
 220     > {
 221         use nix::dir::Dir;
 222         use nix::fcntl::OFlag;
 223         use nix::sys::stat::Mode;
 224
 225         let base_handle = Dir::open(&self.chunk_dir, OFlag::O_RDONLY, Mode::empty())
 226             .map_err(|err| {
 227                 format_err!(
 228                     "unable to open store '{}' chunk dir {:?} - {}",
 229                     self.name,
 230                     self.chunk_dir,
 231                     err,
 232                 )
 233             })?;
 234
 235         let mut done = false;
 236         let mut inner: Option<tools::fs::ReadDir> = None;
 237         let mut at = 0;
 238         let mut percentage = 0;
 239         Ok(std::iter::from_fn(move || {
 240             if done {
 241                 return None;
 242             }
 243
 244             loop {
 245                 if let Some(ref mut inner) = inner {
 246                     match inner.next() {
 247                         Some(Ok(entry)) => {
 248                             // skip files if they're not a hash
 249                             let bytes = entry.file_name().to_bytes();
 250                             if bytes.len() != 64 {
 251                                 continue;
 252                             }
 253                             if !bytes.iter().all(u8::is_ascii_hexdigit) {
 254                                 continue;
 255                             }
 256                             return Some((Ok(entry), percentage));
 257                         }
 258                         Some(Err(err)) => {
 259                             // stop after first error
 260                             done = true;
 261                             // and pass the error through:
 262                             return Some((Err(err), percentage));
 263                         }
 264                         None => (), // open next directory
 265                     }
 266                 }
 267
 268                 inner = None;
 269
 270                 if at == 0x10000 {
 271                     done = true;
 272                     return None;
 273                 }
 274
 275                 let subdir: &str = &format!("{:04x}", at);
 276                 percentage = (at * 100) / 0x10000;
 277                 at += 1;
 278                 match tools::fs::read_subdir(base_handle.as_raw_fd(), subdir) {
 279                     Ok(dir) => {
 280                         inner = Some(dir);
 281                         // start reading:
 282                         continue;
 283                     }
 284                     Err(ref err) if err.as_errno() == Some(nix::errno::Errno::ENOENT) => {
 285                         // non-existing directories are okay, just keep going:
 286                         continue;
 287                     }
 288                     Err(err) => {
 289                         // other errors are fatal, so end our iteration
 290                         done = true;
 291                         // and pass the error through:
 292                         return Some((Err(format_err!("unable to read subdir '{}' - {}", subdir, err)), percentage));
 293                     }
 294                 }
 295             }
 296         }).fuse())
 297     }
 298
 299     pub fn oldest_writer(&self) -> Option<i64> {
 300         tools::ProcessLocker::oldest_shared_lock(self.locker.clone())
 301     }
 302
 303     pub fn sweep_unused_chunks(
 304         &self,
 305         oldest_writer: Option<i64>,
 306         status: &mut GarbageCollectionStatus,
 307         worker: Arc<WorkerTask>,
 308     ) -> Result<(), Error> {
 309         use nix::sys::stat::fstatat;
 310
 311         let now = unsafe { libc::time(std::ptr::null_mut()) };
 312
 313         let mut min_atime = now - 3600*24; // at least 24h (see mount option relatime)
 314
 315         if let Some(stamp) = oldest_writer {
 316             if stamp < min_atime {
 317                 min_atime = stamp;
 318             }
 319         }
 320
 321         min_atime -= 300; // add 5 mins gap for safety
 322
 323         let mut last_percentage = 0;
 324         let mut chunk_count = 0;
 325
 326         for (entry, percentage) in self.get_chunk_iterator()? {
 327             if last_percentage != percentage {
 328                 last_percentage = percentage;
 329                 worker.log(format!("percentage done: {}, chunk count: {}", percentage, chunk_count));
 330             }
 331
 332             tools::fail_on_shutdown()?;
 333
 334             let (dirfd, entry) = match entry {
 335                 Ok(entry) => (entry.parent_fd(), entry),
 336                 Err(err) => bail!("chunk iterator on chunk store '{}' failed - {}", self.name, err),
 337             };
 338
 339             let file_type = match entry.file_type() {
 340                 Some(file_type) => file_type,
 341                 None => bail!("unsupported file system type on chunk store '{}'", self.name),
 342             };
 343             if file_type != nix::dir::Type::File {
 344                 continue;
 345             }
 346
 347             chunk_count += 1;
 348
 349             let filename = entry.file_name();
 350
 351             let lock = self.mutex.lock();
 352
 353             if let Ok(stat) = fstatat(dirfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
 354                 let age = now - stat.st_atime;
 355                 //println!("FOUND {}  {:?}", age/(3600*24), filename);
 356                 if stat.st_atime < min_atime {
 357                     println!("UNLINK {}  {:?}", age/(3600*24), filename);
 358                     let res = unsafe { libc::unlinkat(dirfd, filename.as_ptr(), 0) };
 359                     if res != 0 {
 360                         let err = nix::Error::last();
 361                         bail!(
 362                             "unlink chunk {:?} failed on store '{}' - {}",
 363                             filename,
 364                             self.name,
 365                             err,
 366                         );
 367                     }
 368                     status.removed_chunks += 1;
 369                     status.removed_bytes += stat.st_size as u64;
 370                } else {
 371                     status.disk_chunks += 1;
 372                     status.disk_bytes += stat.st_size as u64;
 373                 }
 374             }
 375             drop(lock);
 376         }
 377
 378         Ok(())
 379     }
 380
 381     pub fn insert_chunk(
 382         &self,
 383         chunk: &DataBlob,
 384         digest: &[u8; 32],
 385     ) -> Result<(bool, u64), Error> {
 386
 387         //println!("DIGEST {}", proxmox::tools::digest_to_hex(digest));
 388
 389         let (chunk_path, digest_str) = self.chunk_path(digest);
 390
 391         let lock = self.mutex.lock();
 392
 393         if let Ok(metadata) = std::fs::metadata(&chunk_path) {
 394             if metadata.is_file() {
 395                 return Ok((true, metadata.len()));
 396             } else {
 397                 bail!("Got unexpected file type on store '{}' for chunk {}", self.name, digest_str);
 398             }
 399         }
 400
 401         let mut tmp_path = chunk_path.clone();
 402         tmp_path.set_extension("tmp");
 403
 404         let mut file = std::fs::File::create(&tmp_path)?;
 405
 406         let raw_data = chunk.raw_data();
 407         let encoded_size = raw_data.len() as u64;
 408
 409         file.write_all(raw_data)?;
 410
 411         if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) {
 412             if let Err(_) = std::fs::remove_file(&tmp_path)  { /* ignore */ }
 413             bail!(
 414                 "Atomic rename on store '{}' failed for chunk {} - {}",
 415                 self.name,
 416                 digest_str,
 417                 err,
 418             );
 419         }
 420
 421         drop(lock);
 422
 423         Ok((false, encoded_size))
 424     }
 425
 426     pub fn chunk_path(&self, digest:&[u8; 32]) -> (PathBuf, String) {
 427         let mut chunk_path = self.chunk_dir.clone();
 428         let prefix = digest_to_prefix(digest);
 429         chunk_path.push(&prefix);
 430         let digest_str = proxmox::tools::digest_to_hex(digest);
 431         chunk_path.push(&digest_str);
 432         (chunk_path, digest_str)
 433     }
 434
 435     pub fn relative_path(&self, path: &Path) -> PathBuf {
 436
 437         let mut full_path = self.base.clone();
 438         full_path.push(path);
 439         full_path
 440     }
 441
 442     pub fn base_path(&self) -> PathBuf {
 443         self.base.clone()
 444     }
 445
 446     pub fn try_shared_lock(&self) -> Result<tools::ProcessLockSharedGuard, Error> {
 447         tools::ProcessLocker::try_shared_lock(self.locker.clone())
 448     }
 449
 450     pub fn try_exclusive_lock(&self) -> Result<tools::ProcessLockExclusiveGuard, Error> {
 451         tools::ProcessLocker::try_exclusive_lock(self.locker.clone())
 452     }
 453 }
 454
 455
 456 #[test]
 457 fn test_chunk_store1() {
 458
 459     let mut path = std::fs::canonicalize(".").unwrap(); // we need absulute path
 460     path.push(".testdir");
 461
 462     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 463
 464     let chunk_store = ChunkStore::open("test", &path);
 465     assert!(chunk_store.is_err());
 466
 467     let chunk_store = ChunkStore::create("test", &path).unwrap();
 468
 469     let (chunk, digest) = super::DataChunkBuilder::new(&[0u8, 1u8]).build().unwrap();
 470
 471     let (exists, _) = chunk_store.insert_chunk(&chunk, &digest).unwrap();
 472     assert!(!exists);
 473
 474     let (exists, _) = chunk_store.insert_chunk(&chunk, &digest).unwrap();
 475     assert!(exists);
 476
 477
 478     let chunk_store = ChunkStore::create("test", &path);
 479     assert!(chunk_store.is_err());
 480
 481     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 482 }