X-Git-Url: http://git.vrable.net/?a=blobdiff_plain;f=scandir.cc;h=9727133235a7054dc05cdfe5023376d4f5f98c61;hb=722040937d7709a7b976bf57b045dac6d703c893;hp=49d8c329351ee92d55ddcbf5e006a210850c96e7;hpb=6bb973b965a30832c3e2d9f6a24e80d3309ef89e;p=cumulus.git diff --git a/scandir.cc b/scandir.cc index 49d8c32..9727133 100644 --- a/scandir.cc +++ b/scandir.cc @@ -28,7 +28,6 @@ #include "metadata.h" #include "store.h" #include "sha1.h" -#include "statcache.h" #include "util.h" using std::list; @@ -55,13 +54,14 @@ static char *block_buf; * invocations to help in creating incremental snapshots. */ LocalDb *db; -/* Stat cache, which stored data locally to speed the backup process by quickly - * skipping files which have not changed. */ -StatCache *statcache; - /* Keep track of all segments which are needed to reconstruct the snapshot. */ std::set segment_list; +/* Snapshot intent: 1=daily, 7=weekly, etc. This is not used directly, but is + * stored in the local database and can help guide segment cleaning and + * snapshot expiration policies. */ +double snapshot_intent = 1.0; + /* Selection of files to include/exclude in the snapshot. */ std::list includes; // Paths in which files should be saved std::list excludes; // Paths which will not be saved @@ -119,9 +119,9 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path, * re-reading the entire contents. */ bool cached = false; - if (statcache->Find(path, &stat_buf)) { + if (metawriter->find(path) && metawriter->is_unchanged(&stat_buf)) { cached = true; - const list &blocks = statcache->get_blocks(); + list blocks = metawriter->get_blocks(); /* If any of the blocks in the object have been expired, then we should * fall back to fully reading in the file. */ @@ -137,12 +137,13 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path, /* If everything looks okay, use the cached information */ if (cached) { - file_info["checksum"] = statcache->get_checksum(); + file_info["checksum"] = metawriter->get_checksum(); for (list::const_iterator i = blocks.begin(); i != blocks.end(); ++i) { const ObjectReference &ref = *i; object_list.push_back(ref.to_string()); - segment_list.insert(ref.get_segment()); + if (ref.is_normal()) + add_segment(ref.get_segment()); db->UseObject(ref); } size = stat_buf.st_size; @@ -165,16 +166,34 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path, hash.process(block_buf, bytes); + // Sparse file processing: if we read a block of all zeroes, encode + // that explicitly. + bool all_zero = true; + for (int i = 0; i < bytes; i++) { + if (block_buf[i] != 0) { + all_zero = false; + break; + } + } + // Either find a copy of this block in an already-existing segment, // or index it so it can be re-used in the future double block_age = 0.0; + ObjectReference ref; + SHA1Checksum block_hash; block_hash.process(block_buf, bytes); string block_csum = block_hash.checksum_str(); - ObjectReference ref = db->FindObject(block_csum, bytes); + + if (all_zero) { + ref = ObjectReference(ObjectReference::REF_ZERO); + ref.set_range(0, bytes); + } else { + ObjectReference ref = db->FindObject(block_csum, bytes); + } // Store a copy of the object if one does not yet exist - if (ref.get_segment().size() == 0) { + if (ref.is_null()) { LbsObject *o = new LbsObject; int object_group; @@ -209,11 +228,13 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path, o->write(tss); ref = o->get_ref(); db->StoreObject(ref, block_csum, bytes, block_age); + ref.set_range(0, bytes); delete o; } object_list.push_back(ref.to_string()); - segment_list.insert(ref.get_segment()); + if (ref.is_normal()) + add_segment(ref.get_segment()); db->UseObject(ref); size += bytes; @@ -227,35 +248,14 @@ int64_t dumpfile(int fd, dictionary &file_info, const string &path, if (status != NULL) printf(" [%s]\n", status); - statcache->Save(path, &stat_buf, file_info["checksum"], object_list); - - /* For files that only need to be broken apart into a few objects, store - * the list of objects directly. For larger files, store the data - * out-of-line and provide a pointer to the indrect object. */ - if (object_list.size() < 8) { - string blocklist = ""; - for (list::iterator i = object_list.begin(); - i != object_list.end(); ++i) { - if (i != object_list.begin()) - blocklist += " "; - blocklist += *i; - } - file_info["data"] = blocklist; - } else { - string blocklist = ""; - for (list::iterator i = object_list.begin(); - i != object_list.end(); ++i) { - blocklist += *i + "\n"; - } - - LbsObject *i = new LbsObject; - i->set_group("metadata"); - i->set_data(blocklist.data(), blocklist.size()); - i->write(tss); - file_info["data"] = "@" + i->get_name(); - segment_list.insert(i->get_ref().get_segment()); - delete i; + string blocklist = ""; + for (list::iterator i = object_list.begin(); + i != object_list.end(); ++i) { + if (i != object_list.begin()) + blocklist += "\n "; + blocklist += *i; } + file_info["data"] = blocklist; return size; } @@ -274,12 +274,20 @@ void dump_inode(const string& path, // Path within snapshot ssize_t len; printf("%s\n", path.c_str()); + metawriter->find(path); + file_info["name"] = uri_encode(path); file_info["mode"] = encode_int(stat_buf.st_mode & 07777, 8); + file_info["ctime"] = encode_int(stat_buf.st_ctime); file_info["mtime"] = encode_int(stat_buf.st_mtime); file_info["user"] = encode_int(stat_buf.st_uid); file_info["group"] = encode_int(stat_buf.st_gid); + time_t now = time(NULL); + if (now - stat_buf.st_ctime < 30 || now - stat_buf.st_mtime < 30) + if ((stat_buf.st_mode & S_IFMT) != S_IFDIR) + file_info["volatile"] = "1"; + struct passwd *pwd = getpwuid(stat_buf.st_uid); if (pwd != NULL) { file_info["user"] += " (" + uri_encode(pwd->pw_name) + ")"; @@ -292,11 +300,12 @@ void dump_inode(const string& path, // Path within snapshot if (stat_buf.st_nlink > 1 && (stat_buf.st_mode & S_IFMT) != S_IFDIR) { file_info["links"] = encode_int(stat_buf.st_nlink); - file_info["inode"] = encode_int(major(stat_buf.st_dev)) - + "/" + encode_int(minor(stat_buf.st_dev)) - + "/" + encode_int(stat_buf.st_ino); } + file_info["inode"] = encode_int(major(stat_buf.st_dev)) + + "/" + encode_int(minor(stat_buf.st_dev)) + + "/" + encode_int(stat_buf.st_ino); + char inode_type; switch (stat_buf.st_mode & S_IFMT) { @@ -343,6 +352,7 @@ void dump_inode(const string& path, // Path within snapshot if (file_size != stat_buf.st_size) { fprintf(stderr, "Warning: Size of %s changed during reading\n", path.c_str()); + file_info["volatile"] = "1"; } break; @@ -357,7 +367,7 @@ void dump_inode(const string& path, // Path within snapshot file_info["type"] = string(1, inode_type); - metawriter->add(path, file_info); + metawriter->add(file_info); } void scanfile(const string& path, bool include) @@ -553,7 +563,9 @@ void usage(const char *program) " (defaults to \".bz2\")\n" " --signature-filter=COMMAND\n" " program though which to filter descriptor\n" - " --scheme=NAME optional name for this snapshot\n", + " --scheme=NAME optional name for this snapshot\n" + " --intent=FLOAT intended backup type: 1=daily, 7=weekly, ...\n" + " (defaults to \"1\")\n", lbs_version, program ); } @@ -574,6 +586,7 @@ int main(int argc, char *argv[]) {"dest", 1, 0, 0}, // 4 {"scheme", 1, 0, 0}, // 5 {"signature-filter", 1, 0, 0}, // 6 + {"intent", 1, 0, 0}, // 7 {NULL, 0, 0, 0}, }; @@ -609,6 +622,11 @@ int main(int argc, char *argv[]) case 6: // --signature-filter signature_filter = optarg; break; + case 7: // --intent + snapshot_intent = atof(optarg); + if (snapshot_intent <= 0) + snapshot_intent = 1; + break; default: fprintf(stderr, "Unhandled long option!\n"); return 1; @@ -679,16 +697,16 @@ int main(int argc, char *argv[]) string database_path = localdb_dir + "/localdb.sqlite"; db = new LocalDb; db->Open(database_path.c_str(), desc_buf, - backup_scheme.size() ? backup_scheme.c_str() : NULL); + backup_scheme.size() ? backup_scheme.c_str() : NULL, + snapshot_intent); tss = new TarSegmentStore(backup_dest, db); - metawriter = new MetadataWriter(tss); - /* Initialize the stat cache, for skipping over unchanged files. */ - statcache = new StatCache; - statcache->Open(localdb_dir.c_str(), desc_buf, - backup_scheme.size() ? backup_scheme.c_str() : NULL); + metawriter = new MetadataWriter(tss, localdb_dir.c_str(), desc_buf, + backup_scheme.size() + ? backup_scheme.c_str() + : NULL); scanfile(".", false); @@ -696,9 +714,6 @@ int main(int argc, char *argv[]) add_segment(root_ref.get_segment()); string backup_root = root_ref.to_string(); - statcache->Close(); - delete statcache; - delete metawriter; tss->sync();