src/backup/datastore.rs

   1 use std::collections::{HashSet, HashMap};
   2 use std::io::{self, Write};
   3 use std::path::{Path, PathBuf};
   4 use std::sync::{Arc, Mutex};
   5
   6 use anyhow::{bail, format_err, Error};
   7 use lazy_static::lazy_static;
   8 use chrono::{DateTime, Utc};
   9
  10 use super::backup_info::{BackupGroup, BackupDir};
  11 use super::chunk_store::ChunkStore;
  12 use super::dynamic_index::{DynamicIndexReader, DynamicIndexWriter};
  13 use super::fixed_index::{FixedIndexReader, FixedIndexWriter};
  14 use super::manifest::{MANIFEST_BLOB_NAME, BackupManifest};
  15 use super::index::*;
  16 use super::{DataBlob, ArchiveType, archive_type};
  17 use crate::config::datastore;
  18 use crate::server::WorkerTask;
  19 use crate::tools;
  20 use crate::api2::types::GarbageCollectionStatus;
  21
  22 lazy_static! {
  23     static ref DATASTORE_MAP: Mutex<HashMap<String, Arc<DataStore>>> = Mutex::new(HashMap::new());
  24 }
  25
  26 /// Datastore Management
  27 ///
  28 /// A Datastore can store severals backups, and provides the
  29 /// management interface for backup.
  30 pub struct DataStore {
  31     chunk_store: Arc<ChunkStore>,
  32     gc_mutex: Mutex<bool>,
  33     last_gc_status: Mutex<GarbageCollectionStatus>,
  34 }
  35
  36 impl DataStore {
  37
  38     pub fn lookup_datastore(name: &str) -> Result<Arc<DataStore>, Error> {
  39
  40         let (config, _digest) = datastore::config()?;
  41         let config: datastore::DataStoreConfig = config.lookup("datastore", name)?;
  42
  43         let mut map = DATASTORE_MAP.lock().unwrap();
  44
  45         if let Some(datastore) = map.get(name) {
  46             // Compare Config - if changed, create new Datastore object!
  47             if datastore.chunk_store.base == PathBuf::from(&config.path) {
  48                 return Ok(datastore.clone());
  49             }
  50         }
  51
  52         let datastore = DataStore::open(name)?;
  53
  54         let datastore = Arc::new(datastore);
  55         map.insert(name.to_string(), datastore.clone());
  56
  57         Ok(datastore)
  58     }
  59
  60     pub fn open(store_name: &str) -> Result<Self, Error> {
  61
  62         let (config, _digest) = datastore::config()?;
  63         let (_, store_config) = config.sections.get(store_name)
  64             .ok_or(format_err!("no such datastore '{}'", store_name))?;
  65
  66         let path = store_config["path"].as_str().unwrap();
  67
  68         let chunk_store = ChunkStore::open(store_name, path)?;
  69
  70         let gc_status = GarbageCollectionStatus::default();
  71
  72         Ok(Self {
  73             chunk_store: Arc::new(chunk_store),
  74             gc_mutex: Mutex::new(false),
  75             last_gc_status: Mutex::new(gc_status),
  76         })
  77     }
  78
  79     pub fn get_chunk_iterator(
  80         &self,
  81     ) -> Result<
  82         impl Iterator<Item = (Result<tools::fs::ReadDirEntry, Error>, usize)>,
  83         Error
  84     > {
  85         self.chunk_store.get_chunk_iterator()
  86     }
  87
  88     pub fn create_fixed_writer<P: AsRef<Path>>(&self, filename: P, size: usize, chunk_size: usize) -> Result<FixedIndexWriter, Error> {
  89
  90         let index = FixedIndexWriter::create(self.chunk_store.clone(), filename.as_ref(), size, chunk_size)?;
  91
  92         Ok(index)
  93     }
  94
  95     pub fn open_fixed_reader<P: AsRef<Path>>(&self, filename: P) -> Result<FixedIndexReader, Error> {
  96
  97         let full_path =  self.chunk_store.relative_path(filename.as_ref());
  98
  99         let index = FixedIndexReader::open(&full_path)?;
 100
 101         Ok(index)
 102     }
 103
 104     pub fn create_dynamic_writer<P: AsRef<Path>>(
 105         &self, filename: P,
 106     ) -> Result<DynamicIndexWriter, Error> {
 107
 108         let index = DynamicIndexWriter::create(
 109             self.chunk_store.clone(), filename.as_ref())?;
 110
 111         Ok(index)
 112     }
 113
 114     pub fn open_dynamic_reader<P: AsRef<Path>>(&self, filename: P) -> Result<DynamicIndexReader, Error> {
 115
 116         let full_path =  self.chunk_store.relative_path(filename.as_ref());
 117
 118         let index = DynamicIndexReader::open(&full_path)?;
 119
 120         Ok(index)
 121     }
 122
 123     pub fn open_index<P>(&self, filename: P) -> Result<Box<dyn IndexFile + Send>, Error>
 124     where
 125         P: AsRef<Path>,
 126     {
 127         let filename = filename.as_ref();
 128         let out: Box<dyn IndexFile + Send> =
 129             match archive_type(filename)? {
 130                 ArchiveType::DynamicIndex => Box::new(self.open_dynamic_reader(filename)?),
 131                 ArchiveType::FixedIndex => Box::new(self.open_fixed_reader(filename)?),
 132                 _ => bail!("cannot open index file of unknown type: {:?}", filename),
 133             };
 134         Ok(out)
 135     }
 136
 137     pub fn base_path(&self) -> PathBuf {
 138         self.chunk_store.base_path()
 139     }
 140
 141     /// Clenaup a backup directory
 142     ///
 143     /// Removes all files not mentioned in the manifest.
 144     pub fn cleanup_backup_dir(&self, backup_dir: &BackupDir, manifest: &BackupManifest
 145     ) ->  Result<(), Error> {
 146
 147         let mut full_path = self.base_path();
 148         full_path.push(backup_dir.relative_path());
 149
 150         let mut wanted_files = HashSet::new();
 151         wanted_files.insert(MANIFEST_BLOB_NAME.to_string());
 152         manifest.files().iter().for_each(|item| { wanted_files.insert(item.filename.clone()); });
 153
 154         for item in tools::fs::read_subdir(libc::AT_FDCWD, &full_path)? {
 155             if let Ok(item) = item {
 156                 if let Some(file_type) = item.file_type() {
 157                     if file_type != nix::dir::Type::File { continue; }
 158                 }
 159                 let file_name = item.file_name().to_bytes();
 160                 if file_name == b"." || file_name == b".." { continue; };
 161
 162                 if let Ok(name) = std::str::from_utf8(file_name) {
 163                     if wanted_files.contains(name) { continue; }
 164                 }
 165                 println!("remove unused file {:?}", item.file_name());
 166                 let dirfd = item.parent_fd();
 167                 let _res = unsafe { libc::unlinkat(dirfd, item.file_name().as_ptr(), 0) };
 168             }
 169         }
 170
 171         Ok(())
 172     }
 173
 174     /// Returns the absolute path for a backup_group
 175     pub fn group_path(&self, backup_group: &BackupGroup) -> PathBuf {
 176         let mut full_path = self.base_path();
 177         full_path.push(backup_group.group_path());
 178         full_path
 179     }
 180
 181     /// Returns the absolute path for backup_dir
 182     pub fn snapshot_path(&self, backup_dir: &BackupDir) -> PathBuf {
 183         let mut full_path = self.base_path();
 184         full_path.push(backup_dir.relative_path());
 185         full_path
 186     }
 187
 188     /// Remove a complete backup group including all snapshots
 189     pub fn remove_backup_group(&self, backup_group: &BackupGroup) ->  Result<(), Error> {
 190
 191         let full_path = self.group_path(backup_group);
 192
 193         log::info!("removing backup group {:?}", full_path);
 194         std::fs::remove_dir_all(&full_path)
 195             .map_err(|err| {
 196                 format_err!(
 197                     "removing backup group {:?} failed - {}",
 198                     full_path,
 199                     err,
 200                 )
 201             })?;
 202
 203         Ok(())
 204     }
 205
 206     /// Remove a backup directory including all content
 207     pub fn remove_backup_dir(&self, backup_dir: &BackupDir) ->  Result<(), Error> {
 208
 209         let full_path = self.snapshot_path(backup_dir);
 210
 211         log::info!("removing backup snapshot {:?}", full_path);
 212         std::fs::remove_dir_all(&full_path)
 213             .map_err(|err| {
 214                 format_err!(
 215                     "removing backup snapshot {:?} failed - {}",
 216                     full_path,
 217                     err,
 218                 )
 219             })?;
 220
 221         Ok(())
 222     }
 223
 224     /// Returns the time of the last successful backup
 225     ///
 226     /// Or None if there is no backup in the group (or the group dir does not exist).
 227     pub fn last_successful_backup(&self, backup_group: &BackupGroup) -> Result<Option<DateTime<Utc>>, Error> {
 228         let base_path = self.base_path();
 229         let mut group_path = base_path.clone();
 230         group_path.push(backup_group.group_path());
 231
 232         if group_path.exists() {
 233             backup_group.last_successful_backup(&base_path)
 234         } else {
 235             Ok(None)
 236         }
 237     }
 238
 239     /// Returns the backup owner.
 240     ///
 241     /// The backup owner is the user who first created the backup group.
 242     pub fn get_owner(&self, backup_group: &BackupGroup) -> Result<String, Error> {
 243         let mut full_path = self.base_path();
 244         full_path.push(backup_group.group_path());
 245         full_path.push("owner");
 246         let owner = proxmox::tools::fs::file_read_firstline(full_path)?;
 247         Ok(owner.trim_end().to_string()) // remove trailing newline
 248     }
 249
 250     /// Set the backup owner.
 251     pub fn set_owner(&self, backup_group: &BackupGroup, userid: &str, force: bool) -> Result<(), Error> {
 252         let mut path = self.base_path();
 253         path.push(backup_group.group_path());
 254         path.push("owner");
 255
 256         let mut open_options = std::fs::OpenOptions::new();
 257         open_options.write(true);
 258         open_options.truncate(true);
 259
 260         if force {
 261             open_options.create(true);
 262         } else {
 263             open_options.create_new(true);
 264         }
 265
 266         let mut file = open_options.open(&path)
 267             .map_err(|err| format_err!("unable to create owner file {:?} - {}", path, err))?;
 268
 269         write!(file, "{}\n", userid)
 270             .map_err(|err| format_err!("unable to write owner file  {:?} - {}", path, err))?;
 271
 272         Ok(())
 273     }
 274
 275     /// Create a backup group if it does not already exists.
 276     ///
 277     /// And set the owner to 'userid'. If the group already exists, it returns the
 278     /// current owner (instead of setting the owner).
 279     pub fn create_backup_group(&self, backup_group: &BackupGroup, userid: &str) -> Result<String, Error> {
 280
 281         // create intermediate path first:
 282         let base_path = self.base_path();
 283
 284         let mut full_path = base_path.clone();
 285         full_path.push(backup_group.backup_type());
 286         std::fs::create_dir_all(&full_path)?;
 287
 288         full_path.push(backup_group.backup_id());
 289
 290         // create the last component now
 291         match std::fs::create_dir(&full_path) {
 292             Ok(_) => {
 293                 self.set_owner(backup_group, userid, false)?;
 294                 let owner = self.get_owner(backup_group)?; // just to be sure
 295                 Ok(owner)
 296             }
 297             Err(ref err) if err.kind() == io::ErrorKind::AlreadyExists => {
 298                 let owner = self.get_owner(backup_group)?; // just to be sure
 299                 Ok(owner)
 300             }
 301             Err(err) => bail!("unable to create backup group {:?} - {}", full_path, err),
 302         }
 303     }
 304
 305     /// Creates a new backup snapshot inside a BackupGroup
 306     ///
 307     /// The BackupGroup directory needs to exist.
 308     pub fn create_backup_dir(&self, backup_dir: &BackupDir) ->  Result<(PathBuf, bool), io::Error> {
 309         let relative_path = backup_dir.relative_path();
 310         let mut full_path = self.base_path();
 311         full_path.push(&relative_path);
 312
 313         match std::fs::create_dir(&full_path) {
 314             Ok(_) => Ok((relative_path, true)),
 315             Err(ref e) if e.kind() == io::ErrorKind::AlreadyExists => Ok((relative_path, false)),
 316             Err(e) => Err(e)
 317         }
 318     }
 319
 320     pub fn list_images(&self) -> Result<Vec<PathBuf>, Error> {
 321         let base = self.base_path();
 322
 323         let mut list = vec![];
 324
 325         use walkdir::WalkDir;
 326
 327         let walker = WalkDir::new(&base).same_file_system(true).into_iter();
 328
 329         // make sure we skip .chunks (and other hidden files to keep it simple)
 330         fn is_hidden(entry: &walkdir::DirEntry) -> bool {
 331             entry.file_name()
 332                 .to_str()
 333                 .map(|s| s.starts_with("."))
 334                 .unwrap_or(false)
 335         }
 336
 337         for entry in walker.filter_entry(|e| !is_hidden(e)) {
 338             let path = entry?.into_path();
 339             if let Ok(archive_type) = archive_type(&path) {
 340                 if archive_type == ArchiveType::FixedIndex || archive_type == ArchiveType::DynamicIndex {
 341                     list.push(path);
 342                 }
 343             }
 344         }
 345
 346         Ok(list)
 347     }
 348
 349     // mark chunks  used by ``index`` as used
 350     fn index_mark_used_chunks<I: IndexFile>(
 351         &self,
 352         index: I,
 353         file_name: &Path, // only used for error reporting
 354         status: &mut GarbageCollectionStatus,
 355     ) -> Result<(), Error> {
 356
 357         status.index_file_count += 1;
 358         status.index_data_bytes += index.index_bytes();
 359
 360         for pos in 0..index.index_count() {
 361             tools::fail_on_shutdown()?;
 362             let digest = index.index_digest(pos).unwrap();
 363             if let Err(err) = self.chunk_store.touch_chunk(digest) {
 364                 bail!("unable to access chunk {}, required by {:?} - {}",
 365                       proxmox::tools::digest_to_hex(digest), file_name, err);
 366             }
 367         }
 368         Ok(())
 369     }
 370
 371     fn mark_used_chunks(&self, status: &mut GarbageCollectionStatus) -> Result<(), Error> {
 372
 373         let image_list = self.list_images()?;
 374
 375         for path in image_list {
 376
 377             tools::fail_on_shutdown()?;
 378
 379             if let Ok(archive_type) = archive_type(&path) {
 380                 if archive_type == ArchiveType::FixedIndex {
 381                     let index = self.open_fixed_reader(&path)?;
 382                     self.index_mark_used_chunks(index, &path, status)?;
 383                 } else if archive_type == ArchiveType::DynamicIndex {
 384                     let index = self.open_dynamic_reader(&path)?;
 385                     self.index_mark_used_chunks(index, &path, status)?;
 386                 }
 387             }
 388         }
 389
 390         Ok(())
 391     }
 392
 393     pub fn last_gc_status(&self) -> GarbageCollectionStatus {
 394         self.last_gc_status.lock().unwrap().clone()
 395     }
 396
 397     pub fn garbage_collection(&self, worker: Arc<WorkerTask>) -> Result<(), Error> {
 398
 399         if let Ok(ref mut _mutex) = self.gc_mutex.try_lock() {
 400
 401             let _exclusive_lock =  self.chunk_store.try_exclusive_lock()?;
 402
 403             let now = unsafe { libc::time(std::ptr::null_mut()) };
 404
 405             let oldest_writer = self.chunk_store.oldest_writer().unwrap_or(now);
 406
 407             let mut gc_status = GarbageCollectionStatus::default();
 408             gc_status.upid = Some(worker.to_string());
 409
 410             worker.log("Start GC phase1 (mark used chunks)");
 411
 412             self.mark_used_chunks(&mut gc_status)?;
 413
 414             worker.log("Start GC phase2 (sweep unused chunks)");
 415             self.chunk_store.sweep_unused_chunks(oldest_writer, &mut gc_status, worker.clone())?;
 416
 417             worker.log(&format!("Removed bytes: {}", gc_status.removed_bytes));
 418             worker.log(&format!("Removed chunks: {}", gc_status.removed_chunks));
 419             if gc_status.pending_bytes > 0 {
 420                 worker.log(&format!("Pending removals: {} bytes ({} chunks)", gc_status.pending_bytes, gc_status.pending_chunks));
 421             }
 422
 423             worker.log(&format!("Original data bytes: {}", gc_status.index_data_bytes));
 424
 425             if gc_status.index_data_bytes > 0 {
 426                 let comp_per = (gc_status.disk_bytes*100)/gc_status.index_data_bytes;
 427                 worker.log(&format!("Disk bytes: {} ({} %)", gc_status.disk_bytes, comp_per));
 428             }
 429
 430             worker.log(&format!("Disk chunks: {}", gc_status.disk_chunks));
 431
 432             if gc_status.disk_chunks > 0 {
 433                 let avg_chunk = gc_status.disk_bytes/(gc_status.disk_chunks as u64);
 434                 worker.log(&format!("Average chunk size: {}", avg_chunk));
 435             }
 436
 437             *self.last_gc_status.lock().unwrap() = gc_status;
 438
 439         } else {
 440             bail!("Start GC failed - (already running/locked)");
 441         }
 442
 443         Ok(())
 444     }
 445
 446     pub fn try_shared_chunk_store_lock(&self) -> Result<tools::ProcessLockSharedGuard, Error> {
 447         self.chunk_store.try_shared_lock()
 448     }
 449
 450     pub fn chunk_path(&self, digest:&[u8; 32]) -> (PathBuf, String) {
 451         self.chunk_store.chunk_path(digest)
 452     }
 453
 454     pub fn cond_touch_chunk(&self, digest: &[u8; 32], fail_if_not_exist: bool) -> Result<bool, Error> {
 455         self.chunk_store.cond_touch_chunk(digest, fail_if_not_exist)
 456     }
 457
 458     pub fn insert_chunk(
 459         &self,
 460         chunk: &DataBlob,
 461         digest: &[u8; 32],
 462     ) -> Result<(bool, u64), Error> {
 463         self.chunk_store.insert_chunk(chunk, digest)
 464     }
 465 }