| Line |
Branch |
Exec |
Source |
| 1 |
|
|
/** |
| 2 |
|
|
* This file is part of the CernVM File System. |
| 3 |
|
|
* |
| 4 |
|
|
* Careful: any real schema migration as of now requires taking care of |
| 5 |
|
|
* hash algorithm |
| 6 |
|
|
*/ |
| 7 |
|
|
|
| 8 |
|
|
#include "swissknife_migrate.h" |
| 9 |
|
|
|
| 10 |
|
|
#include <sys/resource.h> |
| 11 |
|
|
#include <unistd.h> |
| 12 |
|
|
|
| 13 |
|
|
#include "catalog_rw.h" |
| 14 |
|
|
#include "catalog_sql.h" |
| 15 |
|
|
#include "catalog_virtual.h" |
| 16 |
|
|
#include "compression.h" |
| 17 |
|
|
#include "crypto/hash.h" |
| 18 |
|
|
#include "swissknife_history.h" |
| 19 |
|
|
#include "util/concurrency.h" |
| 20 |
|
|
#include "util/logging.h" |
| 21 |
|
|
|
| 22 |
|
|
using namespace std; // NOLINT |
| 23 |
|
|
|
| 24 |
|
|
namespace swissknife { |
| 25 |
|
|
|
| 26 |
|
|
catalog::DirectoryEntry CommandMigrate::nested_catalog_marker_; |
| 27 |
|
|
|
| 28 |
|
✗ |
CommandMigrate::CommandMigrate() : |
| 29 |
|
✗ |
file_descriptor_limit_(8192), |
| 30 |
|
✗ |
catalog_count_(0), |
| 31 |
|
✗ |
has_committed_new_revision_(false), |
| 32 |
|
✗ |
uid_(0), |
| 33 |
|
✗ |
gid_(0), |
| 34 |
|
✗ |
root_catalog_(NULL) |
| 35 |
|
|
{ |
| 36 |
|
✗ |
atomic_init32(&catalogs_processed_); |
| 37 |
|
|
} |
| 38 |
|
|
|
| 39 |
|
|
|
| 40 |
|
✗ |
ParameterList CommandMigrate::GetParams() const { |
| 41 |
|
✗ |
ParameterList r; |
| 42 |
|
✗ |
r.push_back(Parameter::Mandatory('v', |
| 43 |
|
|
"migration base version ( 2.0.x | 2.1.7 | chown | hardlink | bulkhash | " |
| 44 |
|
|
"stats)")); |
| 45 |
|
✗ |
r.push_back(Parameter::Mandatory('r', |
| 46 |
|
|
"repository URL (absolute local path or remote URL)")); |
| 47 |
|
✗ |
r.push_back(Parameter::Mandatory('u', "upstream definition string")); |
| 48 |
|
✗ |
r.push_back(Parameter::Mandatory('o', "manifest output file")); |
| 49 |
|
✗ |
r.push_back(Parameter::Mandatory('t', |
| 50 |
|
|
"temporary directory for catalog decompress")); |
| 51 |
|
✗ |
r.push_back(Parameter::Optional('p', |
| 52 |
|
|
"user id to be used for this repository")); |
| 53 |
|
✗ |
r.push_back(Parameter::Optional('g', |
| 54 |
|
|
"group id to be used for this repository")); |
| 55 |
|
✗ |
r.push_back(Parameter::Optional('n', "fully qualified repository name")); |
| 56 |
|
✗ |
r.push_back(Parameter::Optional('h', "root hash (other than trunk)")); |
| 57 |
|
✗ |
r.push_back(Parameter::Optional('k', "repository master key(s)")); |
| 58 |
|
✗ |
r.push_back(Parameter::Optional('i', "UID map for chown")); |
| 59 |
|
✗ |
r.push_back(Parameter::Optional('j', "GID map for chown")); |
| 60 |
|
✗ |
r.push_back(Parameter::Optional('@', "proxy url")); |
| 61 |
|
✗ |
r.push_back(Parameter::Switch('f', "fix nested catalog transition points")); |
| 62 |
|
✗ |
r.push_back(Parameter::Switch('l', "disable linkcount analysis of files")); |
| 63 |
|
✗ |
r.push_back(Parameter::Switch('s', |
| 64 |
|
|
"enable collection of catalog statistics")); |
| 65 |
|
✗ |
return r; |
| 66 |
|
|
} |
| 67 |
|
|
|
| 68 |
|
|
|
| 69 |
|
✗ |
static void Error(const std::string &message) { |
| 70 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStderr, "%s", message.c_str()); |
| 71 |
|
|
} |
| 72 |
|
|
|
| 73 |
|
|
|
| 74 |
|
✗ |
static void Error(const std::string &message, |
| 75 |
|
|
const CommandMigrate::PendingCatalog *catalog) { |
| 76 |
|
✗ |
const std::string err_msg = message + "\n" |
| 77 |
|
✗ |
"Catalog: " + catalog->root_path(); |
| 78 |
|
✗ |
Error(err_msg); |
| 79 |
|
|
} |
| 80 |
|
|
|
| 81 |
|
|
|
| 82 |
|
✗ |
static void Error(const std::string &message, |
| 83 |
|
|
const catalog::SqlCatalog &statement, |
| 84 |
|
|
const CommandMigrate::PendingCatalog *catalog) { |
| 85 |
|
|
const std::string err_msg = |
| 86 |
|
✗ |
message + "\n" |
| 87 |
|
✗ |
"SQLite: " + StringifyInt(statement.GetLastError()) + |
| 88 |
|
✗ |
" - " + statement.GetLastErrorMsg(); |
| 89 |
|
✗ |
Error(err_msg, catalog); |
| 90 |
|
|
} |
| 91 |
|
|
|
| 92 |
|
|
|
| 93 |
|
✗ |
int CommandMigrate::Main(const ArgumentList &args) { |
| 94 |
|
✗ |
shash::Any manual_root_hash; |
| 95 |
|
✗ |
const std::string &migration_base = *args.find('v')->second; |
| 96 |
|
✗ |
const std::string &repo_url = *args.find('r')->second; |
| 97 |
|
✗ |
const std::string &spooler = *args.find('u')->second; |
| 98 |
|
✗ |
const std::string &manifest_path = *args.find('o')->second; |
| 99 |
|
✗ |
const std::string &tmp_dir = *args.find('t')->second; |
| 100 |
|
✗ |
const std::string &uid = (args.count('p') > 0) ? |
| 101 |
|
✗ |
*args.find('p')->second : |
| 102 |
|
✗ |
""; |
| 103 |
|
✗ |
const std::string &gid = (args.count('g') > 0) ? |
| 104 |
|
✗ |
*args.find('g')->second : |
| 105 |
|
✗ |
""; |
| 106 |
|
✗ |
const std::string &repo_name = (args.count('n') > 0) ? |
| 107 |
|
✗ |
*args.find('n')->second : |
| 108 |
|
✗ |
""; |
| 109 |
|
✗ |
const std::string &repo_keys = (args.count('k') > 0) ? |
| 110 |
|
✗ |
*args.find('k')->second : |
| 111 |
|
✗ |
""; |
| 112 |
|
✗ |
const std::string &uid_map_path = (args.count('i') > 0) ? |
| 113 |
|
✗ |
*args.find('i')->second : |
| 114 |
|
✗ |
""; |
| 115 |
|
✗ |
const std::string &gid_map_path = (args.count('j') > 0) ? |
| 116 |
|
✗ |
*args.find('j')->second : |
| 117 |
|
✗ |
""; |
| 118 |
|
✗ |
const bool fix_transition_points = (args.count('f') > 0); |
| 119 |
|
✗ |
const bool analyze_file_linkcounts = (args.count('l') == 0); |
| 120 |
|
✗ |
const bool collect_catalog_statistics = (args.count('s') > 0); |
| 121 |
|
✗ |
if (args.count('h') > 0) { |
| 122 |
|
✗ |
manual_root_hash = shash::MkFromHexPtr(shash::HexPtr( |
| 123 |
|
✗ |
*args.find('h')->second), shash::kSuffixCatalog); |
| 124 |
|
|
} |
| 125 |
|
|
|
| 126 |
|
|
// We might need a lot of file descriptors |
| 127 |
|
✗ |
if (!RaiseFileDescriptorLimit()) { |
| 128 |
|
✗ |
Error("Failed to raise file descriptor limits"); |
| 129 |
|
✗ |
return 2; |
| 130 |
|
|
} |
| 131 |
|
|
|
| 132 |
|
|
// Put SQLite into multithreaded mode |
| 133 |
|
✗ |
if (!ConfigureSQLite()) { |
| 134 |
|
✗ |
Error("Failed to preconfigure SQLite library"); |
| 135 |
|
✗ |
return 3; |
| 136 |
|
|
} |
| 137 |
|
|
|
| 138 |
|
|
// Create an upstream spooler |
| 139 |
|
✗ |
temporary_directory_ = tmp_dir; |
| 140 |
|
✗ |
const upload::SpoolerDefinition spooler_definition(spooler, shash::kSha1); |
| 141 |
|
✗ |
spooler_ = upload::Spooler::Construct(spooler_definition); |
| 142 |
|
✗ |
if (!spooler_.IsValid()) { |
| 143 |
|
✗ |
Error("Failed to create upstream Spooler."); |
| 144 |
|
✗ |
return 5; |
| 145 |
|
|
} |
| 146 |
|
✗ |
spooler_->RegisterListener(&CommandMigrate::UploadCallback, this); |
| 147 |
|
|
|
| 148 |
|
|
// Load the full catalog hierarchy |
| 149 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "Loading current catalog tree..."); |
| 150 |
|
|
|
| 151 |
|
✗ |
catalog_loading_stopwatch_.Start(); |
| 152 |
|
✗ |
bool loading_successful = false; |
| 153 |
|
✗ |
if (IsHttpUrl(repo_url)) { |
| 154 |
|
|
typedef HttpObjectFetcher<catalog::WritableCatalog> ObjectFetcher; |
| 155 |
|
|
|
| 156 |
|
✗ |
const bool follow_redirects = false; |
| 157 |
|
✗ |
const string proxy = (args.count('@') > 0) ? *args.find('@')->second : ""; |
| 158 |
|
✗ |
if (!this->InitDownloadManager(follow_redirects, proxy) || |
| 159 |
|
✗ |
!this->InitVerifyingSignatureManager(repo_keys)) { |
| 160 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStderr, "Failed to init repo connection"); |
| 161 |
|
✗ |
return 1; |
| 162 |
|
|
} |
| 163 |
|
|
|
| 164 |
|
|
ObjectFetcher fetcher(repo_name, |
| 165 |
|
|
repo_url, |
| 166 |
|
|
tmp_dir, |
| 167 |
|
|
download_manager(), |
| 168 |
|
✗ |
signature_manager()); |
| 169 |
|
|
|
| 170 |
|
✗ |
loading_successful = LoadCatalogs(manual_root_hash, &fetcher); |
| 171 |
|
✗ |
} else { |
| 172 |
|
|
typedef LocalObjectFetcher<catalog::WritableCatalog> ObjectFetcher; |
| 173 |
|
✗ |
ObjectFetcher fetcher(repo_url, tmp_dir); |
| 174 |
|
✗ |
loading_successful = LoadCatalogs(manual_root_hash, &fetcher); |
| 175 |
|
|
} |
| 176 |
|
✗ |
catalog_loading_stopwatch_.Stop(); |
| 177 |
|
|
|
| 178 |
|
✗ |
if (!loading_successful) { |
| 179 |
|
✗ |
Error("Failed to load catalog tree"); |
| 180 |
|
✗ |
return 4; |
| 181 |
|
|
} |
| 182 |
|
|
|
| 183 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "Loaded %d catalogs", catalog_count_); |
| 184 |
|
✗ |
assert(root_catalog_ != NULL); |
| 185 |
|
|
|
| 186 |
|
|
// Do the actual migration step |
| 187 |
|
✗ |
bool migration_succeeded = false; |
| 188 |
|
✗ |
if (migration_base == "2.0.x") { |
| 189 |
|
✗ |
if (!ReadPersona(uid, gid)) { |
| 190 |
|
✗ |
return 1; |
| 191 |
|
|
} |
| 192 |
|
|
|
| 193 |
|
|
// Generate and upload a nested catalog marker |
| 194 |
|
✗ |
if (!GenerateNestedCatalogMarkerChunk()) { |
| 195 |
|
✗ |
Error("Failed to create a nested catalog marker."); |
| 196 |
|
✗ |
return 6; |
| 197 |
|
|
} |
| 198 |
|
✗ |
spooler_->WaitForUpload(); |
| 199 |
|
|
|
| 200 |
|
|
// Configure the concurrent catalog migration facility |
| 201 |
|
✗ |
MigrationWorker_20x::worker_context context(temporary_directory_, |
| 202 |
|
|
collect_catalog_statistics, |
| 203 |
|
|
fix_transition_points, |
| 204 |
|
|
analyze_file_linkcounts, |
| 205 |
|
|
uid_, |
| 206 |
|
✗ |
gid_); |
| 207 |
|
|
migration_succeeded = |
| 208 |
|
✗ |
DoMigrationAndCommit<MigrationWorker_20x>(manifest_path, &context); |
| 209 |
|
✗ |
} else if (migration_base == "2.1.7") { |
| 210 |
|
✗ |
MigrationWorker_217::worker_context context(temporary_directory_, |
| 211 |
|
✗ |
collect_catalog_statistics); |
| 212 |
|
|
migration_succeeded = |
| 213 |
|
✗ |
DoMigrationAndCommit<MigrationWorker_217>(manifest_path, &context); |
| 214 |
|
✗ |
} else if (migration_base == "chown") { |
| 215 |
|
✗ |
UidMap uid_map; |
| 216 |
|
✗ |
GidMap gid_map; |
| 217 |
|
✗ |
if (!ReadPersonaMaps(uid_map_path, gid_map_path, &uid_map, &gid_map)) { |
| 218 |
|
✗ |
Error("Failed to read UID and/or GID map"); |
| 219 |
|
✗ |
return 1; |
| 220 |
|
|
} |
| 221 |
|
✗ |
ChownMigrationWorker::worker_context context(temporary_directory_, |
| 222 |
|
|
collect_catalog_statistics, |
| 223 |
|
|
uid_map, |
| 224 |
|
✗ |
gid_map); |
| 225 |
|
|
migration_succeeded = |
| 226 |
|
✗ |
DoMigrationAndCommit<ChownMigrationWorker>(manifest_path, &context); |
| 227 |
|
✗ |
} else if (migration_base == "hardlink") { |
| 228 |
|
|
HardlinkRemovalMigrationWorker::worker_context |
| 229 |
|
✗ |
context(temporary_directory_, collect_catalog_statistics); |
| 230 |
|
|
migration_succeeded = |
| 231 |
|
✗ |
DoMigrationAndCommit<HardlinkRemovalMigrationWorker>(manifest_path, |
| 232 |
|
|
&context); |
| 233 |
|
✗ |
} else if (migration_base == "bulkhash") { |
| 234 |
|
|
BulkhashRemovalMigrationWorker::worker_context |
| 235 |
|
✗ |
context(temporary_directory_, collect_catalog_statistics); |
| 236 |
|
|
migration_succeeded = |
| 237 |
|
✗ |
DoMigrationAndCommit<BulkhashRemovalMigrationWorker>(manifest_path, |
| 238 |
|
|
&context); |
| 239 |
|
✗ |
} else if (migration_base == "stats") { |
| 240 |
|
|
StatsMigrationWorker::worker_context context( |
| 241 |
|
✗ |
temporary_directory_, collect_catalog_statistics); |
| 242 |
|
|
migration_succeeded = |
| 243 |
|
✗ |
DoMigrationAndCommit<StatsMigrationWorker>(manifest_path, &context); |
| 244 |
|
✗ |
} else { |
| 245 |
|
✗ |
const std::string err_msg = "Unknown migration base: " + migration_base; |
| 246 |
|
✗ |
Error(err_msg); |
| 247 |
|
✗ |
return 1; |
| 248 |
|
|
} |
| 249 |
|
|
|
| 250 |
|
|
// Check if everything went well |
| 251 |
|
✗ |
if (!migration_succeeded) { |
| 252 |
|
✗ |
Error("Migration failed!"); |
| 253 |
|
✗ |
return 5; |
| 254 |
|
|
} |
| 255 |
|
|
|
| 256 |
|
|
// Analyze collected statistics |
| 257 |
|
✗ |
if (collect_catalog_statistics && has_committed_new_revision_) { |
| 258 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "\nCollected statistics results:"); |
| 259 |
|
✗ |
AnalyzeCatalogStatistics(); |
| 260 |
|
|
} |
| 261 |
|
|
|
| 262 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "\nCatalog Migration succeeded"); |
| 263 |
|
✗ |
return 0; |
| 264 |
|
|
} |
| 265 |
|
|
|
| 266 |
|
|
|
| 267 |
|
✗ |
bool CommandMigrate::ReadPersona(const std::string &uid, |
| 268 |
|
|
const std::string &gid) { |
| 269 |
|
✗ |
if (uid.empty()) { |
| 270 |
|
✗ |
Error("Please provide a user ID"); |
| 271 |
|
✗ |
return false; |
| 272 |
|
|
} |
| 273 |
|
✗ |
if (gid.empty()) { |
| 274 |
|
✗ |
Error("Please provide a group ID"); |
| 275 |
|
✗ |
return false; |
| 276 |
|
|
} |
| 277 |
|
|
|
| 278 |
|
✗ |
uid_ = String2Int64(uid); |
| 279 |
|
✗ |
gid_ = String2Int64(gid); |
| 280 |
|
✗ |
return true; |
| 281 |
|
|
} |
| 282 |
|
|
|
| 283 |
|
|
|
| 284 |
|
|
|
| 285 |
|
✗ |
bool CommandMigrate::ReadPersonaMaps(const std::string &uid_map_path, |
| 286 |
|
|
const std::string &gid_map_path, |
| 287 |
|
|
UidMap *uid_map, |
| 288 |
|
|
GidMap *gid_map) const { |
| 289 |
|
✗ |
if (!uid_map->Read(uid_map_path) || !uid_map->IsValid()) { |
| 290 |
|
✗ |
Error("Failed to read UID map"); |
| 291 |
|
✗ |
return false; |
| 292 |
|
|
} |
| 293 |
|
|
|
| 294 |
|
✗ |
if (!gid_map->Read(gid_map_path) || !gid_map->IsValid()) { |
| 295 |
|
✗ |
Error("Failed to read GID map"); |
| 296 |
|
✗ |
return false; |
| 297 |
|
|
} |
| 298 |
|
|
|
| 299 |
|
✗ |
if (uid_map->RuleCount() == 0 && !uid_map->HasDefault()) { |
| 300 |
|
✗ |
Error("UID map appears to be empty"); |
| 301 |
|
✗ |
return false; |
| 302 |
|
|
} |
| 303 |
|
|
|
| 304 |
|
✗ |
if (gid_map->RuleCount() == 0 && !gid_map->HasDefault()) { |
| 305 |
|
✗ |
Error("GID map appears to be empty"); |
| 306 |
|
✗ |
return false; |
| 307 |
|
|
} |
| 308 |
|
|
|
| 309 |
|
✗ |
return true; |
| 310 |
|
|
} |
| 311 |
|
|
|
| 312 |
|
|
|
| 313 |
|
✗ |
void CommandMigrate::UploadHistoryClosure( |
| 314 |
|
|
const upload::SpoolerResult &result, |
| 315 |
|
|
Future<shash::Any> *hash) |
| 316 |
|
|
{ |
| 317 |
|
✗ |
assert(!result.IsChunked()); |
| 318 |
|
✗ |
if (result.return_code != 0) { |
| 319 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStderr, "failed to upload history database (%d)", |
| 320 |
|
|
result.return_code); |
| 321 |
|
✗ |
hash->Set(shash::Any()); |
| 322 |
|
|
} else { |
| 323 |
|
✗ |
hash->Set(result.content_hash); |
| 324 |
|
|
} |
| 325 |
|
|
} |
| 326 |
|
|
|
| 327 |
|
|
|
| 328 |
|
✗ |
bool CommandMigrate::UpdateUndoTags( |
| 329 |
|
|
PendingCatalog *root_catalog, |
| 330 |
|
|
uint64_t revision, |
| 331 |
|
|
time_t timestamp, |
| 332 |
|
|
shash::Any *history_hash) |
| 333 |
|
|
{ |
| 334 |
|
✗ |
string filename_old = history_upstream_->filename(); |
| 335 |
|
✗ |
string filename_new = filename_old + ".new"; |
| 336 |
|
✗ |
bool retval = CopyPath2Path(filename_old, filename_new); |
| 337 |
|
✗ |
if (!retval) return false; |
| 338 |
|
|
UniquePtr<history::SqliteHistory> history( |
| 339 |
|
✗ |
history::SqliteHistory::OpenWritable(filename_new)); |
| 340 |
|
✗ |
history->TakeDatabaseFileOwnership(); |
| 341 |
|
|
|
| 342 |
|
✗ |
history::History::Tag tag_trunk; |
| 343 |
|
✗ |
bool exists = history->GetByName(CommandTag::kHeadTag, &tag_trunk); |
| 344 |
|
✗ |
if (exists) { |
| 345 |
|
✗ |
retval = history->Remove(CommandTag::kHeadTag); |
| 346 |
|
✗ |
if (!retval) return false; |
| 347 |
|
|
|
| 348 |
|
✗ |
history::History::Tag tag_trunk_previous = tag_trunk; |
| 349 |
|
✗ |
tag_trunk_previous.name = CommandTag::kPreviousHeadTag; |
| 350 |
|
✗ |
tag_trunk_previous.description = CommandTag::kPreviousHeadTagDescription; |
| 351 |
|
✗ |
history->Remove(CommandTag::kPreviousHeadTag); |
| 352 |
|
|
|
| 353 |
|
✗ |
tag_trunk.root_hash = root_catalog->new_catalog_hash; |
| 354 |
|
✗ |
tag_trunk.size = root_catalog->new_catalog_size; |
| 355 |
|
✗ |
tag_trunk.revision = revision; |
| 356 |
|
✗ |
tag_trunk.timestamp = timestamp; |
| 357 |
|
|
|
| 358 |
|
✗ |
retval = history->Insert(tag_trunk_previous); |
| 359 |
|
✗ |
if (!retval) return false; |
| 360 |
|
✗ |
retval = history->Insert(tag_trunk); |
| 361 |
|
✗ |
if (!retval) return false; |
| 362 |
|
|
} |
| 363 |
|
|
|
| 364 |
|
✗ |
history->SetPreviousRevision(manifest_upstream_->history()); |
| 365 |
|
✗ |
history->DropDatabaseFileOwnership(); |
| 366 |
|
✗ |
history.Destroy(); |
| 367 |
|
|
|
| 368 |
|
✗ |
Future<shash::Any> history_hash_new; |
| 369 |
|
✗ |
upload::Spooler::CallbackPtr callback = spooler_->RegisterListener( |
| 370 |
|
|
&CommandMigrate::UploadHistoryClosure, this, &history_hash_new); |
| 371 |
|
✗ |
spooler_->ProcessHistory(filename_new); |
| 372 |
|
✗ |
spooler_->WaitForUpload(); |
| 373 |
|
✗ |
spooler_->UnregisterListener(callback); |
| 374 |
|
✗ |
unlink(filename_new.c_str()); |
| 375 |
|
✗ |
*history_hash = history_hash_new.Get(); |
| 376 |
|
✗ |
if (history_hash->IsNull()) { |
| 377 |
|
✗ |
Error("failed to upload tag database"); |
| 378 |
|
✗ |
return false; |
| 379 |
|
|
} |
| 380 |
|
|
|
| 381 |
|
✗ |
return true; |
| 382 |
|
|
} |
| 383 |
|
|
|
| 384 |
|
|
|
| 385 |
|
|
template <class MigratorT> |
| 386 |
|
✗ |
bool CommandMigrate::DoMigrationAndCommit( |
| 387 |
|
|
const std::string &manifest_path, |
| 388 |
|
|
typename MigratorT::worker_context *context |
| 389 |
|
|
) { |
| 390 |
|
|
// Create a concurrent migration context for catalog migration |
| 391 |
|
✗ |
const unsigned int cpus = GetNumberOfCpuCores(); |
| 392 |
|
✗ |
ConcurrentWorkers<MigratorT> concurrent_migration(cpus, cpus * 10, context); |
| 393 |
|
|
|
| 394 |
|
✗ |
if (!concurrent_migration.Initialize()) { |
| 395 |
|
✗ |
Error("Failed to initialize worker migration system."); |
| 396 |
|
✗ |
return false; |
| 397 |
|
|
} |
| 398 |
|
✗ |
concurrent_migration.RegisterListener(&CommandMigrate::MigrationCallback, |
| 399 |
|
|
this); |
| 400 |
|
|
|
| 401 |
|
|
// Migrate catalogs recursively (starting with the deepest nested catalogs) |
| 402 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "\nMigrating catalogs..."); |
| 403 |
|
✗ |
PendingCatalog *root_catalog = new PendingCatalog(root_catalog_); |
| 404 |
|
✗ |
migration_stopwatch_.Start(); |
| 405 |
|
✗ |
ConvertCatalogsRecursively(root_catalog, &concurrent_migration); |
| 406 |
|
✗ |
concurrent_migration.WaitForEmptyQueue(); |
| 407 |
|
✗ |
spooler_->WaitForUpload(); |
| 408 |
|
✗ |
spooler_->UnregisterListeners(); |
| 409 |
|
✗ |
migration_stopwatch_.Stop(); |
| 410 |
|
|
|
| 411 |
|
|
// check for possible errors during the migration process |
| 412 |
|
✗ |
const unsigned int errors = concurrent_migration.GetNumberOfFailedJobs() + |
| 413 |
|
✗ |
spooler_->GetNumberOfErrors(); |
| 414 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, |
| 415 |
|
|
"Catalog Migration finished with %d errors.", errors); |
| 416 |
|
✗ |
if (errors > 0) { |
| 417 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, |
| 418 |
|
|
"\nCatalog Migration produced errors\nAborting..."); |
| 419 |
|
✗ |
return false; |
| 420 |
|
|
} |
| 421 |
|
|
|
| 422 |
|
✗ |
if (root_catalog->was_updated.Get()) { |
| 423 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, |
| 424 |
|
|
"\nCommitting migrated repository revision..."); |
| 425 |
|
✗ |
manifest::Manifest manifest = *manifest_upstream_; |
| 426 |
|
✗ |
manifest.set_catalog_hash(root_catalog->new_catalog_hash); |
| 427 |
|
✗ |
manifest.set_catalog_size(root_catalog->new_catalog_size); |
| 428 |
|
✗ |
manifest.set_root_path(root_catalog->root_path()); |
| 429 |
|
✗ |
const catalog::Catalog* new_catalog = (root_catalog->HasNew()) |
| 430 |
|
✗ |
? root_catalog->new_catalog |
| 431 |
|
|
: root_catalog->old_catalog; |
| 432 |
|
✗ |
manifest.set_ttl(new_catalog->GetTTL()); |
| 433 |
|
✗ |
manifest.set_revision(new_catalog->GetRevision()); |
| 434 |
|
|
|
| 435 |
|
|
// Commit the new (migrated) repository revision... |
| 436 |
|
✗ |
if (history_upstream_.IsValid()) { |
| 437 |
|
✗ |
shash::Any history_hash(manifest_upstream_->history()); |
| 438 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout | kLogNoLinebreak, |
| 439 |
|
|
"Updating repository tag database... "); |
| 440 |
|
✗ |
if (!UpdateUndoTags(root_catalog, |
| 441 |
|
|
new_catalog->GetRevision(), |
| 442 |
|
✗ |
new_catalog->GetLastModified(), |
| 443 |
|
|
&history_hash)) |
| 444 |
|
|
{ |
| 445 |
|
✗ |
Error("Updating tag database failed.\nAborting..."); |
| 446 |
|
✗ |
return false; |
| 447 |
|
|
} |
| 448 |
|
✗ |
manifest.set_history(history_hash); |
| 449 |
|
✗ |
LogCvmfs(kLogCvmfs, kLogStdout, "%s", history_hash.ToString().c_str()); |
| 450 |
|
|
} |
| 451 |
|
|
|
| 452 |
|
✗ |
if (!manifest.Export(manifest_path)) { |
| 453 |
|
✗ |
Error("Manifest export failed.\nAborting..."); |
| 454 |
|
✗ |
return false; |
| 455 |
|
|
} |
| 456 |
|
✗ |
has_committed_new_revision_ = true; |
| 457 |
|
✗ |
} else { |
| 458 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, |
| 459 |
|
|
"\nNo catalogs migrated, skipping the commit..."); |
| 460 |
|
|
} |
| 461 |
|
|
|
| 462 |
|
|
// Get rid of the open root catalog |
| 463 |
|
✗ |
delete root_catalog; |
| 464 |
|
|
|
| 465 |
|
✗ |
return true; |
| 466 |
|
|
} |
| 467 |
|
|
|
| 468 |
|
|
|
| 469 |
|
✗ |
void CommandMigrate::CatalogCallback( |
| 470 |
|
|
const CatalogTraversalData<catalog::WritableCatalog> &data) { |
| 471 |
|
✗ |
std::string tree_indent; |
| 472 |
|
✗ |
std::string hash_string; |
| 473 |
|
✗ |
std::string path; |
| 474 |
|
|
|
| 475 |
|
✗ |
for (unsigned int i = 1; i < data.tree_level; ++i) { |
| 476 |
|
✗ |
tree_indent += "\u2502 "; |
| 477 |
|
|
} |
| 478 |
|
|
|
| 479 |
|
✗ |
if (data.tree_level > 0) { |
| 480 |
|
✗ |
tree_indent += "\u251C\u2500 "; |
| 481 |
|
|
} |
| 482 |
|
|
|
| 483 |
|
✗ |
hash_string = data.catalog_hash.ToString(); |
| 484 |
|
|
|
| 485 |
|
✗ |
path = data.catalog->mountpoint().ToString(); |
| 486 |
|
✗ |
if (path.empty()) { |
| 487 |
|
✗ |
path = "/"; |
| 488 |
|
✗ |
root_catalog_ = data.catalog; |
| 489 |
|
|
} |
| 490 |
|
|
|
| 491 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "%s%s %s", |
| 492 |
|
|
tree_indent.c_str(), |
| 493 |
|
|
hash_string.c_str(), |
| 494 |
|
|
path.c_str()); |
| 495 |
|
|
|
| 496 |
|
✗ |
++catalog_count_; |
| 497 |
|
|
} |
| 498 |
|
|
|
| 499 |
|
|
|
| 500 |
|
✗ |
void CommandMigrate::MigrationCallback(PendingCatalog *const &data) { |
| 501 |
|
|
// Check if the migration of the catalog was successful |
| 502 |
|
✗ |
if (!data->success) { |
| 503 |
|
✗ |
Error("Catalog migration failed! Aborting..."); |
| 504 |
|
✗ |
exit(1); |
| 505 |
|
✗ |
return; |
| 506 |
|
|
} |
| 507 |
|
|
|
| 508 |
|
✗ |
if (!data->HasChanges()) { |
| 509 |
|
✗ |
PrintStatusMessage(data, data->GetOldContentHash(), "preserved"); |
| 510 |
|
✗ |
data->was_updated.Set(false); |
| 511 |
|
✗ |
return; |
| 512 |
|
|
} |
| 513 |
|
|
|
| 514 |
|
✗ |
const string &path = (data->HasNew()) ? data->new_catalog->database_path() |
| 515 |
|
✗ |
: data->old_catalog->database_path(); |
| 516 |
|
|
|
| 517 |
|
|
// Save the processed catalog in the pending map |
| 518 |
|
|
{ |
| 519 |
|
✗ |
LockGuard<PendingCatalogMap> guard(&pending_catalogs_); |
| 520 |
|
✗ |
assert(pending_catalogs_.find(path) == pending_catalogs_.end()); |
| 521 |
|
✗ |
pending_catalogs_[path] = data; |
| 522 |
|
|
} |
| 523 |
|
✗ |
catalog_statistics_list_.Insert(data->statistics); |
| 524 |
|
|
|
| 525 |
|
|
// check the size of the uncompressed catalog file |
| 526 |
|
✗ |
size_t new_catalog_size = GetFileSize(path); |
| 527 |
|
✗ |
if (new_catalog_size <= 0) { |
| 528 |
|
✗ |
Error("Failed to get uncompressed file size of catalog!", data); |
| 529 |
|
✗ |
exit(2); |
| 530 |
|
|
return; |
| 531 |
|
|
} |
| 532 |
|
✗ |
data->new_catalog_size = new_catalog_size; |
| 533 |
|
|
|
| 534 |
|
|
// Schedule the compression and upload of the catalog |
| 535 |
|
✗ |
spooler_->ProcessCatalog(path); |
| 536 |
|
|
} |
| 537 |
|
|
|
| 538 |
|
|
|
| 539 |
|
✗ |
void CommandMigrate::UploadCallback(const upload::SpoolerResult &result) { |
| 540 |
|
✗ |
const string &path = result.local_path; |
| 541 |
|
|
|
| 542 |
|
|
// Check if the upload was successful |
| 543 |
|
✗ |
if (result.return_code != 0) { |
| 544 |
|
✗ |
Error("Failed to upload file " + path + "\nAborting..."); |
| 545 |
|
✗ |
exit(2); |
| 546 |
|
|
return; |
| 547 |
|
|
} |
| 548 |
|
✗ |
assert(result.file_chunks.size() == 0); |
| 549 |
|
|
|
| 550 |
|
|
// Remove the just uploaded file |
| 551 |
|
✗ |
unlink(path.c_str()); |
| 552 |
|
|
|
| 553 |
|
|
// Uploaded nested catalog marker... generate and cache DirectoryEntry for it |
| 554 |
|
✗ |
if (path == nested_catalog_marker_tmp_path_) { |
| 555 |
|
✗ |
CreateNestedCatalogMarkerDirent(result.content_hash); |
| 556 |
|
✗ |
return; |
| 557 |
|
|
} else { |
| 558 |
|
|
// Find the catalog path in the pending catalogs and remove it from the list |
| 559 |
|
|
PendingCatalog *catalog; |
| 560 |
|
|
{ |
| 561 |
|
✗ |
LockGuard<PendingCatalogMap> guard(&pending_catalogs_); |
| 562 |
|
✗ |
PendingCatalogMap::iterator i = pending_catalogs_.find(path); |
| 563 |
|
✗ |
assert(i != pending_catalogs_.end()); |
| 564 |
|
✗ |
catalog = const_cast<PendingCatalog*>(i->second); |
| 565 |
|
✗ |
pending_catalogs_.erase(i); |
| 566 |
|
|
} |
| 567 |
|
|
|
| 568 |
|
✗ |
PrintStatusMessage(catalog, result.content_hash, "migrated and uploaded"); |
| 569 |
|
|
|
| 570 |
|
|
// The catalog is completely processed... fill the content_hash to allow the |
| 571 |
|
|
// processing of parent catalogs (Notified by 'was_updated'-future) |
| 572 |
|
|
// NOTE: From now on, this PendingCatalog structure could be deleted and |
| 573 |
|
|
// should not be used anymore! |
| 574 |
|
✗ |
catalog->new_catalog_hash = result.content_hash; |
| 575 |
|
✗ |
catalog->was_updated.Set(true); |
| 576 |
|
|
} |
| 577 |
|
|
} |
| 578 |
|
|
|
| 579 |
|
|
|
| 580 |
|
✗ |
void CommandMigrate::PrintStatusMessage(const PendingCatalog *catalog, |
| 581 |
|
|
const shash::Any &content_hash, |
| 582 |
|
|
const std::string &message) { |
| 583 |
|
✗ |
atomic_inc32(&catalogs_processed_); |
| 584 |
|
✗ |
const unsigned int processed = (atomic_read32(&catalogs_processed_) * 100) / |
| 585 |
|
✗ |
catalog_count_; |
| 586 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "[%d%%] %s %sC %s", |
| 587 |
|
|
processed, |
| 588 |
|
|
message.c_str(), |
| 589 |
|
|
content_hash.ToString().c_str(), |
| 590 |
|
|
catalog->root_path().c_str()); |
| 591 |
|
|
} |
| 592 |
|
|
|
| 593 |
|
|
|
| 594 |
|
|
template <class MigratorT> |
| 595 |
|
✗ |
void CommandMigrate::ConvertCatalogsRecursively(PendingCatalog *catalog, |
| 596 |
|
|
MigratorT *migrator) { |
| 597 |
|
|
// First migrate all nested catalogs (depth first traversal) |
| 598 |
|
✗ |
const catalog::CatalogList nested_catalogs = |
| 599 |
|
✗ |
catalog->old_catalog->GetChildren(); |
| 600 |
|
✗ |
catalog::CatalogList::const_iterator i = nested_catalogs.begin(); |
| 601 |
|
✗ |
catalog::CatalogList::const_iterator iend = nested_catalogs.end(); |
| 602 |
|
✗ |
catalog->nested_catalogs.reserve(nested_catalogs.size()); |
| 603 |
|
✗ |
for (; i != iend; ++i) { |
| 604 |
|
✗ |
PendingCatalog *new_nested = new PendingCatalog(*i); |
| 605 |
|
✗ |
catalog->nested_catalogs.push_back(new_nested); |
| 606 |
|
✗ |
ConvertCatalogsRecursively(new_nested, migrator); |
| 607 |
|
|
} |
| 608 |
|
|
|
| 609 |
|
|
// Migrate this catalog referencing all its (already migrated) children |
| 610 |
|
✗ |
migrator->Schedule(catalog); |
| 611 |
|
|
} |
| 612 |
|
|
|
| 613 |
|
|
|
| 614 |
|
✗ |
bool CommandMigrate::RaiseFileDescriptorLimit() const { |
| 615 |
|
|
struct rlimit rpl; |
| 616 |
|
✗ |
memset(&rpl, 0, sizeof(rpl)); |
| 617 |
|
✗ |
getrlimit(RLIMIT_NOFILE, &rpl); |
| 618 |
|
✗ |
if (rpl.rlim_cur < file_descriptor_limit_) { |
| 619 |
|
✗ |
if (rpl.rlim_max < file_descriptor_limit_) |
| 620 |
|
✗ |
rpl.rlim_max = file_descriptor_limit_; |
| 621 |
|
✗ |
rpl.rlim_cur = file_descriptor_limit_; |
| 622 |
|
✗ |
const bool retval = setrlimit(RLIMIT_NOFILE, &rpl); |
| 623 |
|
✗ |
if (retval != 0) { |
| 624 |
|
✗ |
return false; |
| 625 |
|
|
} |
| 626 |
|
|
} |
| 627 |
|
✗ |
return true; |
| 628 |
|
|
} |
| 629 |
|
|
|
| 630 |
|
|
|
| 631 |
|
✗ |
bool CommandMigrate::ConfigureSQLite() const { |
| 632 |
|
✗ |
int retval = sqlite3_config(SQLITE_CONFIG_MULTITHREAD); |
| 633 |
|
✗ |
return (retval == SQLITE_OK); |
| 634 |
|
|
} |
| 635 |
|
|
|
| 636 |
|
|
|
| 637 |
|
✗ |
void CommandMigrate::AnalyzeCatalogStatistics() const { |
| 638 |
|
✗ |
const unsigned int number_of_catalogs = catalog_statistics_list_.size(); |
| 639 |
|
✗ |
unsigned int aggregated_entry_count = 0; |
| 640 |
|
✗ |
unsigned int aggregated_max_row_id = 0; |
| 641 |
|
✗ |
unsigned int aggregated_hardlink_count = 0; |
| 642 |
|
✗ |
unsigned int aggregated_linkcounts = 0; |
| 643 |
|
✗ |
double aggregated_migration_time = 0.0; |
| 644 |
|
|
|
| 645 |
|
✗ |
CatalogStatisticsList::const_iterator i = catalog_statistics_list_.begin(); |
| 646 |
|
✗ |
CatalogStatisticsList::const_iterator iend = catalog_statistics_list_.end(); |
| 647 |
|
✗ |
for (; i != iend; ++i) { |
| 648 |
|
✗ |
aggregated_entry_count += i->entry_count; |
| 649 |
|
✗ |
aggregated_max_row_id += i->max_row_id; |
| 650 |
|
✗ |
aggregated_hardlink_count += i->hardlink_group_count; |
| 651 |
|
✗ |
aggregated_linkcounts += i->aggregated_linkcounts; |
| 652 |
|
✗ |
aggregated_migration_time += i->migration_time; |
| 653 |
|
|
} |
| 654 |
|
|
|
| 655 |
|
|
// Inode quantization |
| 656 |
|
✗ |
assert(aggregated_max_row_id > 0); |
| 657 |
|
✗ |
const unsigned int unused_inodes = |
| 658 |
|
|
aggregated_max_row_id - aggregated_entry_count; |
| 659 |
|
✗ |
const float ratio = |
| 660 |
|
✗ |
(static_cast<float>(unused_inodes) / |
| 661 |
|
✗ |
static_cast<float>(aggregated_max_row_id)) * 100.0f; |
| 662 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "Actual Entries: %d\n" |
| 663 |
|
|
"Allocated Inodes: %d\n" |
| 664 |
|
|
" Unused Inodes: %d\n" |
| 665 |
|
|
" Percentage of wasted Inodes: %.1f%%\n", |
| 666 |
|
|
aggregated_entry_count, aggregated_max_row_id, unused_inodes, ratio); |
| 667 |
|
|
|
| 668 |
|
|
// Hardlink statistics |
| 669 |
|
✗ |
const float average_linkcount = (aggregated_hardlink_count > 0) |
| 670 |
|
✗ |
? aggregated_linkcounts / |
| 671 |
|
|
aggregated_hardlink_count |
| 672 |
|
|
: 0.0f; |
| 673 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "Generated Hardlink Groups: %d\n" |
| 674 |
|
|
"Average Linkcount per Group: %.1f\n", |
| 675 |
|
|
aggregated_hardlink_count, average_linkcount); |
| 676 |
|
|
|
| 677 |
|
|
// Performance measures |
| 678 |
|
✗ |
const double average_migration_time = |
| 679 |
|
✗ |
aggregated_migration_time / static_cast<double>(number_of_catalogs); |
| 680 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "Catalog Loading Time: %.2fs\n" |
| 681 |
|
|
"Average Migration Time: %.2fs\n" |
| 682 |
|
|
"Overall Migration Time: %.2fs\n" |
| 683 |
|
|
"Aggregated Migration Time: %.2fs\n", |
| 684 |
|
|
catalog_loading_stopwatch_.GetTime(), |
| 685 |
|
|
average_migration_time, |
| 686 |
|
|
migration_stopwatch_.GetTime(), |
| 687 |
|
|
aggregated_migration_time); |
| 688 |
|
|
} |
| 689 |
|
|
|
| 690 |
|
|
|
| 691 |
|
✗ |
CommandMigrate::PendingCatalog::~PendingCatalog() { |
| 692 |
|
✗ |
delete old_catalog; |
| 693 |
|
✗ |
old_catalog = NULL; |
| 694 |
|
|
|
| 695 |
|
✗ |
if (new_catalog != NULL) { |
| 696 |
|
✗ |
delete new_catalog; |
| 697 |
|
✗ |
new_catalog = NULL; |
| 698 |
|
|
} |
| 699 |
|
|
} |
| 700 |
|
|
|
| 701 |
|
|
|
| 702 |
|
|
template<class DerivedT> |
| 703 |
|
✗ |
CommandMigrate::AbstractMigrationWorker<DerivedT>::AbstractMigrationWorker( |
| 704 |
|
|
const worker_context *context) |
| 705 |
|
✗ |
: temporary_directory_(context->temporary_directory) |
| 706 |
|
✗ |
, collect_catalog_statistics_(context->collect_catalog_statistics) |
| 707 |
|
✗ |
{ } |
| 708 |
|
|
|
| 709 |
|
|
|
| 710 |
|
|
template<class DerivedT> |
| 711 |
|
✗ |
CommandMigrate::AbstractMigrationWorker<DerivedT>::~AbstractMigrationWorker() {} |
| 712 |
|
|
|
| 713 |
|
|
|
| 714 |
|
|
template<class DerivedT> |
| 715 |
|
✗ |
void CommandMigrate::AbstractMigrationWorker<DerivedT>::operator()( |
| 716 |
|
|
const expected_data &data) { |
| 717 |
|
✗ |
migration_stopwatch_.Start(); |
| 718 |
|
✗ |
const bool success = static_cast<DerivedT*>(this)->RunMigration(data) && |
| 719 |
|
✗ |
UpdateNestedCatalogReferences(data) && |
| 720 |
|
✗ |
UpdateCatalogMetadata(data) && |
| 721 |
|
✗ |
CollectAndAggregateStatistics(data) && |
| 722 |
|
✗ |
CleanupNestedCatalogs(data); |
| 723 |
|
✗ |
data->success = success; |
| 724 |
|
✗ |
migration_stopwatch_.Stop(); |
| 725 |
|
|
|
| 726 |
|
✗ |
data->statistics.migration_time = migration_stopwatch_.GetTime(); |
| 727 |
|
✗ |
migration_stopwatch_.Reset(); |
| 728 |
|
|
|
| 729 |
|
|
// Note: MigrationCallback() will take care of the result... |
| 730 |
|
✗ |
if (success) { |
| 731 |
|
✗ |
ConcurrentWorker<DerivedT>::master()->JobSuccessful(data); |
| 732 |
|
|
} else { |
| 733 |
|
✗ |
ConcurrentWorker<DerivedT>::master()->JobFailed(data); |
| 734 |
|
|
} |
| 735 |
|
|
} |
| 736 |
|
|
|
| 737 |
|
|
|
| 738 |
|
|
template<class DerivedT> |
| 739 |
|
✗ |
bool CommandMigrate::AbstractMigrationWorker<DerivedT>:: |
| 740 |
|
|
UpdateNestedCatalogReferences(PendingCatalog *data) const |
| 741 |
|
|
{ |
| 742 |
|
✗ |
const catalog::Catalog *new_catalog = |
| 743 |
|
✗ |
(data->HasNew()) ? data->new_catalog : data->old_catalog; |
| 744 |
|
✗ |
const catalog::CatalogDatabase &writable = new_catalog->database(); |
| 745 |
|
|
|
| 746 |
|
✗ |
catalog::SqlCatalog add_nested_catalog(writable, |
| 747 |
|
|
"INSERT OR REPLACE INTO nested_catalogs (path, sha1, size) " |
| 748 |
|
|
" VALUES (:path, :sha1, :size);"); |
| 749 |
|
|
|
| 750 |
|
|
// go through all nested catalogs and update their references (we are |
| 751 |
|
|
// currently in their parent catalog) |
| 752 |
|
|
// Note: we might need to wait for the nested catalog to be fully processed. |
| 753 |
|
✗ |
PendingCatalogList::const_iterator i = data->nested_catalogs.begin(); |
| 754 |
|
✗ |
PendingCatalogList::const_iterator iend = data->nested_catalogs.end(); |
| 755 |
|
✗ |
for (; i != iend; ++i) { |
| 756 |
|
✗ |
PendingCatalog *nested_catalog = *i; |
| 757 |
|
|
|
| 758 |
|
✗ |
if (!nested_catalog->was_updated.Get()) { |
| 759 |
|
✗ |
continue; |
| 760 |
|
|
} |
| 761 |
|
|
|
| 762 |
|
✗ |
const std::string &root_path = nested_catalog->root_path(); |
| 763 |
|
✗ |
const shash::Any catalog_hash = nested_catalog->new_catalog_hash; |
| 764 |
|
✗ |
const size_t catalog_size = nested_catalog->new_catalog_size; |
| 765 |
|
|
|
| 766 |
|
|
// insert the updated nested catalog reference into the new catalog |
| 767 |
|
✗ |
const bool retval = |
| 768 |
|
✗ |
add_nested_catalog.BindText(1, root_path) && |
| 769 |
|
✗ |
add_nested_catalog.BindText(2, catalog_hash.ToString()) && |
| 770 |
|
✗ |
add_nested_catalog.BindInt64(3, catalog_size) && |
| 771 |
|
✗ |
add_nested_catalog.Execute(); |
| 772 |
|
✗ |
if (!retval) { |
| 773 |
|
✗ |
Error("Failed to add nested catalog link", add_nested_catalog, data); |
| 774 |
|
✗ |
return false; |
| 775 |
|
|
} |
| 776 |
|
✗ |
add_nested_catalog.Reset(); |
| 777 |
|
|
} |
| 778 |
|
|
|
| 779 |
|
✗ |
return true; |
| 780 |
|
|
} |
| 781 |
|
|
|
| 782 |
|
|
|
| 783 |
|
|
template<class DerivedT> |
| 784 |
|
✗ |
bool CommandMigrate::AbstractMigrationWorker<DerivedT>:: |
| 785 |
|
|
UpdateCatalogMetadata(PendingCatalog *data) const |
| 786 |
|
|
{ |
| 787 |
|
✗ |
if (!data->HasChanges()) { |
| 788 |
|
✗ |
return true; |
| 789 |
|
|
} |
| 790 |
|
|
|
| 791 |
|
✗ |
catalog::WritableCatalog *catalog = |
| 792 |
|
✗ |
(data->HasNew()) ? data->new_catalog : GetWritable(data->old_catalog); |
| 793 |
|
|
|
| 794 |
|
|
// Set the previous revision hash in the new catalog to the old catalog |
| 795 |
|
|
// we are doing the whole migration as a new snapshot that does not change |
| 796 |
|
|
// any files, but just applies the necessary data schema migrations |
| 797 |
|
✗ |
catalog->SetPreviousRevision(data->old_catalog->hash()); |
| 798 |
|
✗ |
catalog->IncrementRevision(); |
| 799 |
|
✗ |
catalog->UpdateLastModified(); |
| 800 |
|
|
|
| 801 |
|
✗ |
return true; |
| 802 |
|
|
} |
| 803 |
|
|
|
| 804 |
|
|
|
| 805 |
|
|
template<class DerivedT> |
| 806 |
|
✗ |
bool CommandMigrate::AbstractMigrationWorker<DerivedT>:: |
| 807 |
|
|
CollectAndAggregateStatistics(PendingCatalog *data) const |
| 808 |
|
|
{ |
| 809 |
|
✗ |
if (!collect_catalog_statistics_) { |
| 810 |
|
✗ |
return true; |
| 811 |
|
|
} |
| 812 |
|
|
|
| 813 |
|
✗ |
const catalog::Catalog *new_catalog = |
| 814 |
|
✗ |
(data->HasNew()) ? data->new_catalog : data->old_catalog; |
| 815 |
|
✗ |
const catalog::CatalogDatabase &writable = new_catalog->database(); |
| 816 |
|
|
bool retval; |
| 817 |
|
|
|
| 818 |
|
|
// Find out the discrepancy between MAX(rowid) and COUNT(*) |
| 819 |
|
✗ |
catalog::SqlCatalog wasted_inodes(writable, |
| 820 |
|
|
"SELECT COUNT(*), MAX(rowid) FROM catalog;"); |
| 821 |
|
✗ |
retval = wasted_inodes.FetchRow(); |
| 822 |
|
✗ |
if (!retval) { |
| 823 |
|
✗ |
Error("Failed to count entries in catalog", wasted_inodes, data); |
| 824 |
|
✗ |
return false; |
| 825 |
|
|
} |
| 826 |
|
✗ |
const unsigned int entry_count = wasted_inodes.RetrieveInt64(0); |
| 827 |
|
✗ |
const unsigned int max_row_id = wasted_inodes.RetrieveInt64(1); |
| 828 |
|
|
|
| 829 |
|
|
// Save collected information into the central statistics aggregator |
| 830 |
|
✗ |
data->statistics.root_path = data->root_path(); |
| 831 |
|
✗ |
data->statistics.max_row_id = max_row_id; |
| 832 |
|
✗ |
data->statistics.entry_count = entry_count; |
| 833 |
|
|
|
| 834 |
|
✗ |
return true; |
| 835 |
|
|
} |
| 836 |
|
|
|
| 837 |
|
|
|
| 838 |
|
|
template<class DerivedT> |
| 839 |
|
✗ |
bool CommandMigrate::AbstractMigrationWorker<DerivedT>::CleanupNestedCatalogs( |
| 840 |
|
|
PendingCatalog *data) const |
| 841 |
|
|
{ |
| 842 |
|
|
// All nested catalogs of PendingCatalog 'data' are fully processed and |
| 843 |
|
|
// accounted. It is safe to get rid of their data structures here! |
| 844 |
|
✗ |
PendingCatalogList::const_iterator i = data->nested_catalogs.begin(); |
| 845 |
|
✗ |
PendingCatalogList::const_iterator iend = data->nested_catalogs.end(); |
| 846 |
|
✗ |
for (; i != iend; ++i) { |
| 847 |
|
✗ |
delete *i; |
| 848 |
|
|
} |
| 849 |
|
|
|
| 850 |
|
✗ |
data->nested_catalogs.clear(); |
| 851 |
|
✗ |
return true; |
| 852 |
|
|
} |
| 853 |
|
|
|
| 854 |
|
|
|
| 855 |
|
|
/** |
| 856 |
|
|
* Those values _must_ reflect the schema version in catalog_sql.h so that a |
| 857 |
|
|
* legacy catalog migration generates always the latest catalog revision. |
| 858 |
|
|
* This is a deliberately duplicated piece of information to ensure that always |
| 859 |
|
|
* both the catalog management and migration classes get updated. |
| 860 |
|
|
*/ |
| 861 |
|
|
const float CommandMigrate::MigrationWorker_20x::kSchema = 2.5; |
| 862 |
|
|
const unsigned CommandMigrate::MigrationWorker_20x::kSchemaRevision = 6; |
| 863 |
|
|
|
| 864 |
|
|
|
| 865 |
|
|
template<class DerivedT> |
| 866 |
|
|
catalog::WritableCatalog* |
| 867 |
|
✗ |
CommandMigrate::AbstractMigrationWorker<DerivedT>::GetWritable( |
| 868 |
|
|
const catalog::Catalog *catalog) const { |
| 869 |
|
✗ |
return dynamic_cast<catalog::WritableCatalog*>(const_cast<catalog::Catalog*>( |
| 870 |
|
✗ |
catalog)); |
| 871 |
|
|
} |
| 872 |
|
|
|
| 873 |
|
|
|
| 874 |
|
|
//------------------------------------------------------------------------------ |
| 875 |
|
|
|
| 876 |
|
|
|
| 877 |
|
✗ |
CommandMigrate::MigrationWorker_20x::MigrationWorker_20x( |
| 878 |
|
✗ |
const worker_context *context) |
| 879 |
|
|
: AbstractMigrationWorker<MigrationWorker_20x>(context) |
| 880 |
|
✗ |
, fix_nested_catalog_transitions_(context->fix_nested_catalog_transitions) |
| 881 |
|
✗ |
, analyze_file_linkcounts_(context->analyze_file_linkcounts) |
| 882 |
|
✗ |
, uid_(context->uid) |
| 883 |
|
✗ |
, gid_(context->gid) { } |
| 884 |
|
|
|
| 885 |
|
|
|
| 886 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::RunMigration(PendingCatalog *data) |
| 887 |
|
|
const |
| 888 |
|
|
{ |
| 889 |
|
|
// double-check that we are generating compatible catalogs to the actual |
| 890 |
|
|
// catalog management classes |
| 891 |
|
✗ |
assert(kSchema == catalog::CatalogDatabase::kLatestSupportedSchema); |
| 892 |
|
✗ |
assert(kSchemaRevision == catalog::CatalogDatabase::kLatestSchemaRevision); |
| 893 |
|
|
|
| 894 |
|
✗ |
return CreateNewEmptyCatalog(data) && |
| 895 |
|
✗ |
CheckDatabaseSchemaCompatibility(data) && |
| 896 |
|
✗ |
AttachOldCatalogDatabase(data) && |
| 897 |
|
✗ |
StartDatabaseTransaction(data) && |
| 898 |
|
✗ |
MigrateFileMetadata(data) && |
| 899 |
|
✗ |
MigrateNestedCatalogMountPoints(data) && |
| 900 |
|
✗ |
FixNestedCatalogTransitionPoints(data) && |
| 901 |
|
✗ |
RemoveDanglingNestedMountpoints(data) && |
| 902 |
|
✗ |
GenerateCatalogStatistics(data) && |
| 903 |
|
✗ |
FindRootEntryInformation(data) && |
| 904 |
|
✗ |
CommitDatabaseTransaction(data) && |
| 905 |
|
✗ |
DetachOldCatalogDatabase(data); |
| 906 |
|
|
} |
| 907 |
|
|
|
| 908 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::CreateNewEmptyCatalog( |
| 909 |
|
|
PendingCatalog *data) const |
| 910 |
|
|
{ |
| 911 |
|
✗ |
const string root_path = data->root_path(); |
| 912 |
|
|
|
| 913 |
|
|
// create a new catalog database schema |
| 914 |
|
|
const string clg_db_path = |
| 915 |
|
✗ |
CreateTempPath(temporary_directory_ + "/catalog", 0666); |
| 916 |
|
✗ |
if (clg_db_path.empty()) { |
| 917 |
|
✗ |
Error("Failed to create temporary file for the new catalog database."); |
| 918 |
|
✗ |
return false; |
| 919 |
|
|
} |
| 920 |
|
✗ |
const bool volatile_content = false; |
| 921 |
|
|
|
| 922 |
|
|
{ |
| 923 |
|
|
// TODO(rmeusel): Attach catalog should work with an open catalog database |
| 924 |
|
|
// as well, to remove this inefficiency |
| 925 |
|
|
UniquePtr<catalog::CatalogDatabase> |
| 926 |
|
✗ |
new_clg_db(catalog::CatalogDatabase::Create(clg_db_path)); |
| 927 |
|
✗ |
if (!new_clg_db.IsValid() || |
| 928 |
|
✗ |
!new_clg_db->InsertInitialValues(root_path, volatile_content, "")) { |
| 929 |
|
✗ |
Error("Failed to create database for new catalog"); |
| 930 |
|
✗ |
unlink(clg_db_path.c_str()); |
| 931 |
|
✗ |
return false; |
| 932 |
|
|
} |
| 933 |
|
|
} |
| 934 |
|
|
|
| 935 |
|
|
// Attach the just created nested catalog database |
| 936 |
|
|
catalog::WritableCatalog *writable_catalog = |
| 937 |
|
✗ |
catalog::WritableCatalog::AttachFreely(root_path, clg_db_path, |
| 938 |
|
✗ |
shash::Any(shash::kSha1)); |
| 939 |
|
✗ |
if (writable_catalog == NULL) { |
| 940 |
|
✗ |
Error("Failed to open database for new catalog"); |
| 941 |
|
✗ |
unlink(clg_db_path.c_str()); |
| 942 |
|
✗ |
return false; |
| 943 |
|
|
} |
| 944 |
|
|
|
| 945 |
|
✗ |
data->new_catalog = writable_catalog; |
| 946 |
|
✗ |
return true; |
| 947 |
|
|
} |
| 948 |
|
|
|
| 949 |
|
|
|
| 950 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::CheckDatabaseSchemaCompatibility( |
| 951 |
|
|
PendingCatalog *data) const |
| 952 |
|
|
{ |
| 953 |
|
✗ |
const catalog::CatalogDatabase &old_catalog = data->old_catalog->database(); |
| 954 |
|
✗ |
const catalog::CatalogDatabase &new_catalog = data->new_catalog->database(); |
| 955 |
|
|
|
| 956 |
|
✗ |
if ((new_catalog.schema_version() < |
| 957 |
|
✗ |
catalog::CatalogDatabase::kLatestSupportedSchema - |
| 958 |
|
|
catalog::CatalogDatabase::kSchemaEpsilon |
| 959 |
|
✗ |
|| |
| 960 |
|
✗ |
new_catalog.schema_version() > |
| 961 |
|
✗ |
catalog::CatalogDatabase::kLatestSupportedSchema + |
| 962 |
|
|
catalog::CatalogDatabase::kSchemaEpsilon) |
| 963 |
|
✗ |
|| |
| 964 |
|
✗ |
(old_catalog.schema_version() > 2.1 + |
| 965 |
|
|
catalog::CatalogDatabase::kSchemaEpsilon)) |
| 966 |
|
|
{ |
| 967 |
|
✗ |
Error("Failed to meet database requirements for migration.", data); |
| 968 |
|
✗ |
return false; |
| 969 |
|
|
} |
| 970 |
|
✗ |
return true; |
| 971 |
|
|
} |
| 972 |
|
|
|
| 973 |
|
|
|
| 974 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::AttachOldCatalogDatabase( |
| 975 |
|
|
PendingCatalog *data) const |
| 976 |
|
|
{ |
| 977 |
|
✗ |
const catalog::CatalogDatabase &old_catalog = data->old_catalog->database(); |
| 978 |
|
✗ |
const catalog::CatalogDatabase &new_catalog = data->new_catalog->database(); |
| 979 |
|
|
|
| 980 |
|
|
catalog::SqlCatalog sql_attach_new(new_catalog, |
| 981 |
|
✗ |
"ATTACH '" + old_catalog.filename() + "' AS old;"); |
| 982 |
|
✗ |
bool retval = sql_attach_new.Execute(); |
| 983 |
|
|
|
| 984 |
|
|
// remove the hardlink to the old database file (temporary file), it will not |
| 985 |
|
|
// be needed anymore... data will get deleted when the database is closed |
| 986 |
|
✗ |
unlink(data->old_catalog->database().filename().c_str()); |
| 987 |
|
|
|
| 988 |
|
✗ |
if (!retval) { |
| 989 |
|
✗ |
Error("Failed to attach database of old catalog", sql_attach_new, data); |
| 990 |
|
✗ |
return false; |
| 991 |
|
|
} |
| 992 |
|
✗ |
return true; |
| 993 |
|
|
} |
| 994 |
|
|
|
| 995 |
|
|
|
| 996 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::StartDatabaseTransaction( |
| 997 |
|
|
PendingCatalog *data) const |
| 998 |
|
|
{ |
| 999 |
|
✗ |
assert(data->HasNew()); |
| 1000 |
|
✗ |
data->new_catalog->Transaction(); |
| 1001 |
|
✗ |
return true; |
| 1002 |
|
|
} |
| 1003 |
|
|
|
| 1004 |
|
|
|
| 1005 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::MigrateFileMetadata( |
| 1006 |
|
|
PendingCatalog *data) const |
| 1007 |
|
|
{ |
| 1008 |
|
✗ |
assert(!data->new_catalog->IsDirty()); |
| 1009 |
|
✗ |
assert(data->HasNew()); |
| 1010 |
|
|
bool retval; |
| 1011 |
|
✗ |
const catalog::CatalogDatabase &writable = data->new_catalog->database(); |
| 1012 |
|
|
|
| 1013 |
|
|
// Hardlinks scratch space. |
| 1014 |
|
|
// This temporary table is used for the hardlink analysis results. |
| 1015 |
|
|
// The old catalog format did not have a direct notion of hardlinks and their |
| 1016 |
|
|
// linkcounts, but this information can be partly retrieved from the under- |
| 1017 |
|
|
// lying file system semantics. |
| 1018 |
|
|
// |
| 1019 |
|
|
// Hardlinks: |
| 1020 |
|
|
// groupid : this group id can be used for the new catalog schema |
| 1021 |
|
|
// inode : the inodes that were part of a hardlink group before |
| 1022 |
|
|
// linkcount : the linkcount for hardlink group id members |
| 1023 |
|
|
catalog::SqlCatalog sql_create_hardlinks_table(writable, |
| 1024 |
|
|
"CREATE TEMPORARY TABLE hardlinks " |
| 1025 |
|
|
" ( hardlink_group_id INTEGER PRIMARY KEY AUTOINCREMENT, " |
| 1026 |
|
|
" inode INTEGER, " |
| 1027 |
|
|
" linkcount INTEGER, " |
| 1028 |
|
✗ |
" CONSTRAINT unique_inode UNIQUE (inode) );"); |
| 1029 |
|
✗ |
retval = sql_create_hardlinks_table.Execute(); |
| 1030 |
|
✗ |
if (!retval) { |
| 1031 |
|
✗ |
Error("Failed to create temporary hardlink analysis table", |
| 1032 |
|
|
sql_create_hardlinks_table, data); |
| 1033 |
|
✗ |
return false; |
| 1034 |
|
|
} |
| 1035 |
|
|
|
| 1036 |
|
|
// Directory Linkcount scratch space. |
| 1037 |
|
|
// Directory linkcounts can be obtained from the directory hierarchy reflected |
| 1038 |
|
|
// in the old style catalogs. The new catalog schema asks for this specific |
| 1039 |
|
|
// linkcount. Directory linkcount analysis results will be put into this |
| 1040 |
|
|
// temporary table |
| 1041 |
|
|
catalog::SqlCatalog sql_create_linkcounts_table(writable, |
| 1042 |
|
|
"CREATE TEMPORARY TABLE dir_linkcounts " |
| 1043 |
|
|
" ( inode INTEGER PRIMARY KEY, " |
| 1044 |
|
✗ |
" linkcount INTEGER );"); |
| 1045 |
|
✗ |
retval = sql_create_linkcounts_table.Execute(); |
| 1046 |
|
✗ |
if (!retval) { |
| 1047 |
|
✗ |
Error("Failed to create tmeporary directory linkcount analysis table", |
| 1048 |
|
|
sql_create_linkcounts_table, data); |
| 1049 |
|
|
} |
| 1050 |
|
|
|
| 1051 |
|
|
// It is possible to skip this step. |
| 1052 |
|
|
// In that case all hardlink inodes with a (potential) linkcount > 1 will get |
| 1053 |
|
|
// degraded to files containing the same content |
| 1054 |
|
✗ |
if (analyze_file_linkcounts_) { |
| 1055 |
|
✗ |
retval = AnalyzeFileLinkcounts(data); |
| 1056 |
|
✗ |
if (!retval) { |
| 1057 |
|
✗ |
return false; |
| 1058 |
|
|
} |
| 1059 |
|
|
} |
| 1060 |
|
|
|
| 1061 |
|
|
// Analyze the linkcounts of directories |
| 1062 |
|
|
// - each directory has a linkcount of at least 2 (empty directory) |
| 1063 |
|
|
// (link in parent directory and self reference (cd .) ) |
| 1064 |
|
|
// - for each child directory, the parent's link count is incremented by 1 |
| 1065 |
|
|
// (parent reference in child (cd ..) ) |
| 1066 |
|
|
// |
| 1067 |
|
|
// Note: nested catalog mountpoints will be miscalculated here, since we can't |
| 1068 |
|
|
// check the number of containing directories. They are defined in a the |
| 1069 |
|
|
// linked nested catalog and need to be added later on. |
| 1070 |
|
|
// (see: MigrateNestedCatalogMountPoints() for details) |
| 1071 |
|
|
catalog::SqlCatalog sql_dir_linkcounts(writable, |
| 1072 |
|
|
"INSERT INTO dir_linkcounts " |
| 1073 |
|
|
" SELECT c1.inode as inode, " |
| 1074 |
|
|
" SUM(IFNULL(MIN(c2.inode,1),0)) + 2 as linkcount " |
| 1075 |
|
|
" FROM old.catalog as c1 " |
| 1076 |
|
|
" LEFT JOIN old.catalog as c2 " |
| 1077 |
|
|
" ON c2.parent_1 = c1.md5path_1 AND " |
| 1078 |
|
|
" c2.parent_2 = c1.md5path_2 AND " |
| 1079 |
|
|
" c2.flags & :flag_dir_1 " |
| 1080 |
|
|
" WHERE c1.flags & :flag_dir_2 " |
| 1081 |
|
✗ |
" GROUP BY c1.inode;"); |
| 1082 |
|
✗ |
retval = |
| 1083 |
|
✗ |
sql_dir_linkcounts.BindInt64(1, catalog::SqlDirent::kFlagDir) && |
| 1084 |
|
✗ |
sql_dir_linkcounts.BindInt64(2, catalog::SqlDirent::kFlagDir) && |
| 1085 |
|
✗ |
sql_dir_linkcounts.Execute(); |
| 1086 |
|
✗ |
if (!retval) { |
| 1087 |
|
✗ |
Error("Failed to analyze directory specific linkcounts", |
| 1088 |
|
|
sql_dir_linkcounts, data); |
| 1089 |
|
✗ |
if (sql_dir_linkcounts.GetLastError() == SQLITE_CONSTRAINT) { |
| 1090 |
|
✗ |
Error("Obviously your catalogs are corrupted, since we found a directory" |
| 1091 |
|
|
"inode that is a file inode at the same time!"); |
| 1092 |
|
|
} |
| 1093 |
|
✗ |
return false; |
| 1094 |
|
|
} |
| 1095 |
|
|
|
| 1096 |
|
|
// Copy the old file meta information into the new catalog schema |
| 1097 |
|
|
// here we also add the previously analyzed hardlink/linkcount information |
| 1098 |
|
|
// from both temporary tables "hardlinks" and "dir_linkcounts". |
| 1099 |
|
|
// |
| 1100 |
|
|
// Note: nested catalog mountpoints still need to be treated separately |
| 1101 |
|
|
// (see MigrateNestedCatalogMountPoints() for details) |
| 1102 |
|
|
catalog::SqlCatalog migrate_file_meta_data(writable, |
| 1103 |
|
|
"INSERT INTO catalog " |
| 1104 |
|
|
" SELECT md5path_1, md5path_2, " |
| 1105 |
|
|
" parent_1, parent_2, " |
| 1106 |
|
|
" IFNULL(hardlink_group_id, 0) << 32 | " |
| 1107 |
|
|
" COALESCE(hardlinks.linkcount, dir_linkcounts.linkcount, 1) " |
| 1108 |
|
|
" AS hardlinks, " |
| 1109 |
|
|
" hash, size, mode, mtime, " |
| 1110 |
|
|
" flags, name, symlink, " |
| 1111 |
|
|
" :uid, " |
| 1112 |
|
|
" :gid, " |
| 1113 |
|
|
" NULL " // set empty xattr BLOB (default) |
| 1114 |
|
|
" FROM old.catalog " |
| 1115 |
|
|
" LEFT JOIN hardlinks " |
| 1116 |
|
|
" ON catalog.inode = hardlinks.inode " |
| 1117 |
|
|
" LEFT JOIN dir_linkcounts " |
| 1118 |
|
✗ |
" ON catalog.inode = dir_linkcounts.inode;"); |
| 1119 |
|
✗ |
retval = migrate_file_meta_data.BindInt64(1, uid_) && |
| 1120 |
|
✗ |
migrate_file_meta_data.BindInt64(2, gid_) && |
| 1121 |
|
✗ |
migrate_file_meta_data.Execute(); |
| 1122 |
|
✗ |
if (!retval) { |
| 1123 |
|
✗ |
Error("Failed to migrate the file system meta data", |
| 1124 |
|
|
migrate_file_meta_data, data); |
| 1125 |
|
✗ |
return false; |
| 1126 |
|
|
} |
| 1127 |
|
|
|
| 1128 |
|
|
// If we deal with a nested catalog, we need to add a .cvmfscatalog entry |
| 1129 |
|
|
// since it was not present in the old repository specification but is needed |
| 1130 |
|
|
// now! |
| 1131 |
|
✗ |
if (!data->IsRoot()) { |
| 1132 |
|
|
const catalog::DirectoryEntry &nested_marker = |
| 1133 |
|
✗ |
CommandMigrate::GetNestedCatalogMarkerDirent(); |
| 1134 |
|
✗ |
catalog::SqlDirentInsert insert_nested_marker(writable); |
| 1135 |
|
✗ |
const std::string root_path = data->root_path(); |
| 1136 |
|
✗ |
const std::string file_path = root_path + |
| 1137 |
|
✗ |
"/" + nested_marker.name().ToString(); |
| 1138 |
|
|
const shash::Md5 &path_hash = shash::Md5(file_path.data(), |
| 1139 |
|
✗ |
file_path.size()); |
| 1140 |
|
|
const shash::Md5 &parent_hash = shash::Md5(root_path.data(), |
| 1141 |
|
✗ |
root_path.size()); |
| 1142 |
|
✗ |
retval = insert_nested_marker.BindPathHash(path_hash) && |
| 1143 |
|
✗ |
insert_nested_marker.BindParentPathHash(parent_hash) && |
| 1144 |
|
✗ |
insert_nested_marker.BindDirent(nested_marker) && |
| 1145 |
|
✗ |
insert_nested_marker.BindXattrEmpty() && |
| 1146 |
|
✗ |
insert_nested_marker.Execute(); |
| 1147 |
|
✗ |
if (!retval) { |
| 1148 |
|
✗ |
Error("Failed to insert nested catalog marker into new nested catalog.", |
| 1149 |
|
|
insert_nested_marker, data); |
| 1150 |
|
✗ |
return false; |
| 1151 |
|
|
} |
| 1152 |
|
|
} |
| 1153 |
|
|
|
| 1154 |
|
|
// Copy (and update) the properties fields |
| 1155 |
|
|
// |
| 1156 |
|
|
// Note: The 'schema' is explicitly not copied to the new catalog. |
| 1157 |
|
|
// Each catalog contains a revision, which is also copied here and that |
| 1158 |
|
|
// is later updated by calling catalog->IncrementRevision() |
| 1159 |
|
|
catalog::SqlCatalog copy_properties(writable, |
| 1160 |
|
|
"INSERT OR REPLACE INTO properties " |
| 1161 |
|
|
" SELECT key, value " |
| 1162 |
|
|
" FROM old.properties " |
| 1163 |
|
✗ |
" WHERE key != 'schema';"); |
| 1164 |
|
✗ |
retval = copy_properties.Execute(); |
| 1165 |
|
✗ |
if (!retval) { |
| 1166 |
|
✗ |
Error("Failed to migrate the properties table.", copy_properties, data); |
| 1167 |
|
✗ |
return false; |
| 1168 |
|
|
} |
| 1169 |
|
|
|
| 1170 |
|
✗ |
return true; |
| 1171 |
|
|
} |
| 1172 |
|
|
|
| 1173 |
|
|
|
| 1174 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::AnalyzeFileLinkcounts( |
| 1175 |
|
|
PendingCatalog *data) const |
| 1176 |
|
|
{ |
| 1177 |
|
✗ |
assert(data->HasNew()); |
| 1178 |
|
✗ |
const catalog::CatalogDatabase &writable = data->new_catalog->database(); |
| 1179 |
|
|
bool retval; |
| 1180 |
|
|
|
| 1181 |
|
|
// Analyze the hardlink relationships in the old catalog |
| 1182 |
|
|
// inodes used to be assigned at publishing time, implicitly constituating |
| 1183 |
|
|
// those relationships. We now need them explicitly in the file catalogs |
| 1184 |
|
|
// This looks for directory entries with matching inodes but differing path- |
| 1185 |
|
|
// hashes and saves the results in a temporary table called 'hl_scratch' |
| 1186 |
|
|
// |
| 1187 |
|
|
// Note: We only support hardlink groups that reside in the same directory! |
| 1188 |
|
|
// Therefore we first need to figure out hardlink candidates (which |
| 1189 |
|
|
// might still contain hardlink groups spanning more than one directory) |
| 1190 |
|
|
// In a second step these candidates will be analyzed to kick out un- |
| 1191 |
|
|
// supported hardlink groups. |
| 1192 |
|
|
// Unsupported hardlink groups will be be treated as normal files with |
| 1193 |
|
|
// the same content |
| 1194 |
|
|
catalog::SqlCatalog sql_create_hardlinks_scratch_table(writable, |
| 1195 |
|
|
"CREATE TEMPORARY TABLE hl_scratch AS " |
| 1196 |
|
|
" SELECT c1.inode AS inode, c1.md5path_1, c1.md5path_2, " |
| 1197 |
|
|
" c1.parent_1 as c1p1, c1.parent_2 as c1p2, " |
| 1198 |
|
|
" c2.parent_1 as c2p1, c2.parent_2 as c2p2 " |
| 1199 |
|
|
" FROM old.catalog AS c1 " |
| 1200 |
|
|
" INNER JOIN old.catalog AS c2 " |
| 1201 |
|
|
" ON c1.inode == c2.inode AND " |
| 1202 |
|
|
" (c1.md5path_1 != c2.md5path_1 OR " |
| 1203 |
|
✗ |
" c1.md5path_2 != c2.md5path_2);"); |
| 1204 |
|
✗ |
retval = sql_create_hardlinks_scratch_table.Execute(); |
| 1205 |
|
✗ |
if (!retval) { |
| 1206 |
|
✗ |
Error("Failed to create temporary scratch table for hardlink analysis", |
| 1207 |
|
|
sql_create_hardlinks_scratch_table, data); |
| 1208 |
|
✗ |
return false; |
| 1209 |
|
|
} |
| 1210 |
|
|
|
| 1211 |
|
|
// Figures out which hardlink candidates are supported by CVMFS and can be |
| 1212 |
|
|
// transferred into the new catalog as so called hardlink groups. Unsupported |
| 1213 |
|
|
// hardlinks need to be discarded and treated as normal files containing the |
| 1214 |
|
|
// exact same data |
| 1215 |
|
|
catalog::SqlCatalog fill_linkcount_table_for_files(writable, |
| 1216 |
|
|
"INSERT INTO hardlinks (inode, linkcount)" |
| 1217 |
|
|
" SELECT inode, count(*) as linkcount " |
| 1218 |
|
|
" FROM ( " |
| 1219 |
|
|
// recombine supported hardlink inodes with their actual manifested |
| 1220 |
|
|
// hard-links in the catalog. |
| 1221 |
|
|
// Note: for each directory entry pointing to the same supported |
| 1222 |
|
|
// hardlink inode we have a distinct MD5 path hash |
| 1223 |
|
|
" SELECT DISTINCT hl.inode, hl.md5path_1, hl.md5path_2 " |
| 1224 |
|
|
" FROM ( " |
| 1225 |
|
|
// sort out supported hardlink inodes from unsupported ones by |
| 1226 |
|
|
// locality |
| 1227 |
|
|
// Note: see the next comment for the nested SELECT |
| 1228 |
|
|
" SELECT inode " |
| 1229 |
|
|
" FROM ( " |
| 1230 |
|
|
" SELECT inode, count(*) AS cnt " |
| 1231 |
|
|
" FROM ( " |
| 1232 |
|
|
// go through the potential hardlinks and collect location infor- |
| 1233 |
|
|
// mation about them. |
| 1234 |
|
|
// Note: we only support hardlinks that all reside in the same |
| 1235 |
|
|
// directory, thus having the same parent (c1p* == c2p*) |
| 1236 |
|
|
// --> For supported hardlink candidates the SELECT DISTINCT |
| 1237 |
|
|
// will produce only a single row, whereas others produce more |
| 1238 |
|
|
" SELECT DISTINCT inode,c1p1,c1p1,c2p1,c2p2 " |
| 1239 |
|
|
" FROM hl_scratch AS hl " |
| 1240 |
|
|
" ) " |
| 1241 |
|
|
" GROUP BY inode " |
| 1242 |
|
|
" ) " |
| 1243 |
|
|
" WHERE cnt = 1 " |
| 1244 |
|
|
" ) AS supported_hardlinks " |
| 1245 |
|
|
" LEFT JOIN hl_scratch AS hl " |
| 1246 |
|
|
" ON supported_hardlinks.inode = hl.inode " |
| 1247 |
|
|
" ) " |
| 1248 |
|
✗ |
" GROUP BY inode;"); |
| 1249 |
|
✗ |
retval = fill_linkcount_table_for_files.Execute(); |
| 1250 |
|
✗ |
if (!retval) { |
| 1251 |
|
✗ |
Error("Failed to analyze hardlink relationships for files.", |
| 1252 |
|
|
fill_linkcount_table_for_files, data); |
| 1253 |
|
✗ |
return false; |
| 1254 |
|
|
} |
| 1255 |
|
|
|
| 1256 |
|
|
// The file linkcount and hardlink analysis is finished and the scratch table |
| 1257 |
|
|
// can be deleted... |
| 1258 |
|
|
catalog::SqlCatalog drop_hardlink_scratch_space(writable, |
| 1259 |
|
✗ |
"DROP TABLE hl_scratch;"); |
| 1260 |
|
✗ |
retval = drop_hardlink_scratch_space.Execute(); |
| 1261 |
|
✗ |
if (!retval) { |
| 1262 |
|
✗ |
Error("Failed to remove file linkcount analysis scratch table", |
| 1263 |
|
|
drop_hardlink_scratch_space, data); |
| 1264 |
|
✗ |
return false; |
| 1265 |
|
|
} |
| 1266 |
|
|
|
| 1267 |
|
|
// Do some statistics if asked for... |
| 1268 |
|
✗ |
if (collect_catalog_statistics_) { |
| 1269 |
|
|
catalog::SqlCatalog count_hardlinks(writable, |
| 1270 |
|
✗ |
"SELECT count(*), sum(linkcount) FROM hardlinks;"); |
| 1271 |
|
✗ |
retval = count_hardlinks.FetchRow(); |
| 1272 |
|
✗ |
if (!retval) { |
| 1273 |
|
✗ |
Error("Failed to count the generated file hardlinks for statistics", |
| 1274 |
|
|
count_hardlinks, data); |
| 1275 |
|
✗ |
return false; |
| 1276 |
|
|
} |
| 1277 |
|
|
|
| 1278 |
|
✗ |
data->statistics.hardlink_group_count += count_hardlinks.RetrieveInt64(0); |
| 1279 |
|
✗ |
data->statistics.aggregated_linkcounts += count_hardlinks.RetrieveInt64(1); |
| 1280 |
|
|
} |
| 1281 |
|
|
|
| 1282 |
|
✗ |
return true; |
| 1283 |
|
|
} |
| 1284 |
|
|
|
| 1285 |
|
|
|
| 1286 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::MigrateNestedCatalogMountPoints( |
| 1287 |
|
|
PendingCatalog *data) const |
| 1288 |
|
|
{ |
| 1289 |
|
✗ |
assert(data->HasNew()); |
| 1290 |
|
✗ |
const catalog::CatalogDatabase &writable = data->new_catalog->database(); |
| 1291 |
|
|
bool retval; |
| 1292 |
|
|
|
| 1293 |
|
|
// preparing the SQL statement for nested catalog mountpoint update |
| 1294 |
|
|
catalog::SqlCatalog update_mntpnt_linkcount(writable, |
| 1295 |
|
|
"UPDATE catalog " |
| 1296 |
|
|
"SET hardlinks = :linkcount " |
| 1297 |
|
✗ |
"WHERE md5path_1 = :md5_1 AND md5path_2 = :md5_2;"); |
| 1298 |
|
|
|
| 1299 |
|
|
// update all nested catalog mountpoints |
| 1300 |
|
|
// (Note: we might need to wait for the nested catalog to be processed) |
| 1301 |
|
✗ |
PendingCatalogList::const_iterator i = data->nested_catalogs.begin(); |
| 1302 |
|
✗ |
PendingCatalogList::const_iterator iend = data->nested_catalogs.end(); |
| 1303 |
|
✗ |
for (; i != iend; ++i) { |
| 1304 |
|
|
// collect information about the nested catalog |
| 1305 |
|
✗ |
PendingCatalog *nested_catalog = *i; |
| 1306 |
|
✗ |
const catalog::DirectoryEntry root_entry = nested_catalog->root_entry.Get(); |
| 1307 |
|
✗ |
const string &root_path = nested_catalog->root_path(); |
| 1308 |
|
|
|
| 1309 |
|
|
// update the nested catalog mountpoint directory entry with the correct |
| 1310 |
|
|
// linkcount that was determined while processing the nested catalog |
| 1311 |
|
|
const shash::Md5 mountpoint_hash = shash::Md5(root_path.data(), |
| 1312 |
|
✗ |
root_path.size()); |
| 1313 |
|
✗ |
retval = |
| 1314 |
|
✗ |
update_mntpnt_linkcount.BindInt64(1, root_entry.linkcount()) && |
| 1315 |
|
✗ |
update_mntpnt_linkcount.BindMd5(2, 3, mountpoint_hash) && |
| 1316 |
|
✗ |
update_mntpnt_linkcount.Execute(); |
| 1317 |
|
✗ |
if (!retval) { |
| 1318 |
|
✗ |
Error("Failed to update linkcount of nested catalog mountpoint", |
| 1319 |
|
|
update_mntpnt_linkcount, data); |
| 1320 |
|
✗ |
return false; |
| 1321 |
|
|
} |
| 1322 |
|
✗ |
update_mntpnt_linkcount.Reset(); |
| 1323 |
|
|
} |
| 1324 |
|
|
|
| 1325 |
|
✗ |
return true; |
| 1326 |
|
|
} |
| 1327 |
|
|
|
| 1328 |
|
|
|
| 1329 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::FixNestedCatalogTransitionPoints( |
| 1330 |
|
|
PendingCatalog *data) const |
| 1331 |
|
|
{ |
| 1332 |
|
✗ |
assert(data->HasNew()); |
| 1333 |
|
✗ |
if (!fix_nested_catalog_transitions_) { |
| 1334 |
|
|
// Fixing transition point mismatches is not enabled... |
| 1335 |
|
✗ |
return true; |
| 1336 |
|
|
} |
| 1337 |
|
|
|
| 1338 |
|
|
typedef catalog::DirectoryEntry::Difference Difference; |
| 1339 |
|
|
|
| 1340 |
|
✗ |
const catalog::CatalogDatabase &writable = data->new_catalog->database(); |
| 1341 |
|
|
bool retval; |
| 1342 |
|
|
|
| 1343 |
|
✗ |
catalog::SqlLookupPathHash lookup_mountpoint(writable); |
| 1344 |
|
✗ |
catalog::SqlDirentUpdate update_directory_entry(writable); |
| 1345 |
|
|
|
| 1346 |
|
|
// Unbox the nested catalogs (possibly waiting for migration of them first) |
| 1347 |
|
✗ |
PendingCatalogList::const_iterator i = data->nested_catalogs.begin(); |
| 1348 |
|
✗ |
PendingCatalogList::const_iterator iend = data->nested_catalogs.end(); |
| 1349 |
|
✗ |
for (; i != iend; ++i) { |
| 1350 |
|
|
// Collect information about the nested catalog |
| 1351 |
|
✗ |
PendingCatalog *nested_catalog = *i; |
| 1352 |
|
|
const catalog::DirectoryEntry nested_root_entry = |
| 1353 |
|
✗ |
nested_catalog->root_entry.Get(); |
| 1354 |
|
✗ |
const string &nested_root_path = nested_catalog->root_path(); |
| 1355 |
|
|
const shash::Md5 mountpoint_path_hash = |
| 1356 |
|
✗ |
shash::Md5(nested_root_path.data(), nested_root_path.size()); |
| 1357 |
|
|
|
| 1358 |
|
|
// Retrieve the nested catalog mountpoint from the current catalog |
| 1359 |
|
✗ |
retval = lookup_mountpoint.BindPathHash(mountpoint_path_hash) && |
| 1360 |
|
✗ |
lookup_mountpoint.FetchRow(); |
| 1361 |
|
✗ |
if (!retval) { |
| 1362 |
|
✗ |
Error("Failed to fetch nested catalog mountpoint to check for compatible" |
| 1363 |
|
|
"transition points", lookup_mountpoint, data); |
| 1364 |
|
✗ |
return false; |
| 1365 |
|
|
} |
| 1366 |
|
|
|
| 1367 |
|
|
catalog::DirectoryEntry mountpoint_entry = |
| 1368 |
|
✗ |
lookup_mountpoint.GetDirent(data->new_catalog); |
| 1369 |
|
✗ |
lookup_mountpoint.Reset(); |
| 1370 |
|
|
|
| 1371 |
|
|
// Compare nested catalog mountpoint and nested catalog root entries |
| 1372 |
|
|
catalog::DirectoryEntry::Differences diffs = |
| 1373 |
|
✗ |
mountpoint_entry.CompareTo(nested_root_entry); |
| 1374 |
|
|
|
| 1375 |
|
|
// We MUST deal with two directory entries that are a pair of nested cata- |
| 1376 |
|
|
// log mountpoint and root entry! Thus we expect their transition flags to |
| 1377 |
|
|
// differ and their name to be the same. |
| 1378 |
|
✗ |
assert(diffs & Difference::kNestedCatalogTransitionFlags); |
| 1379 |
|
✗ |
assert((diffs & Difference::kName) == 0); |
| 1380 |
|
|
|
| 1381 |
|
|
// Check if there are other differences except the nested catalog transition |
| 1382 |
|
|
// flags and fix them... |
| 1383 |
|
✗ |
if ((diffs ^ Difference::kNestedCatalogTransitionFlags) != 0) { |
| 1384 |
|
|
// If we found differences, we still assume a couple of directory entry |
| 1385 |
|
|
// fields to be the same, otherwise some severe stuff would be wrong... |
| 1386 |
|
✗ |
if ((diffs & Difference::kChecksum) || |
| 1387 |
|
✗ |
(diffs & Difference::kLinkcount) || |
| 1388 |
|
✗ |
(diffs & Difference::kSymlink) || |
| 1389 |
|
✗ |
(diffs & Difference::kChunkedFileFlag) ) |
| 1390 |
|
|
{ |
| 1391 |
|
✗ |
Error("Found an irreparable mismatch in a nested catalog transition " |
| 1392 |
|
✗ |
"point at '" + nested_root_path + "'\nAborting...\n"); |
| 1393 |
|
|
} |
| 1394 |
|
|
|
| 1395 |
|
|
// Copy the properties from the nested catalog root entry into the mount- |
| 1396 |
|
|
// point entry to bring them in sync again |
| 1397 |
|
✗ |
CommandMigrate::FixNestedCatalogTransitionPoint( |
| 1398 |
|
|
nested_root_entry, &mountpoint_entry); |
| 1399 |
|
|
|
| 1400 |
|
|
// save the nested catalog mountpoint entry into the catalog |
| 1401 |
|
✗ |
retval = update_directory_entry.BindPathHash(mountpoint_path_hash) && |
| 1402 |
|
✗ |
update_directory_entry.BindDirent(mountpoint_entry) && |
| 1403 |
|
✗ |
update_directory_entry.Execute(); |
| 1404 |
|
✗ |
if (!retval) { |
| 1405 |
|
✗ |
Error("Failed to save resynchronized nested catalog mountpoint into " |
| 1406 |
|
|
"catalog database", update_directory_entry, data); |
| 1407 |
|
✗ |
return false; |
| 1408 |
|
|
} |
| 1409 |
|
✗ |
update_directory_entry.Reset(); |
| 1410 |
|
|
|
| 1411 |
|
|
// Fixing of this mountpoint went well... inform the user that this minor |
| 1412 |
|
|
// issue occurred |
| 1413 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, |
| 1414 |
|
|
"NOTE: fixed incompatible nested catalog transition point at: " |
| 1415 |
|
|
"'%s' ", nested_root_path.c_str()); |
| 1416 |
|
|
} |
| 1417 |
|
|
} |
| 1418 |
|
|
|
| 1419 |
|
✗ |
return true; |
| 1420 |
|
|
} |
| 1421 |
|
|
|
| 1422 |
|
|
|
| 1423 |
|
✗ |
void CommandMigrate::FixNestedCatalogTransitionPoint( |
| 1424 |
|
|
const catalog::DirectoryEntry &nested_root, |
| 1425 |
|
|
catalog::DirectoryEntry *mountpoint |
| 1426 |
|
|
) { |
| 1427 |
|
|
// Replace some file system parameters in the mountpoint to resync it with |
| 1428 |
|
|
// the nested root of the corresponding nested catalog |
| 1429 |
|
|
// |
| 1430 |
|
|
// Note: this method relies on CommandMigrate being a friend of DirectoryEntry |
| 1431 |
|
✗ |
mountpoint->mode_ = nested_root.mode_; |
| 1432 |
|
✗ |
mountpoint->uid_ = nested_root.uid_; |
| 1433 |
|
✗ |
mountpoint->gid_ = nested_root.gid_; |
| 1434 |
|
✗ |
mountpoint->size_ = nested_root.size_; |
| 1435 |
|
✗ |
mountpoint->mtime_ = nested_root.mtime_; |
| 1436 |
|
|
} |
| 1437 |
|
|
|
| 1438 |
|
|
|
| 1439 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::RemoveDanglingNestedMountpoints( |
| 1440 |
|
|
PendingCatalog *data) const |
| 1441 |
|
|
{ |
| 1442 |
|
✗ |
assert(data->HasNew()); |
| 1443 |
|
✗ |
const catalog::CatalogDatabase &writable = data->new_catalog->database(); |
| 1444 |
|
✗ |
bool retval = false; |
| 1445 |
|
|
|
| 1446 |
|
|
// build a set of registered nested catalog path hashes |
| 1447 |
|
|
typedef catalog::Catalog::NestedCatalogList NestedCatalogList; |
| 1448 |
|
|
typedef std::map<shash::Md5, catalog::Catalog::NestedCatalog> |
| 1449 |
|
|
NestedCatalogMap; |
| 1450 |
|
|
const NestedCatalogList& nested_clgs = |
| 1451 |
|
✗ |
data->old_catalog->ListNestedCatalogs(); |
| 1452 |
|
✗ |
NestedCatalogList::const_iterator i = nested_clgs.begin(); |
| 1453 |
|
✗ |
const NestedCatalogList::const_iterator iend = nested_clgs.end(); |
| 1454 |
|
✗ |
NestedCatalogMap nested_catalog_path_hashes; |
| 1455 |
|
✗ |
for (; i != iend; ++i) { |
| 1456 |
|
✗ |
const PathString &path = i->mountpoint; |
| 1457 |
|
✗ |
const shash::Md5 hash(path.GetChars(), path.GetLength()); |
| 1458 |
|
✗ |
nested_catalog_path_hashes[hash] = *i; |
| 1459 |
|
|
} |
| 1460 |
|
|
|
| 1461 |
|
|
// Retrieve nested catalog mountpoints that have child entries directly inside |
| 1462 |
|
|
// the current catalog (which is a malformed state) |
| 1463 |
|
✗ |
catalog::SqlLookupDanglingMountpoints sql_dangling_mountpoints(writable); |
| 1464 |
|
✗ |
catalog::SqlDirentUpdate save_updated_mountpoint(writable); |
| 1465 |
|
|
|
| 1466 |
|
✗ |
std::vector<catalog::DirectoryEntry> todo_dirent; |
| 1467 |
|
✗ |
std::vector<shash::Md5> todo_hash; |
| 1468 |
|
|
|
| 1469 |
|
|
// go through the list of dangling nested catalog mountpoints and fix them |
| 1470 |
|
|
// where needed (check if there is no nested catalog registered for them) |
| 1471 |
|
✗ |
while (sql_dangling_mountpoints.FetchRow()) { |
| 1472 |
|
|
catalog::DirectoryEntry dangling_mountpoint = |
| 1473 |
|
✗ |
sql_dangling_mountpoints.GetDirent(data->new_catalog); |
| 1474 |
|
✗ |
const shash::Md5 path_hash = sql_dangling_mountpoints.GetPathHash(); |
| 1475 |
|
✗ |
assert(dangling_mountpoint.IsNestedCatalogMountpoint()); |
| 1476 |
|
|
|
| 1477 |
|
|
// check if the nested catalog mountpoint is registered in the nested cata- |
| 1478 |
|
|
// log list of the currently migrated catalog |
| 1479 |
|
|
const NestedCatalogMap::const_iterator nested_catalog = |
| 1480 |
|
✗ |
nested_catalog_path_hashes.find(path_hash); |
| 1481 |
|
✗ |
if (nested_catalog != nested_catalog_path_hashes.end()) { |
| 1482 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStderr, |
| 1483 |
|
|
"WARNING: found a non-empty nested catalog mountpoint under " |
| 1484 |
|
|
"'%s'", nested_catalog->second.mountpoint.c_str()); |
| 1485 |
|
✗ |
continue; |
| 1486 |
|
|
} |
| 1487 |
|
|
|
| 1488 |
|
|
// the mountpoint was confirmed to be dangling and needs to be removed |
| 1489 |
|
✗ |
dangling_mountpoint.set_is_nested_catalog_mountpoint(false); |
| 1490 |
|
✗ |
todo_dirent.push_back(dangling_mountpoint); |
| 1491 |
|
✗ |
todo_hash.push_back(path_hash); |
| 1492 |
|
|
} |
| 1493 |
|
|
|
| 1494 |
|
✗ |
for (unsigned i = 0; i < todo_dirent.size(); ++i) { |
| 1495 |
|
✗ |
retval = save_updated_mountpoint.BindPathHash(todo_hash[i]) && |
| 1496 |
|
✗ |
save_updated_mountpoint.BindDirent(todo_dirent[i]) && |
| 1497 |
|
✗ |
save_updated_mountpoint.Execute() && |
| 1498 |
|
✗ |
save_updated_mountpoint.Reset(); |
| 1499 |
|
✗ |
if (!retval) { |
| 1500 |
|
✗ |
Error("Failed to remove dangling nested catalog mountpoint entry in " |
| 1501 |
|
|
"catalog", save_updated_mountpoint, data); |
| 1502 |
|
✗ |
return false; |
| 1503 |
|
|
} |
| 1504 |
|
|
|
| 1505 |
|
|
// tell the user that this intervention has been taken place |
| 1506 |
|
✗ |
LogCvmfs(kLogCatalog, kLogStdout, "NOTE: fixed dangling nested catalog " |
| 1507 |
|
|
"mountpoint entry called: '%s' ", |
| 1508 |
|
|
todo_dirent[i].name().c_str()); |
| 1509 |
|
|
} |
| 1510 |
|
|
|
| 1511 |
|
✗ |
return true; |
| 1512 |
|
|
} |
| 1513 |
|
|
|
| 1514 |
|
|
|
| 1515 |
|
✗ |
const catalog::DirectoryEntry& CommandMigrate::GetNestedCatalogMarkerDirent() { |
| 1516 |
|
|
// This is pre-initialized singleton... it MUST be already there... |
| 1517 |
|
✗ |
assert(nested_catalog_marker_.name_.ToString() == ".cvmfscatalog"); |
| 1518 |
|
✗ |
return nested_catalog_marker_; |
| 1519 |
|
|
} |
| 1520 |
|
|
|
| 1521 |
|
✗ |
bool CommandMigrate::GenerateNestedCatalogMarkerChunk() { |
| 1522 |
|
|
// Create an empty nested catalog marker file |
| 1523 |
|
|
nested_catalog_marker_tmp_path_ = |
| 1524 |
|
✗ |
CreateTempPath(temporary_directory_ + "/.cvmfscatalog", 0644); |
| 1525 |
|
✗ |
if (nested_catalog_marker_tmp_path_.empty()) { |
| 1526 |
|
✗ |
Error("Failed to create temp file for nested catalog marker dummy."); |
| 1527 |
|
✗ |
return false; |
| 1528 |
|
|
} |
| 1529 |
|
|
|
| 1530 |
|
|
// Process and upload it to the backend storage |
| 1531 |
|
|
IngestionSource *source = |
| 1532 |
|
✗ |
new FileIngestionSource(nested_catalog_marker_tmp_path_); |
| 1533 |
|
✗ |
spooler_->Process(source); |
| 1534 |
|
✗ |
return true; |
| 1535 |
|
|
} |
| 1536 |
|
|
|
| 1537 |
|
✗ |
void CommandMigrate::CreateNestedCatalogMarkerDirent( |
| 1538 |
|
|
const shash::Any &content_hash) |
| 1539 |
|
|
{ |
| 1540 |
|
|
// Generate it only once |
| 1541 |
|
✗ |
assert(nested_catalog_marker_.name_.ToString() != ".cvmfscatalog"); |
| 1542 |
|
|
|
| 1543 |
|
|
// Fill the DirectoryEntry structure will all needed information |
| 1544 |
|
✗ |
nested_catalog_marker_.name_.Assign(".cvmfscatalog", strlen(".cvmfscatalog")); |
| 1545 |
|
✗ |
nested_catalog_marker_.mode_ = 33188; |
| 1546 |
|
✗ |
nested_catalog_marker_.uid_ = uid_; |
| 1547 |
|
✗ |
nested_catalog_marker_.gid_ = gid_; |
| 1548 |
|
✗ |
nested_catalog_marker_.size_ = 0; |
| 1549 |
|
✗ |
nested_catalog_marker_.mtime_ = time(NULL); |
| 1550 |
|
✗ |
nested_catalog_marker_.linkcount_ = 1; |
| 1551 |
|
✗ |
nested_catalog_marker_.checksum_ = content_hash; |
| 1552 |
|
|
} |
| 1553 |
|
|
|
| 1554 |
|
|
|
| 1555 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::GenerateCatalogStatistics( |
| 1556 |
|
|
PendingCatalog *data) const |
| 1557 |
|
|
{ |
| 1558 |
|
✗ |
assert(data->HasNew()); |
| 1559 |
|
✗ |
bool retval = false; |
| 1560 |
|
✗ |
const catalog::CatalogDatabase &writable = data->new_catalog->database(); |
| 1561 |
|
|
|
| 1562 |
|
|
// Aggregated the statistics counters of all nested catalogs |
| 1563 |
|
|
// Note: we might need to wait until nested catalogs are successfully |
| 1564 |
|
|
// processed |
| 1565 |
|
✗ |
catalog::DeltaCounters stats_counters; |
| 1566 |
|
✗ |
PendingCatalogList::const_iterator i = data->nested_catalogs.begin(); |
| 1567 |
|
✗ |
PendingCatalogList::const_iterator iend = data->nested_catalogs.end(); |
| 1568 |
|
✗ |
for (; i != iend; ++i) { |
| 1569 |
|
✗ |
const PendingCatalog *nested_catalog = *i; |
| 1570 |
|
✗ |
const catalog::DeltaCounters &s = nested_catalog->nested_statistics.Get(); |
| 1571 |
|
✗ |
s.PopulateToParent(&stats_counters); |
| 1572 |
|
|
} |
| 1573 |
|
|
|
| 1574 |
|
|
// Count various directory entry types in the catalog to fill up the catalog |
| 1575 |
|
|
// statistics counters introduced in the current catalog schema |
| 1576 |
|
|
catalog::SqlCatalog count_regular_files(writable, |
| 1577 |
|
|
"SELECT count(*) FROM catalog " |
| 1578 |
|
|
" WHERE flags & :flag_file " |
| 1579 |
|
✗ |
" AND NOT flags & :flag_link;"); |
| 1580 |
|
|
catalog::SqlCatalog count_symlinks(writable, |
| 1581 |
|
✗ |
"SELECT count(*) FROM catalog WHERE flags & :flag_link;"); |
| 1582 |
|
|
catalog::SqlCatalog count_directories(writable, |
| 1583 |
|
✗ |
"SELECT count(*) FROM catalog WHERE flags & :flag_dir;"); |
| 1584 |
|
|
catalog::SqlCatalog aggregate_file_size(writable, |
| 1585 |
|
|
"SELECT sum(size) FROM catalog WHERE flags & :flag_file " |
| 1586 |
|
✗ |
" AND NOT flags & :flag_link"); |
| 1587 |
|
|
|
| 1588 |
|
|
// Run the actual counting queries |
| 1589 |
|
✗ |
retval = |
| 1590 |
|
✗ |
count_regular_files.BindInt64(1, catalog::SqlDirent::kFlagFile) && |
| 1591 |
|
✗ |
count_regular_files.BindInt64(2, catalog::SqlDirent::kFlagLink) && |
| 1592 |
|
✗ |
count_regular_files.FetchRow(); |
| 1593 |
|
✗ |
if (!retval) { |
| 1594 |
|
✗ |
Error("Failed to count regular files.", count_regular_files, data); |
| 1595 |
|
✗ |
return false; |
| 1596 |
|
|
} |
| 1597 |
|
✗ |
retval = |
| 1598 |
|
✗ |
count_symlinks.BindInt64(1, catalog::SqlDirent::kFlagLink) && |
| 1599 |
|
✗ |
count_symlinks.FetchRow(); |
| 1600 |
|
✗ |
if (!retval) { |
| 1601 |
|
✗ |
Error("Failed to count symlinks.", count_symlinks, data); |
| 1602 |
|
✗ |
return false; |
| 1603 |
|
|
} |
| 1604 |
|
✗ |
retval = |
| 1605 |
|
✗ |
count_directories.BindInt64(1, catalog::SqlDirent::kFlagDir) && |
| 1606 |
|
✗ |
count_directories.FetchRow(); |
| 1607 |
|
✗ |
if (!retval) { |
| 1608 |
|
✗ |
Error("Failed to count directories.", count_directories, data); |
| 1609 |
|
✗ |
return false; |
| 1610 |
|
|
} |
| 1611 |
|
✗ |
retval = |
| 1612 |
|
✗ |
aggregate_file_size.BindInt64(1, catalog::SqlDirent::kFlagFile) && |
| 1613 |
|
✗ |
aggregate_file_size.BindInt64(2, catalog::SqlDirent::kFlagLink) && |
| 1614 |
|
✗ |
aggregate_file_size.FetchRow(); |
| 1615 |
|
✗ |
if (!retval) { |
| 1616 |
|
✗ |
Error("Failed to aggregate the file sizes.", aggregate_file_size, data); |
| 1617 |
|
✗ |
return false; |
| 1618 |
|
|
} |
| 1619 |
|
|
|
| 1620 |
|
|
// Insert the counted statistics into the DeltaCounters data structure |
| 1621 |
|
✗ |
stats_counters.self.regular_files = count_regular_files.RetrieveInt64(0); |
| 1622 |
|
✗ |
stats_counters.self.symlinks = count_symlinks.RetrieveInt64(0); |
| 1623 |
|
✗ |
stats_counters.self.directories = count_directories.RetrieveInt64(0); |
| 1624 |
|
✗ |
stats_counters.self.nested_catalogs = data->nested_catalogs.size(); |
| 1625 |
|
✗ |
stats_counters.self.file_size = aggregate_file_size.RetrieveInt64(0); |
| 1626 |
|
|
|
| 1627 |
|
|
// Write back the generated statistics counters into the catalog database |
| 1628 |
|
✗ |
stats_counters.WriteToDatabase(writable); |
| 1629 |
|
|
|
| 1630 |
|
|
// Push the generated statistics counters up to the parent catalog |
| 1631 |
|
✗ |
data->nested_statistics.Set(stats_counters); |
| 1632 |
|
|
|
| 1633 |
|
✗ |
return true; |
| 1634 |
|
|
} |
| 1635 |
|
|
|
| 1636 |
|
|
|
| 1637 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::FindRootEntryInformation( |
| 1638 |
|
|
PendingCatalog *data) const |
| 1639 |
|
|
{ |
| 1640 |
|
✗ |
const catalog::CatalogDatabase &writable = data->new_catalog->database(); |
| 1641 |
|
|
bool retval; |
| 1642 |
|
|
|
| 1643 |
|
✗ |
std::string root_path = data->root_path(); |
| 1644 |
|
✗ |
shash::Md5 root_path_hash = shash::Md5(root_path.data(), root_path.size()); |
| 1645 |
|
|
|
| 1646 |
|
✗ |
catalog::SqlLookupPathHash lookup_root_entry(writable); |
| 1647 |
|
✗ |
retval = lookup_root_entry.BindPathHash(root_path_hash) && |
| 1648 |
|
✗ |
lookup_root_entry.FetchRow(); |
| 1649 |
|
✗ |
if (!retval) { |
| 1650 |
|
✗ |
Error("Failed to retrieve root directory entry of migrated catalog", |
| 1651 |
|
|
lookup_root_entry, data); |
| 1652 |
|
✗ |
return false; |
| 1653 |
|
|
} |
| 1654 |
|
|
|
| 1655 |
|
|
catalog::DirectoryEntry entry = |
| 1656 |
|
✗ |
lookup_root_entry.GetDirent(data->new_catalog); |
| 1657 |
|
✗ |
if (entry.linkcount() < 2 || entry.hardlink_group() > 0) { |
| 1658 |
|
✗ |
Error("Retrieved linkcount of catalog root entry is not sane.", data); |
| 1659 |
|
✗ |
return false; |
| 1660 |
|
|
} |
| 1661 |
|
|
|
| 1662 |
|
✗ |
data->root_entry.Set(entry); |
| 1663 |
|
✗ |
return true; |
| 1664 |
|
|
} |
| 1665 |
|
|
|
| 1666 |
|
|
|
| 1667 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::CommitDatabaseTransaction( |
| 1668 |
|
|
PendingCatalog *data) const |
| 1669 |
|
|
{ |
| 1670 |
|
✗ |
assert(data->HasNew()); |
| 1671 |
|
✗ |
data->new_catalog->Commit(); |
| 1672 |
|
✗ |
return true; |
| 1673 |
|
|
} |
| 1674 |
|
|
|
| 1675 |
|
|
|
| 1676 |
|
✗ |
bool CommandMigrate::MigrationWorker_20x::DetachOldCatalogDatabase( |
| 1677 |
|
|
PendingCatalog *data) const |
| 1678 |
|
|
{ |
| 1679 |
|
✗ |
assert(data->HasNew()); |
| 1680 |
|
✗ |
const catalog::CatalogDatabase &writable = data->new_catalog->database(); |
| 1681 |
|
✗ |
catalog::SqlCatalog detach_old_catalog(writable, "DETACH old;"); |
| 1682 |
|
✗ |
const bool retval = detach_old_catalog.Execute(); |
| 1683 |
|
✗ |
if (!retval) { |
| 1684 |
|
✗ |
Error("Failed to detach old catalog database.", detach_old_catalog, data); |
| 1685 |
|
✗ |
return false; |
| 1686 |
|
|
} |
| 1687 |
|
✗ |
return true; |
| 1688 |
|
|
} |
| 1689 |
|
|
|
| 1690 |
|
|
|
| 1691 |
|
|
//------------------------------------------------------------------------------ |
| 1692 |
|
|
|
| 1693 |
|
|
|
| 1694 |
|
✗ |
CommandMigrate::MigrationWorker_217::MigrationWorker_217( |
| 1695 |
|
✗ |
const worker_context *context) |
| 1696 |
|
✗ |
: AbstractMigrationWorker<MigrationWorker_217>(context) |
| 1697 |
|
✗ |
{ } |
| 1698 |
|
|
|
| 1699 |
|
|
|
| 1700 |
|
✗ |
bool CommandMigrate::MigrationWorker_217::RunMigration(PendingCatalog *data) |
| 1701 |
|
|
const |
| 1702 |
|
|
{ |
| 1703 |
|
✗ |
return CheckDatabaseSchemaCompatibility(data) && |
| 1704 |
|
✗ |
StartDatabaseTransaction(data) && |
| 1705 |
|
✗ |
GenerateNewStatisticsCounters(data) && |
| 1706 |
|
✗ |
UpdateCatalogSchema(data) && |
| 1707 |
|
✗ |
CommitDatabaseTransaction(data); |
| 1708 |
|
|
} |
| 1709 |
|
|
|
| 1710 |
|
|
|
| 1711 |
|
✗ |
bool CommandMigrate::MigrationWorker_217::CheckDatabaseSchemaCompatibility( |
| 1712 |
|
|
PendingCatalog *data) const |
| 1713 |
|
|
{ |
| 1714 |
|
✗ |
assert(!data->HasNew()); |
| 1715 |
|
✗ |
const catalog::CatalogDatabase &old_catalog = data->old_catalog->database(); |
| 1716 |
|
|
|
| 1717 |
|
✗ |
if ((old_catalog.schema_version() < 2.4 - |
| 1718 |
|
|
catalog::CatalogDatabase::kSchemaEpsilon) |
| 1719 |
|
✗ |
|| |
| 1720 |
|
✗ |
(old_catalog.schema_version() > 2.4 + |
| 1721 |
|
|
catalog::CatalogDatabase::kSchemaEpsilon)) |
| 1722 |
|
|
{ |
| 1723 |
|
✗ |
Error("Given Catalog is not Schema 2.4.", data); |
| 1724 |
|
✗ |
return false; |
| 1725 |
|
|
} |
| 1726 |
|
|
|
| 1727 |
|
✗ |
return true; |
| 1728 |
|
|
} |
| 1729 |
|
|
|
| 1730 |
|
|
|
| 1731 |
|
✗ |
bool CommandMigrate::MigrationWorker_217::StartDatabaseTransaction( |
| 1732 |
|
|
PendingCatalog *data) const |
| 1733 |
|
|
{ |
| 1734 |
|
✗ |
assert(!data->HasNew()); |
| 1735 |
|
✗ |
GetWritable(data->old_catalog)->Transaction(); |
| 1736 |
|
✗ |
return true; |
| 1737 |
|
|
} |
| 1738 |
|
|
|
| 1739 |
|
|
|
| 1740 |
|
✗ |
bool CommandMigrate::MigrationWorker_217::GenerateNewStatisticsCounters |
| 1741 |
|
|
(PendingCatalog *data) const { |
| 1742 |
|
✗ |
assert(!data->HasNew()); |
| 1743 |
|
✗ |
bool retval = false; |
| 1744 |
|
|
const catalog::CatalogDatabase &writable = |
| 1745 |
|
✗ |
GetWritable(data->old_catalog)->database(); |
| 1746 |
|
|
|
| 1747 |
|
|
// Aggregated the statistics counters of all nested catalogs |
| 1748 |
|
|
// Note: we might need to wait until nested catalogs are successfully |
| 1749 |
|
|
// processed |
| 1750 |
|
✗ |
catalog::DeltaCounters stats_counters; |
| 1751 |
|
✗ |
PendingCatalogList::const_iterator i = data->nested_catalogs.begin(); |
| 1752 |
|
✗ |
PendingCatalogList::const_iterator iend = data->nested_catalogs.end(); |
| 1753 |
|
✗ |
for (; i != iend; ++i) { |
| 1754 |
|
✗ |
const PendingCatalog *nested_catalog = *i; |
| 1755 |
|
✗ |
const catalog::DeltaCounters &s = nested_catalog->nested_statistics.Get(); |
| 1756 |
|
✗ |
s.PopulateToParent(&stats_counters); |
| 1757 |
|
|
} |
| 1758 |
|
|
|
| 1759 |
|
|
// Count various directory entry types in the catalog to fill up the catalog |
| 1760 |
|
|
// statistics counters introduced in the current catalog schema |
| 1761 |
|
|
catalog::SqlCatalog count_chunked_files(writable, |
| 1762 |
|
|
"SELECT count(*), sum(size) FROM catalog " |
| 1763 |
|
✗ |
" WHERE flags & :flag_chunked_file;"); |
| 1764 |
|
|
catalog::SqlCatalog count_file_chunks(writable, |
| 1765 |
|
✗ |
"SELECT count(*) FROM chunks;"); |
| 1766 |
|
|
catalog::SqlCatalog aggregate_file_size(writable, |
| 1767 |
|
|
"SELECT sum(size) FROM catalog WHERE flags & :flag_file " |
| 1768 |
|
✗ |
" AND NOT flags & :flag_link;"); |
| 1769 |
|
|
|
| 1770 |
|
|
// Run the actual counting queries |
| 1771 |
|
✗ |
retval = |
| 1772 |
|
✗ |
count_chunked_files.BindInt64(1, catalog::SqlDirent::kFlagFileChunk) && |
| 1773 |
|
✗ |
count_chunked_files.FetchRow(); |
| 1774 |
|
✗ |
if (!retval) { |
| 1775 |
|
✗ |
Error("Failed to count chunked files.", count_chunked_files, data); |
| 1776 |
|
✗ |
return false; |
| 1777 |
|
|
} |
| 1778 |
|
✗ |
retval = count_file_chunks.FetchRow(); |
| 1779 |
|
✗ |
if (!retval) { |
| 1780 |
|
✗ |
Error("Failed to count file chunks", count_file_chunks, data); |
| 1781 |
|
✗ |
return false; |
| 1782 |
|
|
} |
| 1783 |
|
✗ |
retval = |
| 1784 |
|
✗ |
aggregate_file_size.BindInt64(1, catalog::SqlDirent::kFlagFile) && |
| 1785 |
|
✗ |
aggregate_file_size.BindInt64(2, catalog::SqlDirent::kFlagLink) && |
| 1786 |
|
✗ |
aggregate_file_size.FetchRow(); |
| 1787 |
|
✗ |
if (!retval) { |
| 1788 |
|
✗ |
Error("Failed to aggregate the file sizes.", aggregate_file_size, data); |
| 1789 |
|
✗ |
return false; |
| 1790 |
|
|
} |
| 1791 |
|
|
|
| 1792 |
|
|
// Insert the counted statistics into the DeltaCounters data structure |
| 1793 |
|
✗ |
stats_counters.self.chunked_files = count_chunked_files.RetrieveInt64(0); |
| 1794 |
|
✗ |
stats_counters.self.chunked_file_size = count_chunked_files.RetrieveInt64(1); |
| 1795 |
|
✗ |
stats_counters.self.file_chunks = count_file_chunks.RetrieveInt64(0); |
| 1796 |
|
✗ |
stats_counters.self.file_size = aggregate_file_size.RetrieveInt64(0); |
| 1797 |
|
|
|
| 1798 |
|
|
// Write back the generated statistics counters into the catalog database |
| 1799 |
|
✗ |
catalog::Counters counters; |
| 1800 |
|
✗ |
retval = counters.ReadFromDatabase(writable, catalog::LegacyMode::kLegacy); |
| 1801 |
|
✗ |
if (!retval) { |
| 1802 |
|
✗ |
Error("Failed to read old catalog statistics counters", data); |
| 1803 |
|
✗ |
return false; |
| 1804 |
|
|
} |
| 1805 |
|
✗ |
counters.ApplyDelta(stats_counters); |
| 1806 |
|
✗ |
retval = counters.InsertIntoDatabase(writable); |
| 1807 |
|
✗ |
if (!retval) { |
| 1808 |
|
✗ |
Error("Failed to write new statistics counters to database", data); |
| 1809 |
|
✗ |
return false; |
| 1810 |
|
|
} |
| 1811 |
|
|
|
| 1812 |
|
|
// Push the generated statistics counters up to the parent catalog |
| 1813 |
|
✗ |
data->nested_statistics.Set(stats_counters); |
| 1814 |
|
|
|
| 1815 |
|
✗ |
return true; |
| 1816 |
|
|
} |
| 1817 |
|
|
|
| 1818 |
|
|
|
| 1819 |
|
✗ |
bool CommandMigrate::MigrationWorker_217::UpdateCatalogSchema |
| 1820 |
|
|
(PendingCatalog *data) const { |
| 1821 |
|
✗ |
assert(!data->HasNew()); |
| 1822 |
|
|
const catalog::CatalogDatabase &writable = |
| 1823 |
|
✗ |
GetWritable(data->old_catalog)->database(); |
| 1824 |
|
|
catalog::SqlCatalog update_schema_version(writable, |
| 1825 |
|
✗ |
"UPDATE properties SET value = :schema_version WHERE key = 'schema';"); |
| 1826 |
|
|
|
| 1827 |
|
|
const bool retval = |
| 1828 |
|
✗ |
update_schema_version.BindDouble(1, 2.5) && |
| 1829 |
|
✗ |
update_schema_version.Execute(); |
| 1830 |
|
✗ |
if (!retval) { |
| 1831 |
|
✗ |
Error("Failed to update catalog schema version", |
| 1832 |
|
|
update_schema_version, |
| 1833 |
|
|
data); |
| 1834 |
|
✗ |
return false; |
| 1835 |
|
|
} |
| 1836 |
|
|
|
| 1837 |
|
✗ |
return true; |
| 1838 |
|
|
} |
| 1839 |
|
|
|
| 1840 |
|
|
|
| 1841 |
|
✗ |
bool CommandMigrate::MigrationWorker_217::CommitDatabaseTransaction |
| 1842 |
|
|
(PendingCatalog *data) const { |
| 1843 |
|
✗ |
assert(!data->HasNew()); |
| 1844 |
|
✗ |
GetWritable(data->old_catalog)->Commit(); |
| 1845 |
|
✗ |
return true; |
| 1846 |
|
|
} |
| 1847 |
|
|
|
| 1848 |
|
|
|
| 1849 |
|
|
//------------------------------------------------------------------------------ |
| 1850 |
|
|
|
| 1851 |
|
|
|
| 1852 |
|
✗ |
CommandMigrate::ChownMigrationWorker::ChownMigrationWorker( |
| 1853 |
|
✗ |
const worker_context *context) |
| 1854 |
|
|
: AbstractMigrationWorker<ChownMigrationWorker>(context) |
| 1855 |
|
✗ |
, uid_map_statement_(GenerateMappingStatement(context->uid_map, "uid")) |
| 1856 |
|
✗ |
, gid_map_statement_(GenerateMappingStatement(context->gid_map, "gid")) |
| 1857 |
|
|
{} |
| 1858 |
|
|
|
| 1859 |
|
✗ |
bool CommandMigrate::ChownMigrationWorker::RunMigration( |
| 1860 |
|
|
PendingCatalog *data) const { |
| 1861 |
|
✗ |
return ApplyPersonaMappings(data); |
| 1862 |
|
|
} |
| 1863 |
|
|
|
| 1864 |
|
|
|
| 1865 |
|
✗ |
bool CommandMigrate::ChownMigrationWorker::ApplyPersonaMappings( |
| 1866 |
|
|
PendingCatalog *data) const { |
| 1867 |
|
✗ |
assert(data->old_catalog != NULL); |
| 1868 |
|
✗ |
assert(data->new_catalog == NULL); |
| 1869 |
|
|
|
| 1870 |
|
✗ |
if (data->old_catalog->mountpoint() == |
| 1871 |
|
✗ |
PathString("/" + string(catalog::VirtualCatalog::kVirtualPath))) |
| 1872 |
|
|
{ |
| 1873 |
|
|
// skipping virtual catalog |
| 1874 |
|
✗ |
return true; |
| 1875 |
|
|
} |
| 1876 |
|
|
|
| 1877 |
|
|
const catalog::CatalogDatabase &db = |
| 1878 |
|
✗ |
GetWritable(data->old_catalog)->database(); |
| 1879 |
|
|
|
| 1880 |
|
✗ |
if (!db.BeginTransaction()) { |
| 1881 |
|
✗ |
return false; |
| 1882 |
|
|
} |
| 1883 |
|
|
|
| 1884 |
|
✗ |
catalog::SqlCatalog uid_sql(db, uid_map_statement_); |
| 1885 |
|
✗ |
if (!uid_sql.Execute()) { |
| 1886 |
|
✗ |
Error("Failed to update UIDs", uid_sql, data); |
| 1887 |
|
✗ |
return false; |
| 1888 |
|
|
} |
| 1889 |
|
|
|
| 1890 |
|
✗ |
catalog::SqlCatalog gid_sql(db, gid_map_statement_); |
| 1891 |
|
✗ |
if (!gid_sql.Execute()) { |
| 1892 |
|
✗ |
Error("Failed to update GIDs", gid_sql, data); |
| 1893 |
|
✗ |
return false; |
| 1894 |
|
|
} |
| 1895 |
|
|
|
| 1896 |
|
✗ |
return db.CommitTransaction(); |
| 1897 |
|
|
} |
| 1898 |
|
|
|
| 1899 |
|
|
|
| 1900 |
|
|
template <class MapT> |
| 1901 |
|
✗ |
std::string CommandMigrate::ChownMigrationWorker::GenerateMappingStatement( |
| 1902 |
|
|
const MapT &map, |
| 1903 |
|
|
const std::string &column) const { |
| 1904 |
|
✗ |
assert(map.RuleCount() > 0 || map.HasDefault()); |
| 1905 |
|
|
|
| 1906 |
|
✗ |
std::string stmt = "UPDATE OR ABORT catalog SET " + column + " = "; |
| 1907 |
|
|
|
| 1908 |
|
✗ |
if (map.RuleCount() == 0) { |
| 1909 |
|
|
// map everything to the same value (just a simple UPDATE clause) |
| 1910 |
|
✗ |
stmt += StringifyInt(map.GetDefault()); |
| 1911 |
|
|
} else { |
| 1912 |
|
|
// apply multiple ID mappings (UPDATE clause with CASE statement) |
| 1913 |
|
✗ |
stmt += "CASE " + column + " "; |
| 1914 |
|
|
typedef typename MapT::map_type::const_iterator map_iterator; |
| 1915 |
|
✗ |
map_iterator i = map.GetRuleMap().begin(); |
| 1916 |
|
✗ |
const map_iterator iend = map.GetRuleMap().end(); |
| 1917 |
|
✗ |
for (; i != iend; ++i) { |
| 1918 |
|
✗ |
stmt += "WHEN " + StringifyInt(i->first) + |
| 1919 |
|
✗ |
" THEN " + StringifyInt(i->second) + " "; |
| 1920 |
|
|
} |
| 1921 |
|
|
|
| 1922 |
|
|
// add a default (if provided) or leave unchanged if no mapping fits |
| 1923 |
|
✗ |
stmt += (map.HasDefault()) |
| 1924 |
|
✗ |
? "ELSE " + StringifyInt(map.GetDefault()) + " " |
| 1925 |
|
|
: "ELSE " + column + " "; |
| 1926 |
|
✗ |
stmt += "END"; |
| 1927 |
|
|
} |
| 1928 |
|
|
|
| 1929 |
|
✗ |
stmt += ";"; |
| 1930 |
|
✗ |
return stmt; |
| 1931 |
|
|
} |
| 1932 |
|
|
|
| 1933 |
|
|
|
| 1934 |
|
|
//------------------------------------------------------------------------------ |
| 1935 |
|
|
|
| 1936 |
|
|
|
| 1937 |
|
✗ |
bool CommandMigrate::HardlinkRemovalMigrationWorker::RunMigration( |
| 1938 |
|
|
PendingCatalog *data) const { |
| 1939 |
|
✗ |
return CheckDatabaseSchemaCompatibility(data) && |
| 1940 |
|
✗ |
BreakUpHardlinks(data); |
| 1941 |
|
|
} |
| 1942 |
|
|
|
| 1943 |
|
|
|
| 1944 |
|
|
bool |
| 1945 |
|
✗ |
CommandMigrate::HardlinkRemovalMigrationWorker::CheckDatabaseSchemaCompatibility |
| 1946 |
|
|
(PendingCatalog *data) const { |
| 1947 |
|
✗ |
assert(data->old_catalog != NULL); |
| 1948 |
|
✗ |
assert(data->new_catalog == NULL); |
| 1949 |
|
|
|
| 1950 |
|
✗ |
const catalog::CatalogDatabase &clg = data->old_catalog->database(); |
| 1951 |
|
✗ |
return clg.schema_version() >= 2.4 - catalog::CatalogDatabase::kSchemaEpsilon; |
| 1952 |
|
|
} |
| 1953 |
|
|
|
| 1954 |
|
|
|
| 1955 |
|
✗ |
bool CommandMigrate::HardlinkRemovalMigrationWorker::BreakUpHardlinks( |
| 1956 |
|
|
PendingCatalog *data) const { |
| 1957 |
|
✗ |
assert(data->old_catalog != NULL); |
| 1958 |
|
✗ |
assert(data->new_catalog == NULL); |
| 1959 |
|
|
|
| 1960 |
|
|
const catalog::CatalogDatabase &db = |
| 1961 |
|
✗ |
GetWritable(data->old_catalog)->database(); |
| 1962 |
|
|
|
| 1963 |
|
✗ |
if (!db.BeginTransaction()) { |
| 1964 |
|
✗ |
return false; |
| 1965 |
|
|
} |
| 1966 |
|
|
|
| 1967 |
|
|
// CernVM-FS catalogs do not contain inodes directly but they are assigned by |
| 1968 |
|
|
// the CVMFS catalog at runtime. Hardlinks are treated with so-called hardlink |
| 1969 |
|
|
// group IDs to indicate hardlink relationships that need to be respected at |
| 1970 |
|
|
// runtime by assigning identical inodes accordingly. |
| 1971 |
|
|
// |
| 1972 |
|
|
// This updates all directory entries of a given catalog that have a linkcount |
| 1973 |
|
|
// greater than 1 and are flagged as a 'file'. Note: Symlinks are flagged both |
| 1974 |
|
|
// as 'file' and as 'symlink', hence they are updated implicitly as well. |
| 1975 |
|
|
// |
| 1976 |
|
|
// The 'hardlinks' field in the catalog contains two 32 bit integers: |
| 1977 |
|
|
// * the linkcount in the lower 32 bits |
| 1978 |
|
|
// * the (so called) hardlink group ID in the higher 32 bits |
| 1979 |
|
|
// |
| 1980 |
|
|
// Files that have a linkcount of exactly 1 do not have any hardlinks and have |
| 1981 |
|
|
// the (implicit) hardlink group ID '0'. Hence, 'hardlinks == 1' means that a |
| 1982 |
|
|
// file doesn't have any hardlinks (linkcount = 1) and doesn't need treatment |
| 1983 |
|
|
// here. |
| 1984 |
|
|
// |
| 1985 |
|
|
// Files that have hardlinks (linkcount > 1) will have a very large integer in |
| 1986 |
|
|
// their 'hardlinks' field (hardlink group ID > 0 in higher 32 bits). Those |
| 1987 |
|
|
// files will be treated by setting their 'hardlinks' field to 1, effectively |
| 1988 |
|
|
// clearing all hardlink information from the directory entry. |
| 1989 |
|
|
const std::string stmt = "UPDATE OR ABORT catalog " |
| 1990 |
|
|
"SET hardlinks = 1 " |
| 1991 |
|
|
"WHERE flags & :file_flag " |
| 1992 |
|
✗ |
" AND hardlinks > 1;"; |
| 1993 |
|
✗ |
catalog::SqlCatalog hardlink_removal_sql(db, stmt); |
| 1994 |
|
✗ |
hardlink_removal_sql.BindInt64(1, catalog::SqlDirent::kFlagFile); |
| 1995 |
|
✗ |
hardlink_removal_sql.Execute(); |
| 1996 |
|
|
|
| 1997 |
|
✗ |
return db.CommitTransaction(); |
| 1998 |
|
|
} |
| 1999 |
|
|
|
| 2000 |
|
|
//------------------------------------------------------------------------------ |
| 2001 |
|
|
|
| 2002 |
|
|
|
| 2003 |
|
✗ |
bool CommandMigrate::BulkhashRemovalMigrationWorker::RunMigration( |
| 2004 |
|
|
PendingCatalog *data) const { |
| 2005 |
|
✗ |
return CheckDatabaseSchemaCompatibility(data) && |
| 2006 |
|
✗ |
RemoveRedundantBulkHashes(data); |
| 2007 |
|
|
} |
| 2008 |
|
|
|
| 2009 |
|
|
|
| 2010 |
|
|
bool |
| 2011 |
|
✗ |
CommandMigrate::BulkhashRemovalMigrationWorker::CheckDatabaseSchemaCompatibility |
| 2012 |
|
|
(PendingCatalog *data) const { |
| 2013 |
|
✗ |
assert(data->old_catalog != NULL); |
| 2014 |
|
✗ |
assert(data->new_catalog == NULL); |
| 2015 |
|
|
|
| 2016 |
|
✗ |
const catalog::CatalogDatabase &clg = data->old_catalog->database(); |
| 2017 |
|
✗ |
return clg.schema_version() >= 2.4 - catalog::CatalogDatabase::kSchemaEpsilon; |
| 2018 |
|
|
} |
| 2019 |
|
|
|
| 2020 |
|
|
|
| 2021 |
|
✗ |
bool CommandMigrate::BulkhashRemovalMigrationWorker::RemoveRedundantBulkHashes( |
| 2022 |
|
|
PendingCatalog *data) const { |
| 2023 |
|
✗ |
assert(data->old_catalog != NULL); |
| 2024 |
|
✗ |
assert(data->new_catalog == NULL); |
| 2025 |
|
|
|
| 2026 |
|
|
const catalog::CatalogDatabase &db = |
| 2027 |
|
✗ |
GetWritable(data->old_catalog)->database(); |
| 2028 |
|
|
|
| 2029 |
|
✗ |
if (!db.BeginTransaction()) { |
| 2030 |
|
✗ |
return false; |
| 2031 |
|
|
} |
| 2032 |
|
|
|
| 2033 |
|
|
// Regular files with both bulk hashes and chunked hashes can drop the bulk |
| 2034 |
|
|
// hash since modern clients >= 2.1.7 won't require them |
| 2035 |
|
|
const std::string stmt = "UPDATE OR ABORT catalog " |
| 2036 |
|
|
"SET hash = NULL " |
| 2037 |
|
✗ |
"WHERE flags & :file_chunked_flag;"; |
| 2038 |
|
✗ |
catalog::SqlCatalog bulkhash_removal_sql(db, stmt); |
| 2039 |
|
✗ |
bulkhash_removal_sql.BindInt64(1, catalog::SqlDirent::kFlagFileChunk); |
| 2040 |
|
✗ |
bulkhash_removal_sql.Execute(); |
| 2041 |
|
|
|
| 2042 |
|
✗ |
return db.CommitTransaction(); |
| 2043 |
|
|
} |
| 2044 |
|
|
|
| 2045 |
|
|
|
| 2046 |
|
|
//------------------------------------------------------------------------------ |
| 2047 |
|
|
|
| 2048 |
|
|
|
| 2049 |
|
✗ |
CommandMigrate::StatsMigrationWorker::StatsMigrationWorker( |
| 2050 |
|
✗ |
const worker_context *context) |
| 2051 |
|
✗ |
: AbstractMigrationWorker<StatsMigrationWorker>(context) |
| 2052 |
|
✗ |
{ } |
| 2053 |
|
|
|
| 2054 |
|
|
|
| 2055 |
|
✗ |
bool CommandMigrate::StatsMigrationWorker::RunMigration(PendingCatalog *data) |
| 2056 |
|
|
const |
| 2057 |
|
|
{ |
| 2058 |
|
✗ |
return CheckDatabaseSchemaCompatibility(data) && |
| 2059 |
|
✗ |
StartDatabaseTransaction(data) && |
| 2060 |
|
✗ |
RepairStatisticsCounters(data) && |
| 2061 |
|
✗ |
CommitDatabaseTransaction(data); |
| 2062 |
|
|
} |
| 2063 |
|
|
|
| 2064 |
|
|
|
| 2065 |
|
✗ |
bool CommandMigrate::StatsMigrationWorker::CheckDatabaseSchemaCompatibility( |
| 2066 |
|
|
PendingCatalog *data) const |
| 2067 |
|
|
{ |
| 2068 |
|
✗ |
assert(data->old_catalog != NULL); |
| 2069 |
|
✗ |
assert(data->new_catalog == NULL); |
| 2070 |
|
|
|
| 2071 |
|
✗ |
const catalog::CatalogDatabase &clg = data->old_catalog->database(); |
| 2072 |
|
✗ |
if (clg.schema_version() < 2.5 - catalog::CatalogDatabase::kSchemaEpsilon) { |
| 2073 |
|
✗ |
Error("Given catalog schema is < 2.5.", data); |
| 2074 |
|
✗ |
return false; |
| 2075 |
|
|
} |
| 2076 |
|
|
|
| 2077 |
|
✗ |
if (clg.schema_revision() < 5) { |
| 2078 |
|
✗ |
Error("Given catalog revision is < 5", data); |
| 2079 |
|
✗ |
return false; |
| 2080 |
|
|
} |
| 2081 |
|
|
|
| 2082 |
|
✗ |
return true; |
| 2083 |
|
|
} |
| 2084 |
|
|
|
| 2085 |
|
|
|
| 2086 |
|
✗ |
bool CommandMigrate::StatsMigrationWorker::StartDatabaseTransaction( |
| 2087 |
|
|
PendingCatalog *data) const |
| 2088 |
|
|
{ |
| 2089 |
|
✗ |
assert(!data->HasNew()); |
| 2090 |
|
✗ |
GetWritable(data->old_catalog)->Transaction(); |
| 2091 |
|
✗ |
return true; |
| 2092 |
|
|
} |
| 2093 |
|
|
|
| 2094 |
|
|
|
| 2095 |
|
✗ |
bool CommandMigrate::StatsMigrationWorker::RepairStatisticsCounters( |
| 2096 |
|
|
PendingCatalog *data) const |
| 2097 |
|
|
{ |
| 2098 |
|
✗ |
assert(!data->HasNew()); |
| 2099 |
|
✗ |
bool retval = false; |
| 2100 |
|
|
const catalog::CatalogDatabase &writable = |
| 2101 |
|
✗ |
GetWritable(data->old_catalog)->database(); |
| 2102 |
|
|
|
| 2103 |
|
|
// Aggregated the statistics counters of all nested catalogs |
| 2104 |
|
|
// Note: we might need to wait until nested catalogs are successfully |
| 2105 |
|
|
// processed |
| 2106 |
|
✗ |
catalog::DeltaCounters stats_counters; |
| 2107 |
|
✗ |
PendingCatalogList::const_iterator i = data->nested_catalogs.begin(); |
| 2108 |
|
✗ |
PendingCatalogList::const_iterator iend = data->nested_catalogs.end(); |
| 2109 |
|
✗ |
for (; i != iend; ++i) { |
| 2110 |
|
✗ |
const PendingCatalog *nested_catalog = *i; |
| 2111 |
|
✗ |
const catalog::DeltaCounters &s = nested_catalog->nested_statistics.Get(); |
| 2112 |
|
✗ |
s.PopulateToParent(&stats_counters); |
| 2113 |
|
|
} |
| 2114 |
|
|
|
| 2115 |
|
|
// Count various directory entry types in the catalog to fill up the catalog |
| 2116 |
|
|
// statistics counters introduced in the current catalog schema |
| 2117 |
|
|
catalog::SqlCatalog count_regular(writable, |
| 2118 |
|
✗ |
std::string("SELECT count(*), sum(size) FROM catalog ") + |
| 2119 |
|
✗ |
"WHERE flags & " + StringifyInt(catalog::SqlDirent::kFlagFile) + |
| 2120 |
|
✗ |
" AND NOT flags & " + StringifyInt(catalog::SqlDirent::kFlagLink) + |
| 2121 |
|
✗ |
" AND NOT flags & " + StringifyInt(catalog::SqlDirent::kFlagFileSpecial) + |
| 2122 |
|
✗ |
";"); |
| 2123 |
|
|
catalog::SqlCatalog count_external(writable, |
| 2124 |
|
✗ |
std::string("SELECT count(*), sum(size) FROM catalog ") + |
| 2125 |
|
✗ |
"WHERE flags & " + StringifyInt(catalog::SqlDirent::kFlagFileExternal) + |
| 2126 |
|
✗ |
";"); |
| 2127 |
|
|
catalog::SqlCatalog count_symlink(writable, |
| 2128 |
|
✗ |
std::string("SELECT count(*) FROM catalog ") + |
| 2129 |
|
✗ |
"WHERE flags & " + StringifyInt(catalog::SqlDirent::kFlagLink) + ";"); |
| 2130 |
|
|
catalog::SqlCatalog count_special(writable, |
| 2131 |
|
✗ |
std::string("SELECT count(*) FROM catalog ") + |
| 2132 |
|
✗ |
"WHERE flags & " + StringifyInt(catalog::SqlDirent::kFlagFileSpecial) + |
| 2133 |
|
✗ |
";"); |
| 2134 |
|
|
catalog::SqlCatalog count_xattr(writable, |
| 2135 |
|
✗ |
std::string("SELECT count(*) FROM catalog ") + |
| 2136 |
|
✗ |
"WHERE xattr IS NOT NULL;"); |
| 2137 |
|
|
catalog::SqlCatalog count_chunk(writable, |
| 2138 |
|
✗ |
std::string("SELECT count(*), sum(size) FROM catalog ") + |
| 2139 |
|
✗ |
"WHERE flags & " + StringifyInt(catalog::SqlDirent::kFlagFileChunk) + ";"); |
| 2140 |
|
|
catalog::SqlCatalog count_dir(writable, |
| 2141 |
|
✗ |
std::string("SELECT count(*) FROM catalog ") + |
| 2142 |
|
✗ |
"WHERE flags & " + StringifyInt(catalog::SqlDirent::kFlagDir) + ";"); |
| 2143 |
|
|
catalog::SqlCatalog count_chunk_blobs(writable, |
| 2144 |
|
✗ |
"SELECT count(*) FROM chunks;"); |
| 2145 |
|
|
|
| 2146 |
|
✗ |
retval = count_regular.FetchRow() && |
| 2147 |
|
✗ |
count_external.FetchRow() && |
| 2148 |
|
✗ |
count_symlink.FetchRow() && |
| 2149 |
|
✗ |
count_special.FetchRow() && |
| 2150 |
|
✗ |
count_xattr.FetchRow() && |
| 2151 |
|
✗ |
count_chunk.FetchRow() && |
| 2152 |
|
✗ |
count_dir.FetchRow() && |
| 2153 |
|
✗ |
count_chunk_blobs.FetchRow(); |
| 2154 |
|
✗ |
if (!retval) { |
| 2155 |
|
✗ |
Error("Failed to collect catalog statistics", data); |
| 2156 |
|
✗ |
return false; |
| 2157 |
|
|
} |
| 2158 |
|
|
|
| 2159 |
|
✗ |
stats_counters.self.regular_files = count_regular.RetrieveInt64(0); |
| 2160 |
|
✗ |
stats_counters.self.symlinks = count_symlink.RetrieveInt64(0); |
| 2161 |
|
✗ |
stats_counters.self.specials = count_special.RetrieveInt64(0); |
| 2162 |
|
✗ |
stats_counters.self.directories = count_dir.RetrieveInt64(0); |
| 2163 |
|
✗ |
stats_counters.self.nested_catalogs = data->nested_catalogs.size(); |
| 2164 |
|
✗ |
stats_counters.self.chunked_files = count_chunk.RetrieveInt64(0); |
| 2165 |
|
✗ |
stats_counters.self.file_chunks = count_chunk_blobs.RetrieveInt64(0); |
| 2166 |
|
✗ |
stats_counters.self.file_size = count_regular.RetrieveInt64(1); |
| 2167 |
|
✗ |
stats_counters.self.chunked_file_size = count_chunk.RetrieveInt64(1); |
| 2168 |
|
✗ |
stats_counters.self.xattrs = count_xattr.RetrieveInt64(0); |
| 2169 |
|
✗ |
stats_counters.self.externals = count_external.RetrieveInt64(0); |
| 2170 |
|
✗ |
stats_counters.self.external_file_size = count_external.RetrieveInt64(1); |
| 2171 |
|
|
|
| 2172 |
|
|
// Write back the generated statistics counters into the catalog database |
| 2173 |
|
✗ |
catalog::Counters counters; |
| 2174 |
|
✗ |
counters.ApplyDelta(stats_counters); |
| 2175 |
|
✗ |
retval = counters.InsertIntoDatabase(writable); |
| 2176 |
|
✗ |
if (!retval) { |
| 2177 |
|
✗ |
Error("Failed to write new statistics counters to database", data); |
| 2178 |
|
✗ |
return false; |
| 2179 |
|
|
} |
| 2180 |
|
|
|
| 2181 |
|
|
// Push the generated statistics counters up to the parent catalog |
| 2182 |
|
✗ |
data->nested_statistics.Set(stats_counters); |
| 2183 |
|
|
|
| 2184 |
|
✗ |
return true; |
| 2185 |
|
|
} |
| 2186 |
|
|
|
| 2187 |
|
|
|
| 2188 |
|
✗ |
bool CommandMigrate::StatsMigrationWorker::CommitDatabaseTransaction( |
| 2189 |
|
|
PendingCatalog *data) const |
| 2190 |
|
|
{ |
| 2191 |
|
✗ |
assert(!data->HasNew()); |
| 2192 |
|
✗ |
GetWritable(data->old_catalog)->Commit(); |
| 2193 |
|
✗ |
return true; |
| 2194 |
|
|
} |
| 2195 |
|
|
|
| 2196 |
|
|
} // namespace swissknife |
| 2197 |
|
|
|