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