src/backup/chunk_store.rs

   1 use anyhow::{bail, format_err, Error};
   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
   8 use proxmox::tools::fs::{CreateOptions, create_path, create_dir};
   9
  10 use crate::tools;
  11 use crate::api2::types::GarbageCollectionStatus;
  12
  13 use super::DataBlob;
  14 use crate::server::WorkerTask;
  15
  16 /// File system based chunk store
  17 pub struct ChunkStore {
  18     name: String, // used for error reporting
  19     pub (crate) base: PathBuf,
  20     chunk_dir: PathBuf,
  21     mutex: Mutex<bool>,
  22     locker: Arc<Mutex<tools::ProcessLocker>>,
  23 }
  24
  25 // TODO: what about sysctl setting vm.vfs_cache_pressure (0 - 100) ?
  26
  27 pub fn verify_chunk_size(size: usize) -> Result<(), Error> {
  28
  29     static SIZES: [usize; 7] = [64*1024, 128*1024, 256*1024, 512*1024, 1024*1024, 2048*1024, 4096*1024];
  30
  31     if !SIZES.contains(&size) {
  32         bail!("Got unsupported chunk size '{}'", size);
  33     }
  34     Ok(())
  35 }
  36
  37 fn digest_to_prefix(digest: &[u8]) -> PathBuf {
  38
  39     let mut buf = Vec::<u8>::with_capacity(2+1+2+1);
  40
  41     const HEX_CHARS: &[u8; 16] = b"0123456789abcdef";
  42
  43     buf.push(HEX_CHARS[(digest[0] as usize) >> 4]);
  44     buf.push(HEX_CHARS[(digest[0] as usize) &0xf]);
  45     buf.push(HEX_CHARS[(digest[1] as usize) >> 4]);
  46     buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]);
  47     buf.push('/' as u8);
  48
  49     let path = unsafe { String::from_utf8_unchecked(buf)};
  50
  51     path.into()
  52 }
  53
  54 impl ChunkStore {
  55
  56     fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf {
  57
  58         let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref());
  59         chunk_dir.push(".chunks");
  60
  61         chunk_dir
  62     }
  63
  64     pub fn create<P>(name: &str, path: P, uid: nix::unistd::Uid, gid: nix::unistd::Gid) -> Result<Self, Error>
  65     where
  66         P: Into<PathBuf>,
  67     {
  68
  69         let base: PathBuf = path.into();
  70
  71         if !base.is_absolute() {
  72             bail!("expected absolute path - got {:?}", base);
  73         }
  74
  75         let chunk_dir = Self::chunk_dir(&base);
  76
  77         let options = CreateOptions::new()
  78             .owner(uid)
  79             .group(gid);
  80
  81         let default_options = CreateOptions::new();
  82
  83         match create_path(&base, Some(default_options.clone()), Some(options.clone())) {
  84             Err(err) => bail!("unable to create chunk store '{}' at {:?} - {}", name, base, err),
  85             Ok(res) => if ! res  { nix::unistd::chown(&base, Some(uid), Some(gid))? },
  86         }
  87
  88         if let Err(err) = create_dir(&chunk_dir, options.clone()) {
  89             bail!("unable to create chunk store '{}' subdir {:?} - {}", name, chunk_dir, err);
  90         }
  91
  92         // create lock file with correct owner/group
  93         let lockfile_path = Self::lockfile_path(&base);
  94         proxmox::tools::fs::replace_file(lockfile_path, b"", options.clone())?;
  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) = create_dir(&l1path, options.clone()) {
 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
 113         Self::open(name, base)
 114     }
 115
 116     fn lockfile_path<P: Into<PathBuf>>(base: P) -> PathBuf {
 117         let base: PathBuf = base.into();
 118
 119         let mut lockfile_path = base.clone();
 120         lockfile_path.push(".lock");
 121
 122         lockfile_path
 123     }
 124
 125     pub fn open<P: Into<PathBuf>>(name: &str, base: P) -> Result<Self, Error> {
 126
 127         let base: PathBuf = base.into();
 128
 129         if !base.is_absolute() {
 130             bail!("expected absolute path - got {:?}", base);
 131         }
 132
 133         let chunk_dir = Self::chunk_dir(&base);
 134
 135         if let Err(err) = std::fs::metadata(&chunk_dir) {
 136             bail!("unable to open chunk store '{}' at {:?} - {}", name, chunk_dir, err);
 137         }
 138
 139         let lockfile_path = Self::lockfile_path(&base);
 140
 141         let locker = tools::ProcessLocker::new(&lockfile_path)?;
 142
 143         Ok(ChunkStore {
 144             name: name.to_owned(),
 145             base,
 146             chunk_dir,
 147             locker,
 148             mutex: Mutex::new(false)
 149         })
 150     }
 151
 152     pub fn touch_chunk(&self, digest: &[u8; 32]) -> Result<(), Error> {
 153         self.cond_touch_chunk(digest, true)?;
 154         Ok(())
 155     }
 156
 157     pub fn cond_touch_chunk(&self, digest: &[u8; 32], fail_if_not_exist: bool) -> Result<bool, Error> {
 158
 159         let (chunk_path, _digest_str) = self.chunk_path(digest);
 160
 161         const UTIME_NOW: i64 = (1 << 30) - 1;
 162         const UTIME_OMIT: i64 = (1 << 30) - 2;
 163
 164         let times: [libc::timespec; 2] = [
 165             libc::timespec { tv_sec: 0, tv_nsec: UTIME_NOW },
 166             libc::timespec { tv_sec: 0, tv_nsec: UTIME_OMIT }
 167         ];
 168
 169         use nix::NixPath;
 170
 171         let res = chunk_path.with_nix_path(|cstr| unsafe {
 172             let tmp = libc::utimensat(-1, cstr.as_ptr(), &times[0], libc::AT_SYMLINK_NOFOLLOW);
 173             nix::errno::Errno::result(tmp)
 174         })?;
 175
 176         if let Err(err) = res {
 177             if !fail_if_not_exist && err.as_errno() == Some(nix::errno::Errno::ENOENT) {
 178                 return Ok(false);
 179             }
 180
 181             bail!("update atime failed for chunk {:?} - {}", chunk_path, err);
 182         }
 183
 184         Ok(true)
 185     }
 186
 187     pub fn get_chunk_iterator(
 188         &self,
 189     ) -> Result<
 190         impl Iterator<Item = (Result<tools::fs::ReadDirEntry, Error>, usize)> + std::iter::FusedIterator,
 191         Error
 192     > {
 193         use nix::dir::Dir;
 194         use nix::fcntl::OFlag;
 195         use nix::sys::stat::Mode;
 196
 197         let base_handle = Dir::open(&self.chunk_dir, OFlag::O_RDONLY, Mode::empty())
 198             .map_err(|err| {
 199                 format_err!(
 200                     "unable to open store '{}' chunk dir {:?} - {}",
 201                     self.name,
 202                     self.chunk_dir,
 203                     err,
 204                 )
 205             })?;
 206
 207         let mut done = false;
 208         let mut inner: Option<tools::fs::ReadDir> = None;
 209         let mut at = 0;
 210         let mut percentage = 0;
 211         Ok(std::iter::from_fn(move || {
 212             if done {
 213                 return None;
 214             }
 215
 216             loop {
 217                 if let Some(ref mut inner) = inner {
 218                     match inner.next() {
 219                         Some(Ok(entry)) => {
 220                             // skip files if they're not a hash
 221                             let bytes = entry.file_name().to_bytes();
 222                             if bytes.len() != 64 {
 223                                 continue;
 224                             }
 225                             if !bytes.iter().all(u8::is_ascii_hexdigit) {
 226                                 continue;
 227                             }
 228                             return Some((Ok(entry), percentage));
 229                         }
 230                         Some(Err(err)) => {
 231                             // stop after first error
 232                             done = true;
 233                             // and pass the error through:
 234                             return Some((Err(err), percentage));
 235                         }
 236                         None => (), // open next directory
 237                     }
 238                 }
 239
 240                 inner = None;
 241
 242                 if at == 0x10000 {
 243                     done = true;
 244                     return None;
 245                 }
 246
 247                 let subdir: &str = &format!("{:04x}", at);
 248                 percentage = (at * 100) / 0x10000;
 249                 at += 1;
 250                 match tools::fs::read_subdir(base_handle.as_raw_fd(), subdir) {
 251                     Ok(dir) => {
 252                         inner = Some(dir);
 253                         // start reading:
 254                         continue;
 255                     }
 256                     Err(ref err) if err.as_errno() == Some(nix::errno::Errno::ENOENT) => {
 257                         // non-existing directories are okay, just keep going:
 258                         continue;
 259                     }
 260                     Err(err) => {
 261                         // other errors are fatal, so end our iteration
 262                         done = true;
 263                         // and pass the error through:
 264                         return Some((Err(format_err!("unable to read subdir '{}' - {}", subdir, err)), percentage));
 265                     }
 266                 }
 267             }
 268         }).fuse())
 269     }
 270
 271     pub fn oldest_writer(&self) -> Option<i64> {
 272         tools::ProcessLocker::oldest_shared_lock(self.locker.clone())
 273     }
 274
 275     pub fn sweep_unused_chunks(
 276         &self,
 277         oldest_writer: i64,
 278         status: &mut GarbageCollectionStatus,
 279         worker: &WorkerTask,
 280     ) -> Result<(), Error> {
 281         use nix::sys::stat::fstatat;
 282
 283         let now = unsafe { libc::time(std::ptr::null_mut()) };
 284
 285         let mut min_atime = now - 3600*24; // at least 24h (see mount option relatime)
 286
 287         if oldest_writer < min_atime {
 288             min_atime = oldest_writer;
 289         }
 290
 291         min_atime -= 300; // add 5 mins gap for safety
 292
 293         let mut last_percentage = 0;
 294         let mut chunk_count = 0;
 295
 296         for (entry, percentage) in self.get_chunk_iterator()? {
 297             if last_percentage != percentage {
 298                 last_percentage = percentage;
 299                 worker.log(format!("percentage done: {}, chunk count: {}", percentage, chunk_count));
 300             }
 301
 302             worker.fail_on_abort()?;
 303             tools::fail_on_shutdown()?;
 304
 305             let (dirfd, entry) = match entry {
 306                 Ok(entry) => (entry.parent_fd(), entry),
 307                 Err(err) => bail!("chunk iterator on chunk store '{}' failed - {}", self.name, err),
 308             };
 309
 310             let file_type = match entry.file_type() {
 311                 Some(file_type) => file_type,
 312                 None => bail!("unsupported file system type on chunk store '{}'", self.name),
 313             };
 314             if file_type != nix::dir::Type::File {
 315                 continue;
 316             }
 317
 318             chunk_count += 1;
 319
 320             let filename = entry.file_name();
 321
 322             let lock = self.mutex.lock();
 323
 324             if let Ok(stat) = fstatat(dirfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
 325                 if stat.st_atime < min_atime {
 326                     //let age = now - stat.st_atime;
 327                     //println!("UNLINK {}  {:?}", age/(3600*24), filename);
 328                     let res = unsafe { libc::unlinkat(dirfd, filename.as_ptr(), 0) };
 329                     if res != 0 {
 330                         let err = nix::Error::last();
 331                         bail!(
 332                             "unlink chunk {:?} failed on store '{}' - {}",
 333                             filename,
 334                             self.name,
 335                             err,
 336                         );
 337                     }
 338                     status.removed_chunks += 1;
 339                     status.removed_bytes += stat.st_size as u64;
 340                 } else {
 341                     if stat.st_atime < oldest_writer {
 342                         status.pending_chunks += 1;
 343                         status.pending_bytes += stat.st_size as u64;
 344                     } else {
 345                         status.disk_chunks += 1;
 346                         status.disk_bytes += stat.st_size as u64;
 347                      }
 348                 }
 349             }
 350             drop(lock);
 351         }
 352
 353         Ok(())
 354     }
 355
 356     pub fn insert_chunk(
 357         &self,
 358         chunk: &DataBlob,
 359         digest: &[u8; 32],
 360     ) -> Result<(bool, u64), Error> {
 361
 362         //println!("DIGEST {}", proxmox::tools::digest_to_hex(digest));
 363
 364         let (chunk_path, digest_str) = self.chunk_path(digest);
 365
 366         let lock = self.mutex.lock();
 367
 368         if let Ok(metadata) = std::fs::metadata(&chunk_path) {
 369             if metadata.is_file() {
 370                 return Ok((true, metadata.len()));
 371             } else {
 372                 bail!("Got unexpected file type on store '{}' for chunk {}", self.name, digest_str);
 373             }
 374         }
 375
 376         let mut tmp_path = chunk_path.clone();
 377         tmp_path.set_extension("tmp");
 378
 379         let mut file = std::fs::File::create(&tmp_path)?;
 380
 381         let raw_data = chunk.raw_data();
 382         let encoded_size = raw_data.len() as u64;
 383
 384         file.write_all(raw_data)?;
 385
 386         if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) {
 387             if let Err(_) = std::fs::remove_file(&tmp_path)  { /* ignore */ }
 388             bail!(
 389                 "Atomic rename on store '{}' failed for chunk {} - {}",
 390                 self.name,
 391                 digest_str,
 392                 err,
 393             );
 394         }
 395
 396         drop(lock);
 397
 398         Ok((false, encoded_size))
 399     }
 400
 401     pub fn chunk_path(&self, digest:&[u8; 32]) -> (PathBuf, String) {
 402         let mut chunk_path = self.chunk_dir.clone();
 403         let prefix = digest_to_prefix(digest);
 404         chunk_path.push(&prefix);
 405         let digest_str = proxmox::tools::digest_to_hex(digest);
 406         chunk_path.push(&digest_str);
 407         (chunk_path, digest_str)
 408     }
 409
 410     pub fn relative_path(&self, path: &Path) -> PathBuf {
 411
 412         let mut full_path = self.base.clone();
 413         full_path.push(path);
 414         full_path
 415     }
 416
 417     pub fn name(&self) -> &str {
 418         &self.name
 419     }
 420
 421     pub fn base_path(&self) -> PathBuf {
 422         self.base.clone()
 423     }
 424
 425     pub fn try_shared_lock(&self) -> Result<tools::ProcessLockSharedGuard, Error> {
 426         tools::ProcessLocker::try_shared_lock(self.locker.clone())
 427     }
 428
 429     pub fn try_exclusive_lock(&self) -> Result<tools::ProcessLockExclusiveGuard, Error> {
 430         tools::ProcessLocker::try_exclusive_lock(self.locker.clone())
 431     }
 432 }
 433
 434
 435 #[test]
 436 fn test_chunk_store1() {
 437
 438     let mut path = std::fs::canonicalize(".").unwrap(); // we need absulute path
 439     path.push(".testdir");
 440
 441     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 442
 443     let chunk_store = ChunkStore::open("test", &path);
 444     assert!(chunk_store.is_err());
 445
 446     let user = nix::unistd::User::from_uid(nix::unistd::Uid::current()).unwrap().unwrap();
 447     let chunk_store = ChunkStore::create("test", &path, user.uid, user.gid).unwrap();
 448
 449     let (chunk, digest) = super::DataChunkBuilder::new(&[0u8, 1u8]).build().unwrap();
 450
 451     let (exists, _) = chunk_store.insert_chunk(&chunk, &digest).unwrap();
 452     assert!(!exists);
 453
 454     let (exists, _) = chunk_store.insert_chunk(&chunk, &digest).unwrap();
 455     assert!(exists);
 456
 457
 458     let chunk_store = ChunkStore::create("test", &path, user.uid, user.gid);
 459     assert!(chunk_store.is_err());
 460
 461     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 462 }