src/backup/chunk_store.rs

   1 use failure::*;
   2 use std::path::{Path, PathBuf};
   3 use std::io::{Read, Write};
   4 use std::time::Duration;
   5
   6 use openssl::sha;
   7 use std::sync::Mutex;
   8
   9 use std::fs::File;
  10 use std::os::unix::io::AsRawFd;
  11
  12 use crate::tools;
  13
  14 pub struct GarbageCollectionStatus {
  15     pub used_bytes: usize,
  16     pub used_chunks: usize,
  17     pub disk_bytes: usize,
  18     pub disk_chunks: usize,
  19 }
  20
  21 impl Default for GarbageCollectionStatus {
  22     fn default() -> Self {
  23         GarbageCollectionStatus {
  24             used_bytes: 0,
  25             used_chunks: 0,
  26             disk_bytes: 0,
  27             disk_chunks: 0,
  28         }
  29     }
  30 }
  31
  32 /// File system based chunk store
  33 pub struct ChunkStore {
  34     name: String, // used for error reporting
  35     pub (crate) base: PathBuf,
  36     chunk_dir: PathBuf,
  37     mutex: Mutex<bool>,
  38     _lockfile: File,
  39 }
  40
  41 // TODO: what about sysctl setting vm.vfs_cache_pressure (0 - 100) ?
  42
  43 pub fn verify_chunk_size(size: u64) -> Result<(), Error> {
  44
  45     static SIZES: [u64; 7] = [64*1024, 128*1024, 256*1024, 512*1024, 1024*1024, 2048*1024, 4096*1024];
  46
  47     if !SIZES.contains(&size) {
  48         bail!("Got unsupported chunk size '{}'", size);
  49     }
  50     Ok(())
  51 }
  52
  53 fn digest_to_prefix(digest: &[u8]) -> PathBuf {
  54
  55     let mut buf = Vec::<u8>::with_capacity(2+1+2+1);
  56
  57     const HEX_CHARS: &'static [u8; 16] = b"0123456789abcdef";
  58
  59     buf.push(HEX_CHARS[(digest[0] as usize) >> 4]);
  60     buf.push(HEX_CHARS[(digest[0] as usize) &0xf]);
  61     buf.push(HEX_CHARS[(digest[1] as usize) >> 4]);
  62     buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]);
  63     buf.push('/' as u8);
  64
  65     let path = unsafe { String::from_utf8_unchecked(buf)};
  66
  67     path.into()
  68 }
  69
  70 impl ChunkStore {
  71
  72     fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf {
  73
  74         let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref());
  75         chunk_dir.push(".chunks");
  76
  77         chunk_dir
  78     }
  79
  80     pub fn create<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
  81
  82         let base: PathBuf = path.into();
  83
  84         if !base.is_absolute() {
  85             bail!("expected absolute path - got {:?}", base);
  86         }
  87
  88         let chunk_dir = Self::chunk_dir(&base);
  89
  90         if let Err(err) = std::fs::create_dir(&base) {
  91             bail!("unable to create chunk store '{}' at {:?} - {}", name, base, err);
  92         }
  93
  94         if let Err(err) = std::fs::create_dir(&chunk_dir) {
  95             bail!("unable to create chunk store '{}' subdir {:?} - {}", name, chunk_dir, err);
  96         }
  97
  98         // create 64*1024 subdirs
  99         let mut last_percentage = 0;
 100
 101         for i in 0..64*1024 {
 102             let mut l1path = chunk_dir.clone();
 103             l1path.push(format!("{:04x}", i));
 104             if let Err(err) = std::fs::create_dir(&l1path) {
 105                 bail!("unable to create chunk store '{}' subdir {:?} - {}", name, l1path, err);
 106             }
 107             let percentage = (i*100)/(64*1024);
 108             if percentage != last_percentage {
 109                 eprintln!("Percentage done: {}", percentage);
 110                 last_percentage = percentage;
 111             }
 112         }
 113
 114         Self::open(name, base)
 115     }
 116
 117     pub fn open<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
 118
 119         let base: PathBuf = path.into();
 120
 121         if !base.is_absolute() {
 122             bail!("expected absolute path - got {:?}", base);
 123         }
 124
 125         let chunk_dir = Self::chunk_dir(&base);
 126
 127         if let Err(err) = std::fs::metadata(&chunk_dir) {
 128             bail!("unable to open chunk store '{}' at {:?} - {}", name, chunk_dir, err);
 129         }
 130
 131         let mut lockfile_path = base.clone();
 132         lockfile_path.push(".lock");
 133
 134         // make sure only one process/thread/task can use it
 135         let lockfile = tools::open_file_locked(
 136             lockfile_path, Duration::from_secs(10))?;
 137
 138         Ok(ChunkStore {
 139             name: name.to_owned(),
 140             base,
 141             chunk_dir,
 142             _lockfile: lockfile,
 143             mutex: Mutex::new(false)
 144         })
 145     }
 146
 147     pub fn touch_chunk(&self, digest:&[u8]) -> Result<(), Error> {
 148
 149         let mut chunk_path = self.chunk_dir.clone();
 150         let prefix = digest_to_prefix(&digest);
 151         chunk_path.push(&prefix);
 152         let digest_str = tools::digest_to_hex(&digest);
 153         chunk_path.push(&digest_str);
 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], buffer: &mut Vec<u8>) -> Result<(), Error> {
 177
 178         let mut chunk_path = self.chunk_dir.clone();
 179         let prefix = digest_to_prefix(&digest);
 180         chunk_path.push(&prefix);
 181         let digest_str = tools::digest_to_hex(&digest);
 182         chunk_path.push(&digest_str);
 183
 184         let f = std::fs::File::open(&chunk_path)?;
 185         let mut decoder = lz4::Decoder::new(f)?;
 186
 187         decoder.read_to_end(buffer)?;
 188
 189         Ok(())
 190     }
 191
 192     pub fn get_chunk_iterator(
 193         &self,
 194         print_percentage: bool,
 195     ) -> Result<
 196         impl Iterator<Item = Result<tools::fs::ReadDirEntry, Error>> + std::iter::FusedIterator,
 197         Error
 198     > {
 199         use nix::dir::Dir;
 200         use nix::fcntl::OFlag;
 201         use nix::sys::stat::Mode;
 202
 203         let base_handle = match Dir::open(
 204             &self.chunk_dir, OFlag::O_RDONLY, Mode::empty()) {
 205             Ok(h) => h,
 206             Err(err) => bail!("unable to open store '{}' chunk dir {:?} - {}",
 207                               self.name, self.chunk_dir, err),
 208         };
 209
 210         let mut verbose = true;
 211         let mut last_percentage = 0;
 212
 213         Ok((0..0x10000).filter_map(move |index| {
 214             if print_percentage {
 215                 let percentage = (index * 100) / 0x10000;
 216                 if last_percentage != percentage {
 217                     last_percentage = percentage;
 218                     eprintln!("percentage done: {}", percentage);
 219                 }
 220             }
 221             let subdir: &str = &format!("{:04x}", index);
 222             match tools::fs::read_subdir(base_handle.as_raw_fd(), subdir) {
 223                 Err(e) => {
 224                     if verbose {
 225                         eprintln!("Error iterating through chunks: {}", e);
 226                         verbose = false;
 227                     }
 228                     None
 229                 }
 230                 Ok(iter) => Some(iter),
 231             }
 232         })
 233         .flatten()
 234         .filter(|entry| {
 235             // Check that the file name is actually a hash! (64 hex digits)
 236             let entry = match entry {
 237                 Err(_) => return true, // pass errors onwards
 238                 Ok(ref entry) => entry,
 239             };
 240             let bytes = entry.file_name().to_bytes();
 241             if bytes.len() != 64 {
 242                 return false;
 243             }
 244             for b in bytes {
 245                 if !b.is_ascii_hexdigit() {
 246                     return false;
 247                 }
 248             }
 249             true
 250         }))
 251     }
 252
 253     pub fn sweep_unused_chunks(&self, status: &mut GarbageCollectionStatus) -> Result<(), Error> {
 254         use nix::sys::stat::fstatat;
 255
 256         let now = unsafe { libc::time(std::ptr::null_mut()) };
 257
 258         for entry in self.get_chunk_iterator(true)? {
 259             let (dirfd, entry) = match entry {
 260                 Ok(entry) => (entry.parent_fd(), entry),
 261                 Err(_) => continue, // ignore errors
 262             };
 263
 264             let file_type = match entry.file_type() {
 265                 Some(file_type) => file_type,
 266                 None => bail!("unsupported file system type on chunk store '{}'", self.name),
 267             };
 268             if file_type != nix::dir::Type::File {
 269                 continue;
 270             }
 271
 272             let filename = entry.file_name();
 273             if let Ok(stat) = fstatat(dirfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
 274                 let age = now - stat.st_atime;
 275                 //println!("FOUND {}  {:?}", age/(3600*24), filename);
 276                 if age/(3600*24) >= 2 {
 277                     println!("UNLINK {}  {:?}", age/(3600*24), filename);
 278                     let res = unsafe { libc::unlinkat(dirfd, filename.as_ptr(), 0) };
 279                     if res != 0 {
 280                         let err = nix::Error::last();
 281                         bail!(
 282                             "unlink chunk {:?} failed on store '{}' - {}",
 283                             filename,
 284                             self.name,
 285                             err,
 286                         );
 287                     }
 288                 } else {
 289                     status.disk_chunks += 1;
 290                     status.disk_bytes += stat.st_size as usize;
 291
 292                 }
 293             }
 294         }
 295         Ok(())
 296     }
 297
 298     pub fn insert_chunk(&self, chunk: &[u8]) -> Result<(bool, [u8; 32], u64), Error> {
 299
 300         // fixme: use Sha512/256 when available
 301         let digest = sha::sha256(chunk);
 302         let (new, csize) = self.insert_chunk_noverify(&digest, chunk)?;
 303         Ok((new, digest, csize))
 304     }
 305
 306     pub fn insert_chunk_noverify(
 307         &self,
 308         digest: &[u8; 32],
 309         chunk: &[u8],
 310     ) -> Result<(bool, u64), Error> {
 311
 312         //println!("DIGEST {}", tools::digest_to_hex(&digest));
 313
 314         let mut chunk_path = self.chunk_dir.clone();
 315         let prefix = digest_to_prefix(digest);
 316         chunk_path.push(&prefix);
 317         let digest_str = tools::digest_to_hex(digest);
 318         chunk_path.push(&digest_str);
 319
 320         let lock = self.mutex.lock();
 321
 322         if let Ok(metadata) = std::fs::metadata(&chunk_path) {
 323             if metadata.is_file() {
 324                 return Ok((true, metadata.len()));
 325             } else {
 326                 bail!("Got unexpected file type on store '{}' for chunk {}", self.name, digest_str);
 327             }
 328         }
 329
 330         let mut tmp_path = chunk_path.clone();
 331         tmp_path.set_extension("tmp");
 332
 333         let f = std::fs::File::create(&tmp_path)?;
 334
 335         // fixme: what is the fasted lz4 encoder available (see lzbench)?
 336         let mut encoder = lz4::EncoderBuilder::new().level(1).build(f)?;
 337
 338         encoder.write_all(chunk)?;
 339         let (f, encode_result) = encoder.finish();
 340         encode_result?;
 341
 342         if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) {
 343             if let Err(_) = std::fs::remove_file(&tmp_path)  { /* ignore */ }
 344             bail!(
 345                 "Atomic rename on store '{}' failed for chunk {} - {}",
 346                 self.name,
 347                 digest_str,
 348                 err,
 349             );
 350         }
 351
 352         // fixme: is there a better way to get the compressed size?
 353         let stat = nix::sys::stat::fstat(f.as_raw_fd())?;
 354         let compressed_size = stat.st_size as u64;
 355
 356         //println!("PATH {:?}", chunk_path);
 357
 358         drop(lock);
 359
 360         Ok((false, compressed_size))
 361     }
 362
 363     pub fn relative_path(&self, path: &Path) -> PathBuf {
 364
 365         let mut full_path = self.base.clone();
 366         full_path.push(path);
 367         full_path
 368     }
 369
 370     pub fn base_path(&self) -> PathBuf {
 371         self.base.clone()
 372     }
 373 }
 374
 375
 376 #[test]
 377 fn test_chunk_store1() {
 378
 379     let mut path = std::fs::canonicalize(".").unwrap(); // we need absulute path
 380     path.push(".testdir");
 381
 382     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 383
 384     let chunk_store = ChunkStore::open("test", &path);
 385     assert!(chunk_store.is_err());
 386
 387     let chunk_store = ChunkStore::create("test", &path).unwrap();
 388     let (exists, _, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 389     assert!(!exists);
 390
 391     let (exists, _, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 392     assert!(exists);
 393
 394
 395     let chunk_store = ChunkStore::create("test", &path);
 396     assert!(chunk_store.is_err());
 397
 398     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 399 }