]>
Commit | Line | Data |
---|---|---|
35cf5daa | 1 | use failure::*; |
43b13033 | 2 | |
35cf5daa | 3 | use std::path::{Path, PathBuf}; |
78216a5a | 4 | use std::io::{Read, Write}; |
43b13033 DM |
5 | use std::sync::{Arc, Mutex}; |
6 | use std::os::unix::io::AsRawFd; | |
128b37fe | 7 | |
7b2b40a8 | 8 | use openssl::sha; |
35cf5daa | 9 | |
365bb90f DM |
10 | use crate::tools; |
11 | ||
64e53b28 DM |
12 | pub struct GarbageCollectionStatus { |
13 | pub used_bytes: usize, | |
14 | pub used_chunks: usize, | |
15 | pub disk_bytes: usize, | |
16 | pub disk_chunks: usize, | |
17 | } | |
18 | ||
19 | impl Default for GarbageCollectionStatus { | |
20 | fn default() -> Self { | |
21 | GarbageCollectionStatus { | |
22 | used_bytes: 0, | |
23 | used_chunks: 0, | |
24 | disk_bytes: 0, | |
25 | disk_chunks: 0, | |
26 | } | |
27 | } | |
28 | } | |
29 | ||
e5064ba6 | 30 | /// File system based chunk store |
35cf5daa | 31 | pub struct ChunkStore { |
277fc5a3 | 32 | name: String, // used for error reporting |
2c32fdde | 33 | pub (crate) base: PathBuf, |
35cf5daa | 34 | chunk_dir: PathBuf, |
c5d82e5f | 35 | mutex: Mutex<bool>, |
43b13033 | 36 | locker: Arc<Mutex<tools::ProcessLocker>>, |
128b37fe DM |
37 | } |
38 | ||
176e4af9 DM |
39 | // TODO: what about sysctl setting vm.vfs_cache_pressure (0 - 100) ? |
40 | ||
247cdbce DM |
41 | pub fn verify_chunk_size(size: u64) -> Result<(), Error> { |
42 | ||
43 | static SIZES: [u64; 7] = [64*1024, 128*1024, 256*1024, 512*1024, 1024*1024, 2048*1024, 4096*1024]; | |
44 | ||
45 | if !SIZES.contains(&size) { | |
46 | bail!("Got unsupported chunk size '{}'", size); | |
47 | } | |
48 | Ok(()) | |
49 | } | |
50 | ||
08481a0b | 51 | fn digest_to_prefix(digest: &[u8]) -> PathBuf { |
128b37fe | 52 | |
e95950e4 | 53 | let mut buf = Vec::<u8>::with_capacity(2+1+2+1); |
128b37fe | 54 | |
22968600 DM |
55 | const HEX_CHARS: &'static [u8; 16] = b"0123456789abcdef"; |
56 | ||
128b37fe DM |
57 | buf.push(HEX_CHARS[(digest[0] as usize) >> 4]); |
58 | buf.push(HEX_CHARS[(digest[0] as usize) &0xf]); | |
e95950e4 | 59 | buf.push(HEX_CHARS[(digest[1] as usize) >> 4]); |
128b37fe | 60 | buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]); |
128b37fe DM |
61 | buf.push('/' as u8); |
62 | ||
63 | let path = unsafe { String::from_utf8_unchecked(buf)}; | |
64 | ||
65 | path.into() | |
35cf5daa DM |
66 | } |
67 | ||
68 | impl ChunkStore { | |
69 | ||
45773720 DM |
70 | fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf { |
71 | ||
72 | let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref()); | |
73 | chunk_dir.push(".chunks"); | |
74 | ||
75 | chunk_dir | |
76 | } | |
77 | ||
277fc5a3 | 78 | pub fn create<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> { |
35cf5daa | 79 | |
45773720 | 80 | let base: PathBuf = path.into(); |
68469eeb DM |
81 | |
82 | if !base.is_absolute() { | |
83 | bail!("expected absolute path - got {:?}", base); | |
84 | } | |
85 | ||
45773720 | 86 | let chunk_dir = Self::chunk_dir(&base); |
35cf5daa | 87 | |
2989f6bf | 88 | if let Err(err) = std::fs::create_dir(&base) { |
277fc5a3 | 89 | bail!("unable to create chunk store '{}' at {:?} - {}", name, base, err); |
2989f6bf DM |
90 | } |
91 | ||
92 | if let Err(err) = std::fs::create_dir(&chunk_dir) { | |
277fc5a3 | 93 | bail!("unable to create chunk store '{}' subdir {:?} - {}", name, chunk_dir, err); |
2989f6bf | 94 | } |
35cf5daa | 95 | |
bc616633 | 96 | // create 64*1024 subdirs |
e95950e4 DM |
97 | let mut last_percentage = 0; |
98 | ||
bc616633 | 99 | for i in 0..64*1024 { |
af3e7d75 | 100 | let mut l1path = chunk_dir.clone(); |
bc616633 | 101 | l1path.push(format!("{:04x}", i)); |
2989f6bf | 102 | if let Err(err) = std::fs::create_dir(&l1path) { |
277fc5a3 | 103 | bail!("unable to create chunk store '{}' subdir {:?} - {}", name, l1path, err); |
2989f6bf | 104 | } |
bc616633 DM |
105 | let percentage = (i*100)/(64*1024); |
106 | if percentage != last_percentage { | |
107 | eprintln!("Percentage done: {}", percentage); | |
108 | last_percentage = percentage; | |
e95950e4 | 109 | } |
35cf5daa DM |
110 | } |
111 | ||
277fc5a3 | 112 | Self::open(name, base) |
35cf5daa DM |
113 | } |
114 | ||
277fc5a3 | 115 | pub fn open<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> { |
35cf5daa | 116 | |
45773720 | 117 | let base: PathBuf = path.into(); |
68469eeb DM |
118 | |
119 | if !base.is_absolute() { | |
120 | bail!("expected absolute path - got {:?}", base); | |
121 | } | |
122 | ||
45773720 DM |
123 | let chunk_dir = Self::chunk_dir(&base); |
124 | ||
ce55dbbc | 125 | if let Err(err) = std::fs::metadata(&chunk_dir) { |
277fc5a3 | 126 | bail!("unable to open chunk store '{}' at {:?} - {}", name, chunk_dir, err); |
ce55dbbc | 127 | } |
45773720 DM |
128 | |
129 | let mut lockfile_path = base.clone(); | |
130 | lockfile_path.push(".lock"); | |
131 | ||
43b13033 | 132 | let locker = tools::ProcessLocker::new(&lockfile_path)?; |
35cf5daa | 133 | |
b8d4766a | 134 | Ok(ChunkStore { |
277fc5a3 | 135 | name: name.to_owned(), |
b8d4766a DM |
136 | base, |
137 | chunk_dir, | |
43b13033 | 138 | locker, |
b8d4766a DM |
139 | mutex: Mutex::new(false) |
140 | }) | |
35cf5daa DM |
141 | } |
142 | ||
7394ca3e | 143 | pub fn touch_chunk(&self, digest:&[u8]) -> Result<(), Error> { |
3d5c11e5 | 144 | |
a198d74f | 145 | let mut chunk_path = self.chunk_dir.clone(); |
08481a0b | 146 | let prefix = digest_to_prefix(&digest); |
3d5c11e5 | 147 | chunk_path.push(&prefix); |
22968600 | 148 | let digest_str = tools::digest_to_hex(&digest); |
3d5c11e5 DM |
149 | chunk_path.push(&digest_str); |
150 | ||
7ee2aa1b DM |
151 | const UTIME_NOW: i64 = ((1 << 30) - 1); |
152 | const UTIME_OMIT: i64 = ((1 << 30) - 2); | |
153 | ||
a198d74f | 154 | let times: [libc::timespec; 2] = [ |
7ee2aa1b DM |
155 | libc::timespec { tv_sec: 0, tv_nsec: UTIME_NOW }, |
156 | libc::timespec { tv_sec: 0, tv_nsec: UTIME_OMIT } | |
157 | ]; | |
158 | ||
159 | use nix::NixPath; | |
160 | ||
161 | let res = chunk_path.with_nix_path(|cstr| unsafe { | |
162 | libc::utimensat(-1, cstr.as_ptr(), ×[0], libc::AT_SYMLINK_NOFOLLOW) | |
163 | })?; | |
164 | ||
165 | if let Err(err) = nix::errno::Errno::result(res) { | |
166 | bail!("updata atime failed for chunk {:?} - {}", chunk_path, err); | |
167 | } | |
168 | ||
3d5c11e5 DM |
169 | Ok(()) |
170 | } | |
171 | ||
df9973e8 | 172 | pub fn read_chunk(&self, digest:&[u8], buffer: &mut Vec<u8>) -> Result<(), Error> { |
96df2fb4 DM |
173 | |
174 | let mut chunk_path = self.chunk_dir.clone(); | |
175 | let prefix = digest_to_prefix(&digest); | |
176 | chunk_path.push(&prefix); | |
22968600 | 177 | let digest_str = tools::digest_to_hex(&digest); |
96df2fb4 DM |
178 | chunk_path.push(&digest_str); |
179 | ||
d2690f74 | 180 | buffer.clear(); |
78216a5a | 181 | let f = std::fs::File::open(&chunk_path)?; |
141f062e | 182 | let mut decoder = zstd::stream::Decoder::new(f)?; |
df9973e8 | 183 | |
78216a5a | 184 | decoder.read_to_end(buffer)?; |
96df2fb4 | 185 | |
df9973e8 | 186 | Ok(()) |
96df2fb4 DM |
187 | } |
188 | ||
9739aca4 WB |
189 | pub fn get_chunk_iterator( |
190 | &self, | |
eff25eca | 191 | print_percentage: bool, |
9739aca4 | 192 | ) -> Result< |
3a50ddd0 | 193 | impl Iterator<Item = Result<tools::fs::ReadDirEntry, Error>> + std::iter::FusedIterator, |
9739aca4 WB |
194 | Error |
195 | > { | |
196 | use nix::dir::Dir; | |
197 | use nix::fcntl::OFlag; | |
198 | use nix::sys::stat::Mode; | |
199 | ||
200 | let base_handle = match Dir::open( | |
201 | &self.chunk_dir, OFlag::O_RDONLY, Mode::empty()) { | |
202 | Ok(h) => h, | |
203 | Err(err) => bail!("unable to open store '{}' chunk dir {:?} - {}", | |
204 | self.name, self.chunk_dir, err), | |
205 | }; | |
206 | ||
c7f481b6 WB |
207 | let mut verbose = true; |
208 | let mut last_percentage = 0; | |
209 | ||
9739aca4 | 210 | Ok((0..0x10000).filter_map(move |index| { |
eff25eca WB |
211 | if print_percentage { |
212 | let percentage = (index * 100) / 0x10000; | |
213 | if last_percentage != percentage { | |
214 | last_percentage = percentage; | |
215 | eprintln!("percentage done: {}", percentage); | |
216 | } | |
c7f481b6 WB |
217 | } |
218 | let subdir: &str = &format!("{:04x}", index); | |
219 | match tools::fs::read_subdir(base_handle.as_raw_fd(), subdir) { | |
220 | Err(e) => { | |
221 | if verbose { | |
222 | eprintln!("Error iterating through chunks: {}", e); | |
223 | verbose = false; | |
224 | } | |
225 | None | |
226 | } | |
227 | Ok(iter) => Some(iter), | |
228 | } | |
229 | }) | |
62f2422f WB |
230 | .flatten() |
231 | .filter(|entry| { | |
232 | // Check that the file name is actually a hash! (64 hex digits) | |
233 | let entry = match entry { | |
234 | Err(_) => return true, // pass errors onwards | |
235 | Ok(ref entry) => entry, | |
236 | }; | |
237 | let bytes = entry.file_name().to_bytes(); | |
238 | if bytes.len() != 64 { | |
239 | return false; | |
240 | } | |
241 | for b in bytes { | |
242 | if !b.is_ascii_hexdigit() { | |
243 | return false; | |
244 | } | |
245 | } | |
246 | true | |
247 | })) | |
c7f481b6 WB |
248 | } |
249 | ||
11861a48 DM |
250 | pub fn oldest_writer(&self) -> Option<i64> { |
251 | tools::ProcessLocker::oldest_shared_lock(self.locker.clone()) | |
252 | } | |
253 | ||
254 | pub fn sweep_unused_chunks( | |
255 | &self, | |
256 | oldest_writer: Option<i64>, | |
257 | status: &mut GarbageCollectionStatus | |
258 | ) -> Result<(), Error> { | |
9349d2a1 | 259 | use nix::sys::stat::fstatat; |
08481a0b | 260 | |
fdd71f52 | 261 | let now = unsafe { libc::time(std::ptr::null_mut()) }; |
c7f481b6 | 262 | |
11861a48 DM |
263 | let mut min_atime = now - 3600*24; // at least 24h (see mount option relatime) |
264 | ||
265 | if let Some(stamp) = oldest_writer { | |
266 | if stamp < min_atime { | |
267 | min_atime = stamp; | |
268 | } | |
269 | } | |
270 | ||
271 | min_atime -= 300; // add 5 mins gap for safety | |
272 | ||
eff25eca | 273 | for entry in self.get_chunk_iterator(true)? { |
fdd71f52 | 274 | let (dirfd, entry) = match entry { |
c7f481b6 WB |
275 | Ok(entry) => (entry.parent_fd(), entry), |
276 | Err(_) => continue, // ignore errors | |
eae8aa3a | 277 | }; |
fdd71f52 | 278 | |
eae8aa3a DM |
279 | let file_type = match entry.file_type() { |
280 | Some(file_type) => file_type, | |
277fc5a3 | 281 | None => bail!("unsupported file system type on chunk store '{}'", self.name), |
eae8aa3a | 282 | }; |
82bc0ad4 WB |
283 | if file_type != nix::dir::Type::File { |
284 | continue; | |
285 | } | |
eae8aa3a | 286 | |
d85987ae DM |
287 | let filename = entry.file_name(); |
288 | ||
15a77c4c DM |
289 | let lock = self.mutex.lock(); |
290 | ||
9349d2a1 | 291 | if let Ok(stat) = fstatat(dirfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) { |
eae8aa3a | 292 | let age = now - stat.st_atime; |
e95950e4 | 293 | //println!("FOUND {} {:?}", age/(3600*24), filename); |
11861a48 | 294 | if stat.st_atime < min_atime { |
eae8aa3a | 295 | println!("UNLINK {} {:?}", age/(3600*24), filename); |
fdd71f52 | 296 | let res = unsafe { libc::unlinkat(dirfd, filename.as_ptr(), 0) }; |
277fc5a3 DM |
297 | if res != 0 { |
298 | let err = nix::Error::last(); | |
9349d2a1 WB |
299 | bail!( |
300 | "unlink chunk {:?} failed on store '{}' - {}", | |
301 | filename, | |
302 | self.name, | |
303 | err, | |
304 | ); | |
277fc5a3 | 305 | } |
64e53b28 DM |
306 | } else { |
307 | status.disk_chunks += 1; | |
308 | status.disk_bytes += stat.st_size as usize; | |
08481a0b | 309 | } |
eae8aa3a | 310 | } |
15a77c4c | 311 | drop(lock); |
eae8aa3a | 312 | } |
277fc5a3 | 313 | Ok(()) |
08481a0b DM |
314 | } |
315 | ||
798f7fa0 | 316 | pub fn insert_chunk(&self, chunk: &[u8]) -> Result<(bool, [u8; 32], u64), Error> { |
128b37fe | 317 | |
7b2b40a8 | 318 | // fixme: use Sha512/256 when available |
9ac6ec86 WB |
319 | let digest = sha::sha256(chunk); |
320 | let (new, csize) = self.insert_chunk_noverify(&digest, chunk)?; | |
321 | Ok((new, digest, csize)) | |
322 | } | |
7b2b40a8 | 323 | |
9ac6ec86 WB |
324 | pub fn insert_chunk_noverify( |
325 | &self, | |
326 | digest: &[u8; 32], | |
327 | chunk: &[u8], | |
328 | ) -> Result<(bool, u64), Error> { | |
7b2b40a8 | 329 | |
22968600 | 330 | //println!("DIGEST {}", tools::digest_to_hex(&digest)); |
128b37fe | 331 | |
af3e7d75 | 332 | let mut chunk_path = self.chunk_dir.clone(); |
9ac6ec86 | 333 | let prefix = digest_to_prefix(digest); |
c5d82e5f | 334 | chunk_path.push(&prefix); |
9ac6ec86 | 335 | let digest_str = tools::digest_to_hex(digest); |
c5d82e5f DM |
336 | chunk_path.push(&digest_str); |
337 | ||
338 | let lock = self.mutex.lock(); | |
339 | ||
340 | if let Ok(metadata) = std::fs::metadata(&chunk_path) { | |
341 | if metadata.is_file() { | |
9ac6ec86 | 342 | return Ok((true, metadata.len())); |
c5d82e5f | 343 | } else { |
f7dd683b | 344 | bail!("Got unexpected file type on store '{}' for chunk {}", self.name, digest_str); |
c5d82e5f DM |
345 | } |
346 | } | |
128b37fe | 347 | |
c5d82e5f DM |
348 | let mut tmp_path = chunk_path.clone(); |
349 | tmp_path.set_extension("tmp"); | |
78216a5a DM |
350 | |
351 | let f = std::fs::File::create(&tmp_path)?; | |
352 | ||
141f062e | 353 | let mut encoder = zstd::stream::Encoder::new(f, 1)?; |
78216a5a DM |
354 | |
355 | encoder.write_all(chunk)?; | |
141f062e | 356 | let f = encoder.finish()?; |
128b37fe | 357 | |
c5d82e5f DM |
358 | if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) { |
359 | if let Err(_) = std::fs::remove_file(&tmp_path) { /* ignore */ } | |
9349d2a1 WB |
360 | bail!( |
361 | "Atomic rename on store '{}' failed for chunk {} - {}", | |
362 | self.name, | |
363 | digest_str, | |
364 | err, | |
365 | ); | |
c5d82e5f DM |
366 | } |
367 | ||
798f7fa0 DM |
368 | // fixme: is there a better way to get the compressed size? |
369 | let stat = nix::sys::stat::fstat(f.as_raw_fd())?; | |
370 | let compressed_size = stat.st_size as u64; | |
371 | ||
77703d95 | 372 | //println!("PATH {:?}", chunk_path); |
128b37fe | 373 | |
c5d82e5f DM |
374 | drop(lock); |
375 | ||
9ac6ec86 | 376 | Ok((false, compressed_size)) |
128b37fe DM |
377 | } |
378 | ||
606ce64b DM |
379 | pub fn relative_path(&self, path: &Path) -> PathBuf { |
380 | ||
381 | let mut full_path = self.base.clone(); | |
382 | full_path.push(path); | |
383 | full_path | |
384 | } | |
385 | ||
3d5c11e5 DM |
386 | pub fn base_path(&self) -> PathBuf { |
387 | self.base.clone() | |
388 | } | |
43b13033 DM |
389 | |
390 | pub fn try_shared_lock(&self) -> Result<tools::ProcessLockSharedGuard, Error> { | |
391 | tools::ProcessLocker::try_shared_lock(self.locker.clone()) | |
392 | } | |
393 | ||
394 | pub fn try_exclusive_lock(&self) -> Result<tools::ProcessLockExclusiveGuard, Error> { | |
395 | tools::ProcessLocker::try_exclusive_lock(self.locker.clone()) | |
396 | } | |
35cf5daa DM |
397 | } |
398 | ||
399 | ||
400 | #[test] | |
401 | fn test_chunk_store1() { | |
402 | ||
332dcc22 DM |
403 | let mut path = std::fs::canonicalize(".").unwrap(); // we need absulute path |
404 | path.push(".testdir"); | |
405 | ||
35cf5daa DM |
406 | if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ } |
407 | ||
332dcc22 | 408 | let chunk_store = ChunkStore::open("test", &path); |
35cf5daa DM |
409 | assert!(chunk_store.is_err()); |
410 | ||
332dcc22 | 411 | let chunk_store = ChunkStore::create("test", &path).unwrap(); |
798f7fa0 | 412 | let (exists, _, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap(); |
391a2e43 DM |
413 | assert!(!exists); |
414 | ||
798f7fa0 | 415 | let (exists, _, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap(); |
391a2e43 | 416 | assert!(exists); |
128b37fe | 417 | |
35cf5daa | 418 | |
332dcc22 | 419 | let chunk_store = ChunkStore::create("test", &path); |
35cf5daa DM |
420 | assert!(chunk_store.is_err()); |
421 | ||
e0a5d1ca | 422 | if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ } |
35cf5daa | 423 | } |