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