src/backup/chunk_store.rs

   1 use failure::*;
   2
   3 use std::path::{Path, PathBuf};
   4 use std::io::{Read, Write};
   5 use std::sync::{Arc, Mutex};
   6 use std::os::unix::io::AsRawFd;
   7
   8 use openssl::sha;
   9
  10 use crate::tools;
  11
  12 pub struct GarbageCollectionStatus {
  13     pub used_bytes: usize,
  14     pub used_chunks: usize,
  15     pub disk_bytes: usize,
  16     pub disk_chunks: usize,
  17 }
  18
  19 impl Default for GarbageCollectionStatus {
  20     fn default() -> Self {
  21         GarbageCollectionStatus {
  22             used_bytes: 0,
  23             used_chunks: 0,
  24             disk_bytes: 0,
  25             disk_chunks: 0,
  26         }
  27     }
  28 }
  29
  30 /// File system based chunk store
  31 pub struct ChunkStore {
  32     name: String, // used for error reporting
  33     pub (crate) base: PathBuf,
  34     chunk_dir: PathBuf,
  35     mutex: Mutex<bool>,
  36     locker: Arc<Mutex<tools::ProcessLocker>>,
  37 }
  38
  39 // TODO: what about sysctl setting vm.vfs_cache_pressure (0 - 100) ?
  40
  41 pub fn verify_chunk_size(size: u64) -> Result<(), Error> {
  42
  43     static SIZES: [u64; 7] = [64*1024, 128*1024, 256*1024, 512*1024, 1024*1024, 2048*1024, 4096*1024];
  44
  45     if !SIZES.contains(&size) {
  46         bail!("Got unsupported chunk size '{}'", size);
  47     }
  48     Ok(())
  49 }
  50
  51 fn digest_to_prefix(digest: &[u8]) -> PathBuf {
  52
  53     let mut buf = Vec::<u8>::with_capacity(2+1+2+1);
  54
  55     const HEX_CHARS: &'static [u8; 16] = b"0123456789abcdef";
  56
  57     buf.push(HEX_CHARS[(digest[0] as usize) >> 4]);
  58     buf.push(HEX_CHARS[(digest[0] as usize) &0xf]);
  59     buf.push(HEX_CHARS[(digest[1] as usize) >> 4]);
  60     buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]);
  61     buf.push('/' as u8);
  62
  63     let path = unsafe { String::from_utf8_unchecked(buf)};
  64
  65     path.into()
  66 }
  67
  68 impl ChunkStore {
  69
  70     fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf {
  71
  72         let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref());
  73         chunk_dir.push(".chunks");
  74
  75         chunk_dir
  76     }
  77
  78     pub fn create<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
  79
  80         let base: PathBuf = path.into();
  81
  82         if !base.is_absolute() {
  83             bail!("expected absolute path - got {:?}", base);
  84         }
  85
  86         let chunk_dir = Self::chunk_dir(&base);
  87
  88         if let Err(err) = std::fs::create_dir(&base) {
  89             bail!("unable to create chunk store '{}' at {:?} - {}", name, base, err);
  90         }
  91
  92         if let Err(err) = std::fs::create_dir(&chunk_dir) {
  93             bail!("unable to create chunk store '{}' subdir {:?} - {}", name, chunk_dir, err);
  94         }
  95
  96         // create 64*1024 subdirs
  97         let mut last_percentage = 0;
  98
  99         for i in 0..64*1024 {
 100             let mut l1path = chunk_dir.clone();
 101             l1path.push(format!("{:04x}", i));
 102             if let Err(err) = std::fs::create_dir(&l1path) {
 103                 bail!("unable to create chunk store '{}' subdir {:?} - {}", name, l1path, err);
 104             }
 105             let percentage = (i*100)/(64*1024);
 106             if percentage != last_percentage {
 107                 eprintln!("Percentage done: {}", percentage);
 108                 last_percentage = percentage;
 109             }
 110         }
 111
 112         Self::open(name, base)
 113     }
 114
 115     pub fn open<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
 116
 117         let base: PathBuf = path.into();
 118
 119         if !base.is_absolute() {
 120             bail!("expected absolute path - got {:?}", base);
 121         }
 122
 123         let chunk_dir = Self::chunk_dir(&base);
 124
 125         if let Err(err) = std::fs::metadata(&chunk_dir) {
 126             bail!("unable to open chunk store '{}' at {:?} - {}", name, chunk_dir, err);
 127         }
 128
 129         let mut lockfile_path = base.clone();
 130         lockfile_path.push(".lock");
 131
 132         let locker = tools::ProcessLocker::new(&lockfile_path)?;
 133
 134         Ok(ChunkStore {
 135             name: name.to_owned(),
 136             base,
 137             chunk_dir,
 138             locker,
 139             mutex: Mutex::new(false)
 140         })
 141     }
 142
 143     pub fn touch_chunk(&self, digest:&[u8]) -> Result<(), Error> {
 144
 145         let mut chunk_path = self.chunk_dir.clone();
 146         let prefix = digest_to_prefix(&digest);
 147         chunk_path.push(&prefix);
 148         let digest_str = tools::digest_to_hex(&digest);
 149         chunk_path.push(&digest_str);
 150
 151         const UTIME_NOW: i64 = ((1 << 30) - 1);
 152         const UTIME_OMIT: i64 = ((1 << 30) - 2);
 153
 154         let times: [libc::timespec; 2] = [
 155             libc::timespec { tv_sec: 0, tv_nsec: UTIME_NOW },
 156             libc::timespec { tv_sec: 0, tv_nsec: UTIME_OMIT }
 157         ];
 158
 159         use nix::NixPath;
 160
 161         let res = chunk_path.with_nix_path(|cstr| unsafe {
 162             libc::utimensat(-1, cstr.as_ptr(), &times[0], libc::AT_SYMLINK_NOFOLLOW)
 163         })?;
 164
 165         if let Err(err) = nix::errno::Errno::result(res) {
 166             bail!("updata atime failed for chunk {:?} - {}", chunk_path, err);
 167         }
 168
 169         Ok(())
 170     }
 171
 172     pub fn read_chunk(&self, digest:&[u8], buffer: &mut Vec<u8>) -> Result<(), Error> {
 173
 174         let mut chunk_path = self.chunk_dir.clone();
 175         let prefix = digest_to_prefix(&digest);
 176         chunk_path.push(&prefix);
 177         let digest_str = tools::digest_to_hex(&digest);
 178         chunk_path.push(&digest_str);
 179
 180         buffer.clear();
 181         let f = std::fs::File::open(&chunk_path)?;
 182         let mut decoder = zstd::stream::Decoder::new(f)?;
 183
 184         decoder.read_to_end(buffer)?;
 185
 186         Ok(())
 187     }
 188
 189     pub fn get_chunk_iterator(
 190         &self,
 191         print_percentage: bool,
 192     ) -> Result<
 193         impl Iterator<Item = Result<tools::fs::ReadDirEntry, Error>> + std::iter::FusedIterator,
 194         Error
 195     > {
 196         use nix::dir::Dir;
 197         use nix::fcntl::OFlag;
 198         use nix::sys::stat::Mode;
 199
 200         let base_handle = match Dir::open(
 201             &self.chunk_dir, OFlag::O_RDONLY, Mode::empty()) {
 202             Ok(h) => h,
 203             Err(err) => bail!("unable to open store '{}' chunk dir {:?} - {}",
 204                               self.name, self.chunk_dir, err),
 205         };
 206
 207         let mut verbose = true;
 208         let mut last_percentage = 0;
 209
 210         Ok((0..0x10000).filter_map(move |index| {
 211             if print_percentage {
 212                 let percentage = (index * 100) / 0x10000;
 213                 if last_percentage != percentage {
 214                     last_percentage = percentage;
 215                     eprintln!("percentage done: {}", percentage);
 216                 }
 217             }
 218             let subdir: &str = &format!("{:04x}", index);
 219             match tools::fs::read_subdir(base_handle.as_raw_fd(), subdir) {
 220                 Err(e) => {
 221                     if verbose {
 222                         eprintln!("Error iterating through chunks: {}", e);
 223                         verbose = false;
 224                     }
 225                     None
 226                 }
 227                 Ok(iter) => Some(iter),
 228             }
 229         })
 230         .flatten()
 231         .filter(|entry| {
 232             // Check that the file name is actually a hash! (64 hex digits)
 233             let entry = match entry {
 234                 Err(_) => return true, // pass errors onwards
 235                 Ok(ref entry) => entry,
 236             };
 237             let bytes = entry.file_name().to_bytes();
 238             if bytes.len() != 64 {
 239                 return false;
 240             }
 241             for b in bytes {
 242                 if !b.is_ascii_hexdigit() {
 243                     return false;
 244                 }
 245             }
 246             true
 247         }))
 248     }
 249
 250     pub fn oldest_writer(&self) -> Option<i64> {
 251         tools::ProcessLocker::oldest_shared_lock(self.locker.clone())
 252     }
 253
 254     pub fn sweep_unused_chunks(
 255         &self,
 256         oldest_writer: Option<i64>,
 257         status: &mut GarbageCollectionStatus
 258     ) -> Result<(), Error> {
 259         use nix::sys::stat::fstatat;
 260
 261         let now = unsafe { libc::time(std::ptr::null_mut()) };
 262
 263         let mut min_atime = now - 3600*24; // at least 24h (see mount option relatime)
 264
 265         if let Some(stamp) = oldest_writer {
 266             if stamp < min_atime {
 267                 min_atime = stamp;
 268             }
 269         }
 270
 271         min_atime -= 300; // add 5 mins gap for safety
 272
 273         for entry in self.get_chunk_iterator(true)? {
 274
 275             tools::fail_on_shutdown()?;
 276
 277             let (dirfd, entry) = match entry {
 278                 Ok(entry) => (entry.parent_fd(), entry),
 279                 Err(_) => continue, // ignore errors
 280             };
 281
 282             let file_type = match entry.file_type() {
 283                 Some(file_type) => file_type,
 284                 None => bail!("unsupported file system type on chunk store '{}'", self.name),
 285             };
 286             if file_type != nix::dir::Type::File {
 287                 continue;
 288             }
 289
 290             let filename = entry.file_name();
 291
 292             let lock = self.mutex.lock();
 293
 294             if let Ok(stat) = fstatat(dirfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
 295                 let age = now - stat.st_atime;
 296                 //println!("FOUND {}  {:?}", age/(3600*24), filename);
 297                 if stat.st_atime < min_atime {
 298                     println!("UNLINK {}  {:?}", age/(3600*24), filename);
 299                     let res = unsafe { libc::unlinkat(dirfd, filename.as_ptr(), 0) };
 300                     if res != 0 {
 301                         let err = nix::Error::last();
 302                         bail!(
 303                             "unlink chunk {:?} failed on store '{}' - {}",
 304                             filename,
 305                             self.name,
 306                             err,
 307                         );
 308                     }
 309                 } else {
 310                     status.disk_chunks += 1;
 311                     status.disk_bytes += stat.st_size as usize;
 312                 }
 313             }
 314             drop(lock);
 315         }
 316         Ok(())
 317     }
 318
 319     pub fn insert_chunk(&self, chunk: &[u8]) -> Result<(bool, [u8; 32], u64), Error> {
 320
 321         // fixme: use Sha512/256 when available
 322         let digest = sha::sha256(chunk);
 323         let (new, csize) = self.insert_chunk_noverify(&digest, chunk)?;
 324         Ok((new, digest, csize))
 325     }
 326
 327     pub fn insert_chunk_noverify(
 328         &self,
 329         digest: &[u8; 32],
 330         chunk: &[u8],
 331     ) -> Result<(bool, u64), Error> {
 332
 333         //println!("DIGEST {}", tools::digest_to_hex(&digest));
 334
 335         let mut chunk_path = self.chunk_dir.clone();
 336         let prefix = digest_to_prefix(digest);
 337         chunk_path.push(&prefix);
 338         let digest_str = tools::digest_to_hex(digest);
 339         chunk_path.push(&digest_str);
 340
 341         let lock = self.mutex.lock();
 342
 343         if let Ok(metadata) = std::fs::metadata(&chunk_path) {
 344             if metadata.is_file() {
 345                 return Ok((true, metadata.len()));
 346             } else {
 347                 bail!("Got unexpected file type on store '{}' for chunk {}", self.name, digest_str);
 348             }
 349         }
 350
 351         let mut tmp_path = chunk_path.clone();
 352         tmp_path.set_extension("tmp");
 353
 354         let f = std::fs::File::create(&tmp_path)?;
 355
 356         let mut encoder = zstd::stream::Encoder::new(f, 1)?;
 357
 358         encoder.write_all(chunk)?;
 359         let f = encoder.finish()?;
 360
 361         if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) {
 362             if let Err(_) = std::fs::remove_file(&tmp_path)  { /* ignore */ }
 363             bail!(
 364                 "Atomic rename on store '{}' failed for chunk {} - {}",
 365                 self.name,
 366                 digest_str,
 367                 err,
 368             );
 369         }
 370
 371         // fixme: is there a better way to get the compressed size?
 372         let stat = nix::sys::stat::fstat(f.as_raw_fd())?;
 373         let compressed_size = stat.st_size as u64;
 374
 375         //println!("PATH {:?}", chunk_path);
 376
 377         drop(lock);
 378
 379         Ok((false, compressed_size))
 380     }
 381
 382     pub fn relative_path(&self, path: &Path) -> PathBuf {
 383
 384         let mut full_path = self.base.clone();
 385         full_path.push(path);
 386         full_path
 387     }
 388
 389     pub fn base_path(&self) -> PathBuf {
 390         self.base.clone()
 391     }
 392
 393     pub fn try_shared_lock(&self) -> Result<tools::ProcessLockSharedGuard, Error> {
 394         tools::ProcessLocker::try_shared_lock(self.locker.clone())
 395     }
 396
 397     pub fn try_exclusive_lock(&self) -> Result<tools::ProcessLockExclusiveGuard, Error> {
 398         tools::ProcessLocker::try_exclusive_lock(self.locker.clone())
 399     }
 400 }
 401
 402
 403 #[test]
 404 fn test_chunk_store1() {
 405
 406     let mut path = std::fs::canonicalize(".").unwrap(); // we need absulute path
 407     path.push(".testdir");
 408
 409     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 410
 411     let chunk_store = ChunkStore::open("test", &path);
 412     assert!(chunk_store.is_err());
 413
 414     let chunk_store = ChunkStore::create("test", &path).unwrap();
 415     let (exists, _, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 416     assert!(!exists);
 417
 418     let (exists, _, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 419     assert!(exists);
 420
 421
 422     let chunk_store = ChunkStore::create("test", &path);
 423     assert!(chunk_store.is_err());
 424
 425     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 426 }