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::DataBlob;
  11 use crate::server::WorkerTask;
  12
  13 #[derive(Clone, Serialize)]
  14 pub struct GarbageCollectionStatus {
  15     pub upid: Option<String>,
  16     pub index_file_count: usize,
  17     pub index_data_bytes: u64,
  18     pub disk_bytes: u64,
  19     pub disk_chunks: usize,
  20     pub removed_bytes: u64,
  21     pub removed_chunks: usize,
  22 }
  23
  24 impl Default for GarbageCollectionStatus {
  25     fn default() -> Self {
  26         GarbageCollectionStatus {
  27             upid: None,
  28             index_file_count: 0,
  29             index_data_bytes: 0,
  30             disk_bytes: 0,
  31             disk_chunks: 0,
  32             removed_bytes: 0,
  33             removed_chunks: 0,
  34         }
  35     }
  36 }
  37
  38 /// File system based chunk store
  39 pub struct ChunkStore {
  40     name: String, // used for error reporting
  41     pub (crate) base: PathBuf,
  42     chunk_dir: PathBuf,
  43     mutex: Mutex<bool>,
  44     locker: Arc<Mutex<tools::ProcessLocker>>,
  45 }
  46
  47 // TODO: what about sysctl setting vm.vfs_cache_pressure (0 - 100) ?
  48
  49 pub fn verify_chunk_size(size: usize) -> Result<(), Error> {
  50
  51     static SIZES: [usize; 7] = [64*1024, 128*1024, 256*1024, 512*1024, 1024*1024, 2048*1024, 4096*1024];
  52
  53     if !SIZES.contains(&size) {
  54         bail!("Got unsupported chunk size '{}'", size);
  55     }
  56     Ok(())
  57 }
  58
  59 fn digest_to_prefix(digest: &[u8]) -> PathBuf {
  60
  61     let mut buf = Vec::<u8>::with_capacity(2+1+2+1);
  62
  63     const HEX_CHARS: &'static [u8; 16] = b"0123456789abcdef";
  64
  65     buf.push(HEX_CHARS[(digest[0] as usize) >> 4]);
  66     buf.push(HEX_CHARS[(digest[0] as usize) &0xf]);
  67     buf.push(HEX_CHARS[(digest[1] as usize) >> 4]);
  68     buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]);
  69     buf.push('/' as u8);
  70
  71     let path = unsafe { String::from_utf8_unchecked(buf)};
  72
  73     path.into()
  74 }
  75
  76 impl ChunkStore {
  77
  78     fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf {
  79
  80         let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref());
  81         chunk_dir.push(".chunks");
  82
  83         chunk_dir
  84     }
  85
  86     pub fn create<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
  87
  88         let base: PathBuf = path.into();
  89
  90         if !base.is_absolute() {
  91             bail!("expected absolute path - got {:?}", base);
  92         }
  93
  94         let chunk_dir = Self::chunk_dir(&base);
  95
  96         if let Err(err) = std::fs::create_dir(&base) {
  97             bail!("unable to create chunk store '{}' at {:?} - {}", name, base, err);
  98         }
  99
 100         if let Err(err) = std::fs::create_dir(&chunk_dir) {
 101             bail!("unable to create chunk store '{}' subdir {:?} - {}", name, chunk_dir, err);
 102         }
 103
 104         // create 64*1024 subdirs
 105         let mut last_percentage = 0;
 106
 107         for i in 0..64*1024 {
 108             let mut l1path = chunk_dir.clone();
 109             l1path.push(format!("{:04x}", i));
 110             if let Err(err) = std::fs::create_dir(&l1path) {
 111                 bail!("unable to create chunk store '{}' subdir {:?} - {}", name, l1path, err);
 112             }
 113             let percentage = (i*100)/(64*1024);
 114             if percentage != last_percentage {
 115                 eprintln!("Percentage done: {}", percentage);
 116                 last_percentage = percentage;
 117             }
 118         }
 119
 120         Self::open(name, base)
 121     }
 122
 123     pub fn open<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
 124
 125         let base: PathBuf = path.into();
 126
 127         if !base.is_absolute() {
 128             bail!("expected absolute path - got {:?}", base);
 129         }
 130
 131         let chunk_dir = Self::chunk_dir(&base);
 132
 133         if let Err(err) = std::fs::metadata(&chunk_dir) {
 134             bail!("unable to open chunk store '{}' at {:?} - {}", name, chunk_dir, err);
 135         }
 136
 137         let mut lockfile_path = base.clone();
 138         lockfile_path.push(".lock");
 139
 140         let locker = tools::ProcessLocker::new(&lockfile_path)?;
 141
 142         Ok(ChunkStore {
 143             name: name.to_owned(),
 144             base,
 145             chunk_dir,
 146             locker,
 147             mutex: Mutex::new(false)
 148         })
 149     }
 150
 151     pub fn touch_chunk(&self, digest: &[u8; 32]) -> Result<(), Error> {
 152
 153         let (chunk_path, _digest_str) = self.chunk_path(digest);
 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; 32]) -> Result<DataBlob, Error> {
 177
 178         let (chunk_path, digest_str) = self.chunk_path(digest);
 179         let mut file = std::fs::File::open(&chunk_path)
 180             .map_err(|err| {
 181                 format_err!(
 182                     "store '{}', unable to read chunk '{}' - {}",
 183                     self.name,
 184                     digest_str,
 185                     err,
 186                 )
 187             })?;
 188
 189         DataBlob::load(&mut file)
 190     }
 191
 192     pub fn get_chunk_iterator(
 193         &self,
 194     ) -> Result<
 195         impl Iterator<Item = (Result<tools::fs::ReadDirEntry, Error>, usize)> + std::iter::FusedIterator,
 196         Error
 197     > {
 198         use nix::dir::Dir;
 199         use nix::fcntl::OFlag;
 200         use nix::sys::stat::Mode;
 201
 202         let base_handle = Dir::open(&self.chunk_dir, OFlag::O_RDONLY, Mode::empty())
 203             .map_err(|err| {
 204                 format_err!(
 205                     "unable to open store '{}' chunk dir {:?} - {}",
 206                     self.name,
 207                     self.chunk_dir,
 208                     err,
 209                 )
 210             })?;
 211
 212         let mut done = false;
 213         let mut inner: Option<tools::fs::ReadDir> = None;
 214         let mut at = 0;
 215         let mut percentage = 0;
 216         Ok(std::iter::from_fn(move || {
 217             if done {
 218                 return None;
 219             }
 220
 221             loop {
 222                 if let Some(ref mut inner) = inner {
 223                     match inner.next() {
 224                         Some(Ok(entry)) => {
 225                             // skip files if they're not a hash
 226                             let bytes = entry.file_name().to_bytes();
 227                             if bytes.len() != 64 {
 228                                 continue;
 229                             }
 230                             if !bytes.iter().all(u8::is_ascii_hexdigit) {
 231                                 continue;
 232                             }
 233                             return Some((Ok(entry), percentage));
 234                         }
 235                         Some(Err(err)) => {
 236                             // stop after first error
 237                             done = true;
 238                             // and pass the error through:
 239                             return Some((Err(err), percentage));
 240                         }
 241                         None => (), // open next directory
 242                     }
 243                 }
 244
 245                 inner = None;
 246
 247                 if at == 0x10000 {
 248                     done = true;
 249                     return None;
 250                 }
 251
 252                 let subdir: &str = &format!("{:04x}", at);
 253                 percentage = (at * 100) / 0x10000;
 254                 at += 1;
 255                 match tools::fs::read_subdir(base_handle.as_raw_fd(), subdir) {
 256                     Ok(dir) => {
 257                         inner = Some(dir);
 258                         // start reading:
 259                         continue;
 260                     }
 261                     Err(ref err) if err.as_errno() == Some(nix::errno::Errno::ENOENT) => {
 262                         // non-existing directories are okay, just keep going:
 263                         continue;
 264                     }
 265                     Err(err) => {
 266                         // other errors are fatal, so end our iteration
 267                         done = true;
 268                         // and pass the error through:
 269                         return Some((Err(format_err!("unable to read subdir '{}' - {}", subdir, err)), percentage));
 270                     }
 271                 }
 272             }
 273         }).fuse())
 274     }
 275
 276     pub fn oldest_writer(&self) -> Option<i64> {
 277         tools::ProcessLocker::oldest_shared_lock(self.locker.clone())
 278     }
 279
 280     pub fn sweep_unused_chunks(
 281         &self,
 282         oldest_writer: Option<i64>,
 283         status: &mut GarbageCollectionStatus,
 284         worker: Arc<WorkerTask>,
 285     ) -> Result<(), Error> {
 286         use nix::sys::stat::fstatat;
 287
 288         let now = unsafe { libc::time(std::ptr::null_mut()) };
 289
 290         let mut min_atime = now - 3600*24; // at least 24h (see mount option relatime)
 291
 292         if let Some(stamp) = oldest_writer {
 293             if stamp < min_atime {
 294                 min_atime = stamp;
 295             }
 296         }
 297
 298         min_atime -= 300; // add 5 mins gap for safety
 299
 300         let mut last_percentage = 0;
 301         let mut chunk_count = 0;
 302
 303         for (entry, percentage) in self.get_chunk_iterator()? {
 304             if last_percentage != percentage {
 305                 last_percentage = percentage;
 306                 worker.log(format!("percentage done: {}, chunk count: {}", percentage, chunk_count));
 307             }
 308
 309             tools::fail_on_shutdown()?;
 310
 311             let (dirfd, entry) = match entry {
 312                 Ok(entry) => (entry.parent_fd(), entry),
 313                 Err(err) => bail!("chunk iterator on chunk store '{}' failed - {}", self.name, err),
 314             };
 315
 316             let file_type = match entry.file_type() {
 317                 Some(file_type) => file_type,
 318                 None => bail!("unsupported file system type on chunk store '{}'", self.name),
 319             };
 320             if file_type != nix::dir::Type::File {
 321                 continue;
 322             }
 323
 324             chunk_count += 1;
 325
 326             let filename = entry.file_name();
 327
 328             let lock = self.mutex.lock();
 329
 330             if let Ok(stat) = fstatat(dirfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
 331                 let age = now - stat.st_atime;
 332                 //println!("FOUND {}  {:?}", age/(3600*24), filename);
 333                 if stat.st_atime < min_atime {
 334                     println!("UNLINK {}  {:?}", age/(3600*24), filename);
 335                     let res = unsafe { libc::unlinkat(dirfd, filename.as_ptr(), 0) };
 336                     if res != 0 {
 337                         let err = nix::Error::last();
 338                         bail!(
 339                             "unlink chunk {:?} failed on store '{}' - {}",
 340                             filename,
 341                             self.name,
 342                             err,
 343                         );
 344                     }
 345                     status.removed_chunks += 1;
 346                     status.removed_bytes += stat.st_size as u64;
 347                } else {
 348                     status.disk_chunks += 1;
 349                     status.disk_bytes += stat.st_size as u64;
 350                 }
 351             }
 352             drop(lock);
 353         }
 354
 355         Ok(())
 356     }
 357
 358     pub fn insert_chunk(
 359         &self,
 360         chunk: &DataBlob,
 361         digest: &[u8; 32],
 362     ) -> Result<(bool, u64), Error> {
 363
 364         //println!("DIGEST {}", proxmox::tools::digest_to_hex(digest));
 365
 366         let (chunk_path, digest_str) = self.chunk_path(digest);
 367
 368         let lock = self.mutex.lock();
 369
 370         if let Ok(metadata) = std::fs::metadata(&chunk_path) {
 371             if metadata.is_file() {
 372                 return Ok((true, metadata.len()));
 373             } else {
 374                 bail!("Got unexpected file type on store '{}' for chunk {}", self.name, digest_str);
 375             }
 376         }
 377
 378         let mut tmp_path = chunk_path.clone();
 379         tmp_path.set_extension("tmp");
 380
 381         let mut file = std::fs::File::create(&tmp_path)?;
 382
 383         let raw_data = chunk.raw_data();
 384         let encoded_size = raw_data.len() as u64;
 385
 386         file.write_all(raw_data)?;
 387
 388         if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) {
 389             if let Err(_) = std::fs::remove_file(&tmp_path)  { /* ignore */ }
 390             bail!(
 391                 "Atomic rename on store '{}' failed for chunk {} - {}",
 392                 self.name,
 393                 digest_str,
 394                 err,
 395             );
 396         }
 397
 398         drop(lock);
 399
 400         Ok((false, encoded_size))
 401     }
 402
 403     pub fn chunk_path(&self, digest:&[u8; 32]) -> (PathBuf, String) {
 404         let mut chunk_path = self.chunk_dir.clone();
 405         let prefix = digest_to_prefix(digest);
 406         chunk_path.push(&prefix);
 407         let digest_str = proxmox::tools::digest_to_hex(digest);
 408         chunk_path.push(&digest_str);
 409         (chunk_path, digest_str)
 410     }
 411
 412     pub fn relative_path(&self, path: &Path) -> PathBuf {
 413
 414         let mut full_path = self.base.clone();
 415         full_path.push(path);
 416         full_path
 417     }
 418
 419     pub fn base_path(&self) -> PathBuf {
 420         self.base.clone()
 421     }
 422
 423     pub fn try_shared_lock(&self) -> Result<tools::ProcessLockSharedGuard, Error> {
 424         tools::ProcessLocker::try_shared_lock(self.locker.clone())
 425     }
 426
 427     pub fn try_exclusive_lock(&self) -> Result<tools::ProcessLockExclusiveGuard, Error> {
 428         tools::ProcessLocker::try_exclusive_lock(self.locker.clone())
 429     }
 430 }
 431
 432
 433 #[test]
 434 fn test_chunk_store1() {
 435
 436     let mut path = std::fs::canonicalize(".").unwrap(); // we need absulute path
 437     path.push(".testdir");
 438
 439     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 440
 441     let chunk_store = ChunkStore::open("test", &path);
 442     assert!(chunk_store.is_err());
 443
 444     let chunk_store = ChunkStore::create("test", &path).unwrap();
 445
 446     let (chunk, digest) = super::DataChunkBuilder::new(&[0u8, 1u8]).build().unwrap();
 447
 448     let (exists, _) = chunk_store.insert_chunk(&chunk, &digest).unwrap();
 449     assert!(!exists);
 450
 451     let (exists, _) = chunk_store.insert_chunk(&chunk, &digest).unwrap();
 452     assert!(exists);
 453
 454
 455     let chunk_store = ChunkStore::create("test", &path);
 456     assert!(chunk_store.is_err());
 457
 458     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 459 }