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 use serde_derive::Serialize;
   8
   9 use openssl::sha;
  10
  11 use crate::tools;
  12
  13 #[derive(Clone, Serialize)]
  14 pub struct GarbageCollectionStatus {
  15     pub upid: Option<String>,
  16     pub used_bytes: usize,
  17     pub used_chunks: usize,
  18     pub disk_bytes: usize,
  19     pub disk_chunks: usize,
  20 }
  21
  22 impl Default for GarbageCollectionStatus {
  23     fn default() -> Self {
  24         GarbageCollectionStatus {
  25             upid: None,
  26             used_bytes: 0,
  27             used_chunks: 0,
  28             disk_bytes: 0,
  29             disk_chunks: 0,
  30         }
  31     }
  32 }
  33
  34 /// File system based chunk store
  35 pub struct ChunkStore {
  36     name: String, // used for error reporting
  37     pub (crate) base: PathBuf,
  38     chunk_dir: PathBuf,
  39     mutex: Mutex<bool>,
  40     locker: Arc<Mutex<tools::ProcessLocker>>,
  41 }
  42
  43 // TODO: what about sysctl setting vm.vfs_cache_pressure (0 - 100) ?
  44
  45 pub fn verify_chunk_size(size: usize) -> Result<(), Error> {
  46
  47     static SIZES: [usize; 7] = [64*1024, 128*1024, 256*1024, 512*1024, 1024*1024, 2048*1024, 4096*1024];
  48
  49     if !SIZES.contains(&size) {
  50         bail!("Got unsupported chunk size '{}'", size);
  51     }
  52     Ok(())
  53 }
  54
  55 fn digest_to_prefix(digest: &[u8]) -> PathBuf {
  56
  57     let mut buf = Vec::<u8>::with_capacity(2+1+2+1);
  58
  59     const HEX_CHARS: &'static [u8; 16] = b"0123456789abcdef";
  60
  61     buf.push(HEX_CHARS[(digest[0] as usize) >> 4]);
  62     buf.push(HEX_CHARS[(digest[0] as usize) &0xf]);
  63     buf.push(HEX_CHARS[(digest[1] as usize) >> 4]);
  64     buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]);
  65     buf.push('/' as u8);
  66
  67     let path = unsafe { String::from_utf8_unchecked(buf)};
  68
  69     path.into()
  70 }
  71
  72 impl ChunkStore {
  73
  74     fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf {
  75
  76         let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref());
  77         chunk_dir.push(".chunks");
  78
  79         chunk_dir
  80     }
  81
  82     pub fn create<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
  83
  84         let base: PathBuf = path.into();
  85
  86         if !base.is_absolute() {
  87             bail!("expected absolute path - got {:?}", base);
  88         }
  89
  90         let chunk_dir = Self::chunk_dir(&base);
  91
  92         if let Err(err) = std::fs::create_dir(&base) {
  93             bail!("unable to create chunk store '{}' at {:?} - {}", name, base, err);
  94         }
  95
  96         if let Err(err) = std::fs::create_dir(&chunk_dir) {
  97             bail!("unable to create chunk store '{}' subdir {:?} - {}", name, chunk_dir, err);
  98         }
  99
 100         // create 64*1024 subdirs
 101         let mut last_percentage = 0;
 102
 103         for i in 0..64*1024 {
 104             let mut l1path = chunk_dir.clone();
 105             l1path.push(format!("{:04x}", i));
 106             if let Err(err) = std::fs::create_dir(&l1path) {
 107                 bail!("unable to create chunk store '{}' subdir {:?} - {}", name, l1path, err);
 108             }
 109             let percentage = (i*100)/(64*1024);
 110             if percentage != last_percentage {
 111                 eprintln!("Percentage done: {}", percentage);
 112                 last_percentage = percentage;
 113             }
 114         }
 115
 116         Self::open(name, base)
 117     }
 118
 119     pub fn open<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
 120
 121         let base: PathBuf = path.into();
 122
 123         if !base.is_absolute() {
 124             bail!("expected absolute path - got {:?}", base);
 125         }
 126
 127         let chunk_dir = Self::chunk_dir(&base);
 128
 129         if let Err(err) = std::fs::metadata(&chunk_dir) {
 130             bail!("unable to open chunk store '{}' at {:?} - {}", name, chunk_dir, err);
 131         }
 132
 133         let mut lockfile_path = base.clone();
 134         lockfile_path.push(".lock");
 135
 136         let locker = tools::ProcessLocker::new(&lockfile_path)?;
 137
 138         Ok(ChunkStore {
 139             name: name.to_owned(),
 140             base,
 141             chunk_dir,
 142             locker,
 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         buffer.clear();
 185         let f = std::fs::File::open(&chunk_path)?;
 186         let mut decoder = zstd::stream::Decoder::new(f)?;
 187
 188         decoder.read_to_end(buffer)?;
 189
 190         Ok(())
 191     }
 192
 193     pub fn get_chunk_iterator(
 194         &self,
 195         print_percentage: bool,
 196     ) -> Result<
 197         impl Iterator<Item = Result<tools::fs::ReadDirEntry, Error>> + std::iter::FusedIterator,
 198         Error
 199     > {
 200         use nix::dir::Dir;
 201         use nix::fcntl::OFlag;
 202         use nix::sys::stat::Mode;
 203
 204         let base_handle = match Dir::open(
 205             &self.chunk_dir, OFlag::O_RDONLY, Mode::empty()) {
 206             Ok(h) => h,
 207             Err(err) => bail!("unable to open store '{}' chunk dir {:?} - {}",
 208                               self.name, self.chunk_dir, err),
 209         };
 210
 211         let mut verbose = true;
 212         let mut last_percentage = 0;
 213
 214         Ok((0..0x10000).filter_map(move |index| {
 215             if print_percentage {
 216                 let percentage = (index * 100) / 0x10000;
 217                 if last_percentage != percentage {
 218                     last_percentage = percentage;
 219                     eprintln!("percentage done: {}", percentage);
 220                 }
 221             }
 222             let subdir: &str = &format!("{:04x}", index);
 223             match tools::fs::read_subdir(base_handle.as_raw_fd(), subdir) {
 224                 Err(e) => {
 225                     if verbose {
 226                         eprintln!("Error iterating through chunks: {}", e);
 227                         verbose = false;
 228                     }
 229                     None
 230                 }
 231                 Ok(iter) => Some(iter),
 232             }
 233         })
 234         .flatten()
 235         .filter(|entry| {
 236             // Check that the file name is actually a hash! (64 hex digits)
 237             let entry = match entry {
 238                 Err(_) => return true, // pass errors onwards
 239                 Ok(ref entry) => entry,
 240             };
 241             let bytes = entry.file_name().to_bytes();
 242             if bytes.len() != 64 {
 243                 return false;
 244             }
 245             for b in bytes {
 246                 if !b.is_ascii_hexdigit() {
 247                     return false;
 248                 }
 249             }
 250             true
 251         }))
 252     }
 253
 254     pub fn oldest_writer(&self) -> Option<i64> {
 255         tools::ProcessLocker::oldest_shared_lock(self.locker.clone())
 256     }
 257
 258     pub fn sweep_unused_chunks(
 259         &self,
 260         oldest_writer: Option<i64>,
 261         status: &mut GarbageCollectionStatus
 262     ) -> Result<(), Error> {
 263         use nix::sys::stat::fstatat;
 264
 265         let now = unsafe { libc::time(std::ptr::null_mut()) };
 266
 267         let mut min_atime = now - 3600*24; // at least 24h (see mount option relatime)
 268
 269         if let Some(stamp) = oldest_writer {
 270             if stamp < min_atime {
 271                 min_atime = stamp;
 272             }
 273         }
 274
 275         min_atime -= 300; // add 5 mins gap for safety
 276
 277         for entry in self.get_chunk_iterator(true)? {
 278
 279             tools::fail_on_shutdown()?;
 280
 281             let (dirfd, entry) = match entry {
 282                 Ok(entry) => (entry.parent_fd(), entry),
 283                 Err(_) => continue, // ignore errors
 284             };
 285
 286             let file_type = match entry.file_type() {
 287                 Some(file_type) => file_type,
 288                 None => bail!("unsupported file system type on chunk store '{}'", self.name),
 289             };
 290             if file_type != nix::dir::Type::File {
 291                 continue;
 292             }
 293
 294             let filename = entry.file_name();
 295
 296             let lock = self.mutex.lock();
 297
 298             if let Ok(stat) = fstatat(dirfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
 299                 let age = now - stat.st_atime;
 300                 //println!("FOUND {}  {:?}", age/(3600*24), filename);
 301                 if stat.st_atime < min_atime {
 302                     println!("UNLINK {}  {:?}", age/(3600*24), filename);
 303                     let res = unsafe { libc::unlinkat(dirfd, filename.as_ptr(), 0) };
 304                     if res != 0 {
 305                         let err = nix::Error::last();
 306                         bail!(
 307                             "unlink chunk {:?} failed on store '{}' - {}",
 308                             filename,
 309                             self.name,
 310                             err,
 311                         );
 312                     }
 313                 } else {
 314                     status.disk_chunks += 1;
 315                     status.disk_bytes += stat.st_size as usize;
 316                 }
 317             }
 318             drop(lock);
 319         }
 320         Ok(())
 321     }
 322
 323     pub fn insert_chunk(&self, chunk: &[u8]) -> Result<(bool, [u8; 32], u64), Error> {
 324
 325         // fixme: use Sha512/256 when available
 326         let digest = sha::sha256(chunk);
 327         let (new, csize) = self.insert_chunk_noverify(&digest, chunk)?;
 328         Ok((new, digest, csize))
 329     }
 330
 331     pub fn insert_chunk_noverify(
 332         &self,
 333         digest: &[u8; 32],
 334         chunk: &[u8],
 335     ) -> Result<(bool, u64), Error> {
 336
 337         //println!("DIGEST {}", tools::digest_to_hex(&digest));
 338
 339         let mut chunk_path = self.chunk_dir.clone();
 340         let prefix = digest_to_prefix(digest);
 341         chunk_path.push(&prefix);
 342         let digest_str = tools::digest_to_hex(digest);
 343         chunk_path.push(&digest_str);
 344
 345         let lock = self.mutex.lock();
 346
 347         if let Ok(metadata) = std::fs::metadata(&chunk_path) {
 348             if metadata.is_file() {
 349                 return Ok((true, metadata.len()));
 350             } else {
 351                 bail!("Got unexpected file type on store '{}' for chunk {}", self.name, digest_str);
 352             }
 353         }
 354
 355         let mut tmp_path = chunk_path.clone();
 356         tmp_path.set_extension("tmp");
 357
 358         let f = std::fs::File::create(&tmp_path)?;
 359
 360         let mut encoder = zstd::stream::Encoder::new(f, 1)?;
 361
 362         encoder.write_all(chunk)?;
 363         let f = encoder.finish()?;
 364
 365         if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) {
 366             if let Err(_) = std::fs::remove_file(&tmp_path)  { /* ignore */ }
 367             bail!(
 368                 "Atomic rename on store '{}' failed for chunk {} - {}",
 369                 self.name,
 370                 digest_str,
 371                 err,
 372             );
 373         }
 374
 375         // fixme: is there a better way to get the compressed size?
 376         let stat = nix::sys::stat::fstat(f.as_raw_fd())?;
 377         let compressed_size = stat.st_size as u64;
 378
 379         //println!("PATH {:?}", chunk_path);
 380
 381         drop(lock);
 382
 383         Ok((false, compressed_size))
 384     }
 385
 386     pub fn relative_path(&self, path: &Path) -> PathBuf {
 387
 388         let mut full_path = self.base.clone();
 389         full_path.push(path);
 390         full_path
 391     }
 392
 393     pub fn base_path(&self) -> PathBuf {
 394         self.base.clone()
 395     }
 396
 397     pub fn try_shared_lock(&self) -> Result<tools::ProcessLockSharedGuard, Error> {
 398         tools::ProcessLocker::try_shared_lock(self.locker.clone())
 399     }
 400
 401     pub fn try_exclusive_lock(&self) -> Result<tools::ProcessLockExclusiveGuard, Error> {
 402         tools::ProcessLocker::try_exclusive_lock(self.locker.clone())
 403     }
 404 }
 405
 406
 407 #[test]
 408 fn test_chunk_store1() {
 409
 410     let mut path = std::fs::canonicalize(".").unwrap(); // we need absulute path
 411     path.push(".testdir");
 412
 413     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 414
 415     let chunk_store = ChunkStore::open("test", &path);
 416     assert!(chunk_store.is_err());
 417
 418     let chunk_store = ChunkStore::create("test", &path).unwrap();
 419     let (exists, _, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 420     assert!(!exists);
 421
 422     let (exists, _, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 423     assert!(exists);
 424
 425
 426     let chunk_store = ChunkStore::create("test", &path);
 427     assert!(chunk_store.is_err());
 428
 429     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 430 }