| Line |
Branch |
Exec |
Source |
| 1 |
|
|
/** |
| 2 |
|
|
* This file is part of the CernVM File System |
| 3 |
|
|
*/ |
| 4 |
|
|
|
| 5 |
|
|
#include "swissknife_ingest.h" |
| 6 |
|
|
|
| 7 |
|
|
#include <fcntl.h> |
| 8 |
|
|
#include <unistd.h> |
| 9 |
|
|
|
| 10 |
|
|
#include <memory> |
| 11 |
|
|
#include <string> |
| 12 |
|
|
#include <vector> |
| 13 |
|
|
|
| 14 |
|
|
#include "catalog_virtual.h" |
| 15 |
|
|
#include "manifest.h" |
| 16 |
|
|
#include "statistics.h" |
| 17 |
|
|
#include "statistics_database.h" |
| 18 |
|
|
#include "swissknife_ingest_gc.h" |
| 19 |
|
|
#include "sync_mediator.h" |
| 20 |
|
|
#include "sync_union.h" |
| 21 |
|
|
#include "sync_union_tarball.h" |
| 22 |
|
|
#include "util/capabilities.h" |
| 23 |
|
|
#include "util/logging.h" |
| 24 |
|
|
#include "util/posix.h" |
| 25 |
|
|
|
| 26 |
|
|
/* |
| 27 |
|
|
* Many of the options possible to set in the ArgumentList are not actually used |
| 28 |
|
|
* by the ingest command since they are not part of its interface, hence those |
| 29 |
|
|
* unused options cannot be set by the shell script. Of course if there is the |
| 30 |
|
|
* necessitty those parameters can be added and managed. |
| 31 |
|
|
* At the moment this approach worked fine and didn't add much complexity, |
| 32 |
|
|
* however if yet another command will need to use a similar approach it would |
| 33 |
|
|
* be good to consider creating different options handler for each command. |
| 34 |
|
|
*/ |
| 35 |
|
✗ |
int swissknife::Ingest::Main(const swissknife::ArgumentList &args) { |
| 36 |
|
✗ |
const std::string start_time = GetGMTimestamp(); |
| 37 |
|
|
|
| 38 |
|
✗ |
SyncParameters params; |
| 39 |
|
✗ |
params.dir_rdonly = MakeCanonicalPath(*args.find('c')->second); |
| 40 |
|
✗ |
params.dir_temp = MakeCanonicalPath(*args.find('t')->second); |
| 41 |
|
✗ |
params.base_hash = shash::MkFromHexPtr(shash::HexPtr(*args.find('b')->second), |
| 42 |
|
|
shash::kSuffixCatalog); |
| 43 |
|
✗ |
params.stratum0 = *args.find('w')->second; |
| 44 |
|
✗ |
params.manifest_path = *args.find('o')->second; |
| 45 |
|
✗ |
params.spooler_definition = *args.find('r')->second; |
| 46 |
|
|
|
| 47 |
|
✗ |
params.public_keys = *args.find('K')->second; |
| 48 |
|
✗ |
params.repo_name = *args.find('N')->second; |
| 49 |
|
|
|
| 50 |
|
✗ |
if (args.find('T') != args.end()) { |
| 51 |
|
✗ |
params.tar_file = *args.find('T')->second; |
| 52 |
|
|
} |
| 53 |
|
✗ |
if (args.find('B') != args.end()) { |
| 54 |
|
✗ |
params.base_directory = *args.find('B')->second; |
| 55 |
|
|
} |
| 56 |
|
✗ |
if (args.find('D') != args.end()) { |
| 57 |
|
✗ |
params.to_delete = *args.find('D')->second; |
| 58 |
|
|
} |
| 59 |
|
|
|
| 60 |
|
✗ |
std::string gc_db_path; |
| 61 |
|
✗ |
std::vector<int64_t> gc_batch_ids; |
| 62 |
|
✗ |
if (args.find('Q') != args.end()) { |
| 63 |
|
✗ |
gc_db_path = *args.find('Q')->second; |
| 64 |
|
|
// Batch size comes from the -X flag (populated by the shell from |
| 65 |
|
|
// CVMFS_GC_DB_BATCH_SIZE in the repository server.conf). 0 / missing / |
| 66 |
|
|
// negative means "no limit": read all pending rows. Default is 1000. |
| 67 |
|
✗ |
int gc_db_batch_size = 1000; |
| 68 |
|
✗ |
if (args.find('X') != args.end()) { |
| 69 |
|
✗ |
gc_db_batch_size = static_cast<int>( |
| 70 |
|
✗ |
String2Int64(*args.find('X')->second)); |
| 71 |
|
✗ |
if (gc_db_batch_size < 0) |
| 72 |
|
✗ |
gc_db_batch_size = 0; |
| 73 |
|
|
} |
| 74 |
|
✗ |
std::vector<std::string> gc_paths; |
| 75 |
|
✗ |
if (!ReadGCDatabase(gc_db_path, &gc_paths, &gc_batch_ids, |
| 76 |
|
|
gc_db_batch_size)) { |
| 77 |
|
✗ |
PrintError("Swissknife Ingest: failed to read GC database"); |
| 78 |
|
✗ |
return 1; |
| 79 |
|
|
} |
| 80 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStdout, |
| 81 |
|
|
"Swissknife Ingest: Read %lu paths from GC database %s " |
| 82 |
|
|
"(batch size %d)", |
| 83 |
|
|
gc_paths.size(), gc_db_path.c_str(), gc_db_batch_size); |
| 84 |
|
|
// Append GC paths to the existing to_delete string using /// delimiter |
| 85 |
|
✗ |
for (size_t i = 0; i < gc_paths.size(); ++i) { |
| 86 |
|
✗ |
if (!params.to_delete.empty()) { |
| 87 |
|
✗ |
params.to_delete += "///"; |
| 88 |
|
|
} |
| 89 |
|
✗ |
params.to_delete += gc_paths[i]; |
| 90 |
|
|
} |
| 91 |
|
|
// GC database implies fast delete |
| 92 |
|
✗ |
if (!gc_paths.empty()) { |
| 93 |
|
✗ |
params.fast_delete = true; |
| 94 |
|
|
} |
| 95 |
|
|
} |
| 96 |
|
|
|
| 97 |
|
✗ |
if (args.find('O') != args.end()) { |
| 98 |
|
✗ |
params.generate_legacy_bulk_chunks = true; |
| 99 |
|
|
} |
| 100 |
|
✗ |
if (args.find('j') != args.end()) { |
| 101 |
|
✗ |
params.enable_mtime_ns = true; |
| 102 |
|
|
} |
| 103 |
|
✗ |
if (args.find('f') != args.end()) { |
| 104 |
|
✗ |
params.fast_delete = true; |
| 105 |
|
|
} |
| 106 |
|
✗ |
if (args.find('m') != args.end()) { |
| 107 |
|
✗ |
params.tolerate_missing_hardlinks = true; |
| 108 |
|
|
} |
| 109 |
|
✗ |
shash::Algorithms hash_algorithm = shash::kSha1; |
| 110 |
|
✗ |
if (args.find('e') != args.end()) { |
| 111 |
|
✗ |
hash_algorithm = shash::ParseHashAlgorithm(*args.find('e')->second); |
| 112 |
|
✗ |
if (hash_algorithm == shash::kAny) { |
| 113 |
|
✗ |
PrintError("Swissknife Ingest: unknown hash algorithm"); |
| 114 |
|
✗ |
return 1; |
| 115 |
|
|
} |
| 116 |
|
|
} |
| 117 |
|
✗ |
if (args.find('Z') != args.end()) { |
| 118 |
|
✗ |
params.compression_alg = zlib::ParseCompressionAlgorithm( |
| 119 |
|
✗ |
*args.find('Z')->second); |
| 120 |
|
|
} |
| 121 |
|
✗ |
if (args.find('U') != args.end()) { |
| 122 |
|
✗ |
params.uid = static_cast<uid_t>(String2Int64(*args.find('U')->second)); |
| 123 |
|
|
} |
| 124 |
|
✗ |
if (args.find('G') != args.end()) { |
| 125 |
|
✗ |
params.gid = static_cast<gid_t>(String2Int64(*args.find('G')->second)); |
| 126 |
|
|
} |
| 127 |
|
|
|
| 128 |
|
✗ |
const bool create_catalog = args.find('C') != args.end(); |
| 129 |
|
|
|
| 130 |
|
✗ |
params.nested_kcatalog_limit = SyncParameters::kDefaultNestedKcatalogLimit; |
| 131 |
|
✗ |
params.root_kcatalog_limit = SyncParameters::kDefaultRootKcatalogLimit; |
| 132 |
|
✗ |
params.file_mbyte_limit = SyncParameters::kDefaultFileMbyteLimit; |
| 133 |
|
|
|
| 134 |
|
✗ |
params.branched_catalog = false; // could be true? |
| 135 |
|
|
|
| 136 |
|
✗ |
if (args.find('P') != args.end()) { |
| 137 |
|
✗ |
params.session_token_file = *args.find('P')->second; |
| 138 |
|
|
} |
| 139 |
|
|
|
| 140 |
|
✗ |
if (args.find('H') != args.end()) { |
| 141 |
|
✗ |
params.key_file = *args.find('H')->second; |
| 142 |
|
|
} |
| 143 |
|
|
|
| 144 |
|
✗ |
const bool upload_statsdb = (args.count('I') > 0); |
| 145 |
|
|
|
| 146 |
|
✗ |
perf::StatisticsTemplate publish_statistics("publish", this->statistics()); |
| 147 |
|
✗ |
StatisticsDatabase *stats_db = StatisticsDatabase::OpenStandardDB( |
| 148 |
|
|
params.repo_name); |
| 149 |
|
|
|
| 150 |
|
|
upload::SpoolerDefinition spooler_definition( |
| 151 |
|
|
params.spooler_definition, hash_algorithm, params.compression_alg, |
| 152 |
|
✗ |
params.generate_legacy_bulk_chunks, params.use_file_chunking, |
| 153 |
|
|
params.min_file_chunk_size, params.avg_file_chunk_size, |
| 154 |
|
✗ |
params.max_file_chunk_size, params.session_token_file, params.key_file); |
| 155 |
|
✗ |
if (params.max_concurrent_write_jobs > 0) { |
| 156 |
|
|
spooler_definition |
| 157 |
|
✗ |
.number_of_concurrent_uploads = params.max_concurrent_write_jobs; |
| 158 |
|
|
} |
| 159 |
|
|
|
| 160 |
|
|
// Sanitize base_directory, removing any leading or trailing slashes |
| 161 |
|
|
// from non-root (!= "/") paths |
| 162 |
|
✗ |
params.base_directory = TrimString(params.base_directory, "/", kTrimAll); |
| 163 |
|
|
|
| 164 |
|
|
const upload::SpoolerDefinition spooler_definition_catalogs( |
| 165 |
|
✗ |
spooler_definition.Dup2DefaultCompression()); |
| 166 |
|
|
|
| 167 |
|
✗ |
params.spooler = upload::Spooler::Construct(spooler_definition, |
| 168 |
|
|
&publish_statistics); |
| 169 |
|
✗ |
if (NULL == params.spooler) |
| 170 |
|
✗ |
return 3; |
| 171 |
|
|
const std::unique_ptr<upload::Spooler> spooler_catalogs( |
| 172 |
|
|
upload::Spooler::Construct(spooler_definition_catalogs, |
| 173 |
|
✗ |
&publish_statistics)); |
| 174 |
|
✗ |
if (spooler_catalogs.get() == nullptr) |
| 175 |
|
✗ |
return 3; |
| 176 |
|
|
|
| 177 |
|
✗ |
const bool follow_redirects = (args.count('L') > 0); |
| 178 |
|
✗ |
const string proxy = (args.count('@') > 0) ? *args.find('@')->second : ""; |
| 179 |
|
✗ |
if (!InitDownloadManager(follow_redirects, proxy)) { |
| 180 |
|
✗ |
return 3; |
| 181 |
|
|
} |
| 182 |
|
|
|
| 183 |
|
✗ |
if (!InitSignatureManager(params.public_keys)) { |
| 184 |
|
✗ |
return 3; |
| 185 |
|
|
} |
| 186 |
|
|
|
| 187 |
|
✗ |
const bool with_gateway = spooler_definition.driver_type |
| 188 |
|
|
== upload::SpoolerDefinition::Gateway; |
| 189 |
|
|
|
| 190 |
|
|
// This may fail, in which case a warning is printed and the process continues |
| 191 |
|
✗ |
ObtainDacReadSearchCapability(); |
| 192 |
|
|
|
| 193 |
|
✗ |
std::unique_ptr<manifest::Manifest> manifest; |
| 194 |
|
✗ |
if (params.branched_catalog) { |
| 195 |
|
|
// Throw-away manifest |
| 196 |
|
✗ |
manifest.reset(new manifest::Manifest(shash::Any(), 0, "")); |
| 197 |
|
|
} else { |
| 198 |
|
✗ |
if (with_gateway) { |
| 199 |
|
✗ |
manifest.reset( |
| 200 |
|
✗ |
FetchRemoteManifest(params.stratum0, params.repo_name, shash::Any())); |
| 201 |
|
|
} else { |
| 202 |
|
✗ |
manifest.reset(FetchRemoteManifest(params.stratum0, params.repo_name, |
| 203 |
|
|
params.base_hash)); |
| 204 |
|
|
} |
| 205 |
|
|
} |
| 206 |
|
✗ |
if (manifest.get() == nullptr) { |
| 207 |
|
✗ |
return 3; |
| 208 |
|
|
} |
| 209 |
|
|
|
| 210 |
|
✗ |
const std::string old_root_hash = manifest->catalog_hash().ToString(true); |
| 211 |
|
|
|
| 212 |
|
|
catalog::WritableCatalogManager catalog_manager( |
| 213 |
|
|
params.base_hash, params.stratum0, params.dir_temp, |
| 214 |
|
✗ |
spooler_catalogs.get(), download_manager(), params.enforce_limits, |
| 215 |
|
|
params.nested_kcatalog_limit, params.root_kcatalog_limit, |
| 216 |
|
✗ |
params.file_mbyte_limit, statistics(), params.is_balanced, |
| 217 |
|
✗ |
params.max_weight, params.min_weight); |
| 218 |
|
✗ |
catalog_manager.Init(); |
| 219 |
|
|
|
| 220 |
|
✗ |
publish::SyncMediator mediator(&catalog_manager, ¶ms, publish_statistics); |
| 221 |
|
✗ |
LogCvmfs(kLogPublish, kLogStdout, "Swissknife Ingest: Processing changes..."); |
| 222 |
|
|
|
| 223 |
|
|
publish::SyncUnion *sync = new publish::SyncUnionTarball( |
| 224 |
|
|
&mediator, params.dir_rdonly, params.tar_file, params.base_directory, |
| 225 |
|
|
params.uid, params.gid, params.to_delete, create_catalog, |
| 226 |
|
✗ |
params.fast_delete, "///", params.tolerate_missing_hardlinks); |
| 227 |
|
|
|
| 228 |
|
✗ |
if (!sync->Initialize()) { |
| 229 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 230 |
|
|
"Swissknife Ingest: Initialization of the synchronisation " |
| 231 |
|
|
"engine failed"); |
| 232 |
|
✗ |
return 4; |
| 233 |
|
|
} |
| 234 |
|
|
|
| 235 |
|
✗ |
sync->Traverse(); |
| 236 |
|
|
|
| 237 |
|
✗ |
if (!params.authz_file.empty()) { |
| 238 |
|
|
LogCvmfs(kLogCvmfs, kLogDebug, |
| 239 |
|
|
"Swissknife Ingest: Adding contents of authz file %s to" |
| 240 |
|
|
" root catalog.", |
| 241 |
|
|
params.authz_file.c_str()); |
| 242 |
|
✗ |
const int fd = open(params.authz_file.c_str(), O_RDONLY); |
| 243 |
|
✗ |
if (fd == -1) { |
| 244 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 245 |
|
|
"Swissknife Ingest: Unable to open authz file (%s)" |
| 246 |
|
|
"from the publication process: %s", |
| 247 |
|
|
params.authz_file.c_str(), strerror(errno)); |
| 248 |
|
✗ |
return 7; |
| 249 |
|
|
} |
| 250 |
|
|
|
| 251 |
|
✗ |
std::string new_authz; |
| 252 |
|
✗ |
const bool read_successful = SafeReadToString(fd, &new_authz); |
| 253 |
|
✗ |
close(fd); |
| 254 |
|
|
|
| 255 |
|
✗ |
if (!read_successful) { |
| 256 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 257 |
|
|
"Swissknife Ingest: Failed to read authz file (%s): %s", |
| 258 |
|
|
params.authz_file.c_str(), strerror(errno)); |
| 259 |
|
✗ |
return 8; |
| 260 |
|
|
} |
| 261 |
|
|
|
| 262 |
|
✗ |
catalog_manager.SetVOMSAuthz(new_authz); |
| 263 |
|
|
} |
| 264 |
|
|
|
| 265 |
|
✗ |
if (!mediator.Commit(manifest.get())) { |
| 266 |
|
✗ |
PrintError("Swissknife Ingest: something went wrong during sync"); |
| 267 |
|
✗ |
stats_db->StorePublishStatistics(this->statistics(), start_time, false); |
| 268 |
|
✗ |
if (upload_statsdb) { |
| 269 |
|
✗ |
stats_db->UploadStatistics(params.spooler); |
| 270 |
|
|
} |
| 271 |
|
✗ |
return 5; |
| 272 |
|
|
} |
| 273 |
|
|
|
| 274 |
|
✗ |
perf::Counter *revision_counter = statistics()->Register( |
| 275 |
|
|
"publish.revision", "Published revision number"); |
| 276 |
|
✗ |
revision_counter->Set(catalog_manager.GetRootCatalog()->revision()); |
| 277 |
|
|
|
| 278 |
|
|
// finalize the spooler |
| 279 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStdout, |
| 280 |
|
|
"Swissknife Ingest: Wait for all uploads to finish"); |
| 281 |
|
✗ |
params.spooler->WaitForUpload(); |
| 282 |
|
✗ |
spooler_catalogs->WaitForUpload(); |
| 283 |
|
✗ |
params.spooler->FinalizeSession(false); |
| 284 |
|
|
|
| 285 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStdout, |
| 286 |
|
|
"Swissknife Ingest: Exporting repository manifest"); |
| 287 |
|
|
|
| 288 |
|
|
// We call FinalizeSession(true) this time, to also trigger the commit |
| 289 |
|
|
// operation on the gateway machine (if the upstream is of type "gw"). |
| 290 |
|
|
|
| 291 |
|
|
// Get the path of the new root catalog |
| 292 |
|
✗ |
const std::string new_root_hash = manifest->catalog_hash().ToString(true); |
| 293 |
|
|
|
| 294 |
|
✗ |
if (!spooler_catalogs->FinalizeSession(true, old_root_hash, new_root_hash, |
| 295 |
|
|
params.repo_tag)) { |
| 296 |
|
✗ |
PrintError("Swissknife Ingest: Failed to commit the transaction."); |
| 297 |
|
✗ |
stats_db->StorePublishStatistics(this->statistics(), start_time, false); |
| 298 |
|
✗ |
if (upload_statsdb) { |
| 299 |
|
✗ |
stats_db->UploadStatistics(params.spooler); |
| 300 |
|
|
} |
| 301 |
|
✗ |
return 9; |
| 302 |
|
|
} |
| 303 |
|
|
|
| 304 |
|
✗ |
stats_db->StorePublishStatistics(this->statistics(), start_time, true); |
| 305 |
|
✗ |
if (upload_statsdb) { |
| 306 |
|
✗ |
stats_db->UploadStatistics(params.spooler); |
| 307 |
|
|
} |
| 308 |
|
|
|
| 309 |
|
✗ |
delete params.spooler; |
| 310 |
|
|
|
| 311 |
|
✗ |
if (!manifest->Export(params.manifest_path)) { |
| 312 |
|
✗ |
PrintError("Swissknife Ingest: Failed to create new repository"); |
| 313 |
|
✗ |
return 6; |
| 314 |
|
|
} |
| 315 |
|
|
|
| 316 |
|
|
// Mark GC paths as deleted in the database after successful publication. |
| 317 |
|
|
// Only the rows that were just read (gc_batch_ids) are marked, so rows |
| 318 |
|
|
// added after the read by a concurrent scanner are preserved for the next |
| 319 |
|
|
// invocation. |
| 320 |
|
✗ |
if (!gc_db_path.empty()) { |
| 321 |
|
✗ |
if (MarkGCPathsDeleted(gc_db_path, gc_batch_ids)) { |
| 322 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStdout, |
| 323 |
|
|
"Swissknife Ingest: Marked %lu GC paths as deleted in %s", |
| 324 |
|
|
gc_batch_ids.size(), gc_db_path.c_str()); |
| 325 |
|
|
} else { |
| 326 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 327 |
|
|
"Swissknife Ingest: WARNING - Failed to mark GC paths as " |
| 328 |
|
|
"deleted in %s. Paths were removed from the repository but " |
| 329 |
|
|
"the database was not updated.", |
| 330 |
|
|
gc_db_path.c_str()); |
| 331 |
|
|
} |
| 332 |
|
|
} |
| 333 |
|
|
|
| 334 |
|
✗ |
return 0; |
| 335 |
|
|
} |
| 336 |
|
|
|