| Line |
Branch |
Exec |
Source |
| 1 |
|
|
/** |
| 2 |
|
|
* This file is part of the CernVM File System. |
| 3 |
|
|
* |
| 4 |
|
|
* This command processes a repository's catalog structure to detect and remove |
| 5 |
|
|
* outdated and/or unneeded data objects. |
| 6 |
|
|
*/ |
| 7 |
|
|
|
| 8 |
|
|
|
| 9 |
|
|
#include "swissknife_gc.h" |
| 10 |
|
|
|
| 11 |
|
|
#include <string> |
| 12 |
|
|
|
| 13 |
|
|
#include "garbage_collection/garbage_collector.h" |
| 14 |
|
|
#include "garbage_collection/gc_aux.h" |
| 15 |
|
|
#include "garbage_collection/hash_filter.h" |
| 16 |
|
|
#include "manifest.h" |
| 17 |
|
|
#include "path_filters/inclusion_spec.h" |
| 18 |
|
|
#include "reflog.h" |
| 19 |
|
|
#include "statistics_database.h" |
| 20 |
|
|
#include "upload_facility.h" |
| 21 |
|
|
#include "util/posix.h" |
| 22 |
|
|
#include "util/string.h" |
| 23 |
|
|
|
| 24 |
|
|
namespace swissknife { |
| 25 |
|
|
|
| 26 |
|
|
typedef HttpObjectFetcher<> ObjectFetcher; |
| 27 |
|
|
typedef CatalogTraversalParallel<ObjectFetcher> ReadonlyCatalogTraversal; |
| 28 |
|
|
typedef SmallhashFilter HashFilter; |
| 29 |
|
|
typedef GarbageCollector<ReadonlyCatalogTraversal, HashFilter> GC; |
| 30 |
|
|
typedef GarbageCollectorAux<ReadonlyCatalogTraversal, HashFilter> GCAux; |
| 31 |
|
|
typedef GC::Configuration GcConfig; |
| 32 |
|
|
|
| 33 |
|
|
|
| 34 |
|
✗ |
ParameterList CommandGc::GetParams() const { |
| 35 |
|
✗ |
ParameterList r; |
| 36 |
|
✗ |
r.push_back(Parameter::Mandatory('r', "repository url")); |
| 37 |
|
✗ |
r.push_back(Parameter::Mandatory('u', "spooler definition string")); |
| 38 |
|
✗ |
r.push_back(Parameter::Mandatory('n', "fully qualified repository name")); |
| 39 |
|
✗ |
r.push_back(Parameter::Mandatory('R', "path to reflog.chksum file")); |
| 40 |
|
✗ |
r.push_back(Parameter::Optional('h', "conserve <h> revisions")); |
| 41 |
|
✗ |
r.push_back(Parameter::Optional('z', "conserve revisions younger than <z>")); |
| 42 |
|
✗ |
r.push_back(Parameter::Optional('k', "repository master key(s) / dir")); |
| 43 |
|
✗ |
r.push_back(Parameter::Optional('t', "temporary directory")); |
| 44 |
|
✗ |
r.push_back(Parameter::Optional('L', "path to deletion log file")); |
| 45 |
|
✗ |
r.push_back(Parameter::Optional('N', "number of threads to use")); |
| 46 |
|
✗ |
r.push_back(Parameter::Optional('@', "proxy url")); |
| 47 |
|
✗ |
r.push_back(Parameter::Switch('d', "dry run")); |
| 48 |
|
✗ |
r.push_back(Parameter::Switch('l', "list objects to be removed")); |
| 49 |
|
✗ |
r.push_back(Parameter::Switch('I', "upload updated statistics DB file")); |
| 50 |
|
✗ |
r.push_back( |
| 51 |
|
✗ |
Parameter::Optional('E', "inclusion spec for partial replication")); |
| 52 |
|
✗ |
return r; |
| 53 |
|
|
} |
| 54 |
|
|
|
| 55 |
|
|
|
| 56 |
|
✗ |
int CommandGc::Main(const ArgumentList &args) { |
| 57 |
|
✗ |
const std::string start_time = GetGMTimestamp(); |
| 58 |
|
|
|
| 59 |
|
✗ |
const std::string &repo_url = *args.find('r')->second; |
| 60 |
|
✗ |
const std::string &spooler = *args.find('u')->second; |
| 61 |
|
✗ |
const std::string &repo_name = *args.find('n')->second; |
| 62 |
|
✗ |
const std::string &reflog_chksum_path = *args.find('R')->second; |
| 63 |
|
✗ |
shash::Any reflog_hash; |
| 64 |
|
✗ |
if (!manifest::Reflog::ReadChecksum(reflog_chksum_path, &reflog_hash)) { |
| 65 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "Could not read reflog checksum"); |
| 66 |
|
✗ |
return 1; |
| 67 |
|
|
} |
| 68 |
|
|
|
| 69 |
|
✗ |
const uint64_t revisions = (args.count('h') > 0) |
| 70 |
|
✗ |
? String2Int64(*args.find('h')->second) |
| 71 |
|
✗ |
: GcConfig::kFullHistory; |
| 72 |
|
✗ |
const time_t timestamp = (args.count('z') > 0) |
| 73 |
|
✗ |
? static_cast<time_t>( |
| 74 |
|
✗ |
String2Int64(*args.find('z')->second)) |
| 75 |
|
✗ |
: GcConfig::kNoTimestamp; |
| 76 |
|
✗ |
std::string repo_keys = (args.count('k') > 0) ? *args.find('k')->second : ""; |
| 77 |
|
✗ |
if (DirectoryExists(repo_keys)) |
| 78 |
|
✗ |
repo_keys = JoinStrings(FindFilesBySuffix(repo_keys, ".pub"), ":"); |
| 79 |
|
✗ |
const bool dry_run = (args.count('d') > 0); |
| 80 |
|
✗ |
const bool list_condemned_objects = (args.count('l') > 0); |
| 81 |
|
✗ |
const std::string temp_directory = (args.count('t') > 0) |
| 82 |
|
✗ |
? *args.find('t')->second |
| 83 |
|
✗ |
: "/tmp"; |
| 84 |
|
✗ |
const std::string deletion_log_path = (args.count('L') > 0) |
| 85 |
|
✗ |
? *args.find('L')->second |
| 86 |
|
✗ |
: ""; |
| 87 |
|
✗ |
const bool upload_statsdb = (args.count('I') > 0); |
| 88 |
|
✗ |
const unsigned int num_threads = (args.count('N') > 0) |
| 89 |
|
✗ |
? String2Uint64(*args.find('N')->second) |
| 90 |
|
✗ |
: 8; |
| 91 |
|
|
|
| 92 |
|
|
// Partial replication: no functional changes needed. Excluded subtrees are |
| 93 |
|
|
// pruned during snapshot, so both their catalogs and their objects are |
| 94 |
|
|
// absent. The catalog traversal already runs with ignore_load_failure=true |
| 95 |
|
|
// (see GarbageCollector::GetTraversalParams), so pruned nested catalogs are |
| 96 |
|
|
// silently skipped along with their entire subtree; and the sweep phase is a |
| 97 |
|
|
// no-op for objects that were never downloaded. |
| 98 |
|
✗ |
if (args.count('E') > 0) { |
| 99 |
|
✗ |
catalog::InclusionSpec *inclusion_spec = catalog::InclusionSpec::Create( |
| 100 |
|
✗ |
*args.find('E')->second); |
| 101 |
|
✗ |
if (inclusion_spec != NULL && inclusion_spec->IsValid()) { |
| 102 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStdout, |
| 103 |
|
|
"Partial replication mode: GC will skip pruned (excluded) " |
| 104 |
|
|
"catalogs and their objects"); |
| 105 |
|
|
} else { |
| 106 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 107 |
|
|
"Warning: could not parse inclusion spec '%s', " |
| 108 |
|
|
"continuing without partial replication awareness", |
| 109 |
|
|
args.find('E')->second->c_str()); |
| 110 |
|
|
} |
| 111 |
|
✗ |
delete inclusion_spec; |
| 112 |
|
|
} |
| 113 |
|
|
|
| 114 |
|
✗ |
if (timestamp == GcConfig::kNoTimestamp |
| 115 |
|
✗ |
&& revisions == GcConfig::kFullHistory) { |
| 116 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 117 |
|
|
"neither a timestamp nor history threshold given"); |
| 118 |
|
✗ |
return 1; |
| 119 |
|
|
} |
| 120 |
|
|
|
| 121 |
|
✗ |
const bool follow_redirects = false; |
| 122 |
|
✗ |
const std::string proxy = ((args.count('@') > 0) ? *args.find('@')->second |
| 123 |
|
✗ |
: ""); |
| 124 |
|
✗ |
if (!this->InitDownloadManager(follow_redirects, proxy) |
| 125 |
|
✗ |
|| !this->InitSignatureManager(repo_keys)) { |
| 126 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStderr, "failed to init repo connection"); |
| 127 |
|
✗ |
return 1; |
| 128 |
|
|
} |
| 129 |
|
|
|
| 130 |
|
|
ObjectFetcher object_fetcher(repo_name, |
| 131 |
|
|
repo_url, |
| 132 |
|
|
temp_directory, |
| 133 |
|
|
download_manager(), |
| 134 |
|
✗ |
signature_manager()); |
| 135 |
|
|
|
| 136 |
|
✗ |
std::unique_ptr<manifest::Manifest> manifest; |
| 137 |
|
✗ |
const ObjectFetcher::Failures retval = object_fetcher.FetchManifest( |
| 138 |
|
|
&manifest); |
| 139 |
|
✗ |
if (retval != ObjectFetcher::kFailOk) { |
| 140 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 141 |
|
|
"failed to load repository manifest " |
| 142 |
|
|
"(%d - %s)", |
| 143 |
|
|
retval, Code2Ascii(retval)); |
| 144 |
|
✗ |
return 1; |
| 145 |
|
|
} |
| 146 |
|
|
|
| 147 |
|
✗ |
if (!manifest->garbage_collectable()) { |
| 148 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 149 |
|
|
"repository does not allow garbage collection " |
| 150 |
|
|
"(manifest does not have the garbage collection flag set; " |
| 151 |
|
|
"ensure CVMFS_GARBAGE_COLLECTION=true in server.conf and " |
| 152 |
|
|
"run a transaction + publish to update the manifest)"); |
| 153 |
|
✗ |
return 1; |
| 154 |
|
|
} |
| 155 |
|
|
|
| 156 |
|
✗ |
std::unique_ptr<manifest::Reflog> reflog; |
| 157 |
|
✗ |
reflog.reset(FetchReflog(&object_fetcher, repo_name, reflog_hash)); |
| 158 |
|
✗ |
assert(reflog.get() != nullptr); |
| 159 |
|
|
|
| 160 |
|
✗ |
const upload::SpoolerDefinition spooler_definition(spooler, shash::kAny); |
| 161 |
|
|
const std::unique_ptr<upload::AbstractUploader> uploader( |
| 162 |
|
✗ |
upload::AbstractUploader::Construct(spooler_definition)); |
| 163 |
|
|
|
| 164 |
|
✗ |
if (uploader.get() == nullptr) { |
| 165 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "failed to initialize spooler for '%s'", |
| 166 |
|
|
spooler.c_str()); |
| 167 |
|
✗ |
return 1; |
| 168 |
|
|
} |
| 169 |
|
|
|
| 170 |
|
✗ |
FILE *deletion_log_file = NULL; |
| 171 |
|
✗ |
if (!deletion_log_path.empty()) { |
| 172 |
|
✗ |
deletion_log_file = fopen(deletion_log_path.c_str(), "a+"); |
| 173 |
|
✗ |
if (NULL == deletion_log_file) { |
| 174 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 175 |
|
|
"failed to open deletion log file " |
| 176 |
|
|
"(errno: %d)", |
| 177 |
|
|
errno); |
| 178 |
|
✗ |
uploader->TearDown(); |
| 179 |
|
✗ |
return 1; |
| 180 |
|
|
} |
| 181 |
|
|
} |
| 182 |
|
|
|
| 183 |
|
✗ |
const bool extended_stats = StatisticsDatabase::GcExtendedStats(repo_name); |
| 184 |
|
|
|
| 185 |
|
✗ |
reflog->BeginTransaction(); |
| 186 |
|
|
|
| 187 |
|
✗ |
GcConfig config; |
| 188 |
|
✗ |
config.uploader = uploader.get(); |
| 189 |
|
✗ |
config.keep_history_depth = revisions; |
| 190 |
|
✗ |
config.keep_history_timestamp = timestamp; |
| 191 |
|
✗ |
config.dry_run = dry_run; |
| 192 |
|
✗ |
config.verbose = list_condemned_objects; |
| 193 |
|
✗ |
config.object_fetcher = &object_fetcher; |
| 194 |
|
✗ |
config.reflog = reflog.get(); |
| 195 |
|
✗ |
config.deleted_objects_logfile = deletion_log_file; |
| 196 |
|
✗ |
config.statistics = statistics(); |
| 197 |
|
✗ |
config.extended_stats = extended_stats; |
| 198 |
|
✗ |
config.num_threads = num_threads; |
| 199 |
|
|
|
| 200 |
|
✗ |
if (deletion_log_file != NULL) { |
| 201 |
|
✗ |
const int bytes_written = fprintf(deletion_log_file, |
| 202 |
|
|
"# Garbage Collection started at %s\n", |
| 203 |
|
✗ |
StringifyTime(time(NULL), true).c_str()); |
| 204 |
|
✗ |
if (bytes_written < 0) { |
| 205 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 206 |
|
|
"failed to write to deletion log '%s' " |
| 207 |
|
|
"(errno: %d)", |
| 208 |
|
|
deletion_log_path.c_str(), errno); |
| 209 |
|
✗ |
uploader->TearDown(); |
| 210 |
|
✗ |
return 1; |
| 211 |
|
|
} |
| 212 |
|
|
} |
| 213 |
|
|
|
| 214 |
|
✗ |
StatisticsDatabase *stats_db = StatisticsDatabase::OpenStandardDB(repo_name); |
| 215 |
|
|
|
| 216 |
|
|
// File catalogs |
| 217 |
|
✗ |
GC collector(config); |
| 218 |
|
✗ |
collector.UseReflogTimestamps(); |
| 219 |
|
✗ |
bool success = collector.Collect(); |
| 220 |
|
|
|
| 221 |
|
✗ |
if (!success) { |
| 222 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "garbage collection failed"); |
| 223 |
|
✗ |
if (!dry_run) { |
| 224 |
|
✗ |
stats_db->StoreGCStatistics(this->statistics(), start_time, false); |
| 225 |
|
✗ |
if (upload_statsdb) { |
| 226 |
|
✗ |
stats_db->UploadStatistics(uploader.get()); |
| 227 |
|
|
} |
| 228 |
|
|
} |
| 229 |
|
✗ |
uploader->TearDown(); |
| 230 |
|
✗ |
return 1; |
| 231 |
|
|
} |
| 232 |
|
|
|
| 233 |
|
|
// Tag databases, meta infos, certificates |
| 234 |
|
✗ |
HashFilter preserved_objects; |
| 235 |
|
✗ |
preserved_objects.Fill(manifest->certificate()); |
| 236 |
|
✗ |
preserved_objects.Fill(manifest->history()); |
| 237 |
|
✗ |
preserved_objects.Fill(manifest->meta_info()); |
| 238 |
|
✗ |
GCAux collector_aux(config); |
| 239 |
|
✗ |
success = collector_aux.CollectOlderThan(collector.oldest_trunk_catalog(), |
| 240 |
|
|
preserved_objects); |
| 241 |
|
✗ |
if (!success) { |
| 242 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, |
| 243 |
|
|
"garbage collection of auxiliary files failed"); |
| 244 |
|
✗ |
if (!dry_run) { |
| 245 |
|
✗ |
stats_db->StoreGCStatistics(this->statistics(), start_time, false); |
| 246 |
|
✗ |
if (upload_statsdb) { |
| 247 |
|
✗ |
stats_db->UploadStatistics(uploader.get()); |
| 248 |
|
|
} |
| 249 |
|
|
} |
| 250 |
|
✗ |
uploader->TearDown(); |
| 251 |
|
✗ |
return 1; |
| 252 |
|
|
} |
| 253 |
|
|
|
| 254 |
|
|
// As of here: garbage collection succeeded, cleanup & commit |
| 255 |
|
|
|
| 256 |
|
✗ |
if (deletion_log_file != NULL) { |
| 257 |
|
✗ |
const int bytes_written = fprintf(deletion_log_file, |
| 258 |
|
|
"# Garbage Collection finished at %s\n\n", |
| 259 |
|
✗ |
StringifyTime(time(NULL), true).c_str()); |
| 260 |
|
✗ |
assert(bytes_written >= 0); |
| 261 |
|
✗ |
fclose(deletion_log_file); |
| 262 |
|
|
} |
| 263 |
|
|
|
| 264 |
|
✗ |
reflog->CommitTransaction(); |
| 265 |
|
|
// Has to be outside the transaction |
| 266 |
|
✗ |
success = reflog->Vacuum(); |
| 267 |
|
✗ |
assert(success); |
| 268 |
|
✗ |
reflog->DropDatabaseFileOwnership(); |
| 269 |
|
✗ |
const std::string reflog_db = reflog->database_file(); |
| 270 |
|
✗ |
reflog.reset(); |
| 271 |
|
|
|
| 272 |
|
✗ |
if (!dry_run) { |
| 273 |
|
✗ |
uploader->UploadFile(reflog_db, ".cvmfsreflog"); |
| 274 |
|
✗ |
manifest::Reflog::HashDatabase(reflog_db, &reflog_hash); |
| 275 |
|
✗ |
uploader->WaitForUpload(); |
| 276 |
|
✗ |
manifest::Reflog::WriteChecksum(reflog_chksum_path, reflog_hash); |
| 277 |
|
|
} |
| 278 |
|
|
|
| 279 |
|
✗ |
unlink(reflog_db.c_str()); |
| 280 |
|
|
|
| 281 |
|
✗ |
if (uploader->GetNumberOfErrors() > 0 && !dry_run) { |
| 282 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "failed to upload updated Reflog"); |
| 283 |
|
|
|
| 284 |
|
✗ |
stats_db->StoreGCStatistics(this->statistics(), start_time, false); |
| 285 |
|
✗ |
if (upload_statsdb) { |
| 286 |
|
✗ |
stats_db->UploadStatistics(uploader.get()); |
| 287 |
|
|
} |
| 288 |
|
|
|
| 289 |
|
✗ |
uploader->TearDown(); |
| 290 |
|
✗ |
return 1; |
| 291 |
|
|
} |
| 292 |
|
|
|
| 293 |
|
✗ |
if (!dry_run) { |
| 294 |
|
✗ |
stats_db->StoreGCStatistics(this->statistics(), start_time, true); |
| 295 |
|
✗ |
if (upload_statsdb) { |
| 296 |
|
✗ |
stats_db->UploadStatistics(uploader.get()); |
| 297 |
|
|
} |
| 298 |
|
|
} |
| 299 |
|
✗ |
uploader->TearDown(); |
| 300 |
|
✗ |
return 0; |
| 301 |
|
|
} |
| 302 |
|
|
|
| 303 |
|
|
} // namespace swissknife |
| 304 |
|
|
|