virtual void validateFile(IDistributedFile *file)
{
IHThorDiskReadBaseArg *helper = (IHThorDiskReadBaseArg *)queryHelper();
bool codeGenGrouped = 0 != (TDXgrouped & helper->getFlags());
bool isGrouped = fileDesc->isGrouped();
if (isGrouped != codeGenGrouped)
{
Owned<IException> e = MakeActivityWarning(&container, TE_GroupMismatch, "DFS and code generated group info. differs: DFS(%s), CodeGen(%s), using DFS info", isGrouped?"grouped":"ungrouped", codeGenGrouped?"grouped":"ungrouped");
queryJobChannel().fireException(e);
}
IOutputMetaData *recordSize = helper->queryDiskRecordSize()->querySerializedDiskMeta();
if (recordSize->isFixedSize()) // fixed size
{
if (0 != fileDesc->queryProperties().getPropInt("@recordSize"))
{
size32_t rSz = fileDesc->queryProperties().getPropInt("@recordSize");
if (isGrouped)
rSz--; // eog byte not to be included in this test.
if (rSz != recordSize->getMinRecordSize())
throw MakeThorException(TE_RecordSizeMismatch, "Published record size %d for file %s, does not match coded record size %d", rSz, fileName.get(), recordSize->getMinRecordSize());
}
if (!fileDesc->isCompressed() && (TDXcompress & helper->getFlags()))
{
size32_t rSz = recordSize->getMinRecordSize();
if (isGrouped) rSz++;
if (rSz >= MIN_ROWCOMPRESS_RECSIZE)
{
Owned<IException> e = MakeActivityWarning(&container, TE_CompressionMismatch, "Ignoring compression attribute on file '%s', which is not published as compressed in DFS", fileName.get());
queryJobChannel().fireException(e);
}
}
}
if (0 == (TDRnocrccheck & helper->getFlags()))
checkFormatCrc(this, file, helper->getFormatCrc(), false);
}
virtual void init()
{
CMasterActivity::init();
OwnedRoxieString fname(helper->getFileName());
Owned<IDistributedFile> fetchFile = queryThorFileManager().lookup(container.queryJob(), fname, false, 0 != (helper->getFetchFlags() & FFdatafileoptional), true);
if (fetchFile)
{
if (isFileKey(fetchFile))
throw MakeActivityException(this, 0, "Attempting to read index as a flat file: %s", fname.get());
Owned<IFileDescriptor> fileDesc = getConfiguredFileDescriptor(*fetchFile);
void *ekey;
size32_t ekeylen;
helper->getFileEncryptKey(ekeylen,ekey);
bool encrypted = fileDesc->queryProperties().getPropBool("@encrypted");
if (0 != ekeylen)
{
memset(ekey,0,ekeylen);
free(ekey);
if (!encrypted)
{
Owned<IException> e = MakeActivityWarning(&container, TE_EncryptionMismatch, "Ignoring encryption key provided as file '%s' was not published as encrypted", fetchFile->queryLogicalName());
queryJobChannel().fireException(e);
}
}
else if (encrypted)
throw MakeActivityException(this, 0, "File '%s' was published as encrypted but no encryption key provided", fetchFile->queryLogicalName());
mapping.setown(getFileSlaveMaps(fetchFile->queryLogicalName(), *fileDesc, container.queryJob().queryUserDescriptor(), container.queryJob().querySlaveGroup(), container.queryLocalOrGrouped(), false, NULL, fetchFile->querySuperFile()));
mapping->serializeFileOffsetMap(offsetMapMb);
addReadFile(fetchFile);
}
}
void CDiskReadMasterBase::init()
{
IHThorDiskReadBaseArg *helper = (IHThorDiskReadBaseArg *) queryHelper();
fileName.setown(helper->getFileName());
file.setown(queryThorFileManager().lookup(container.queryJob(), fileName, 0 != ((TDXtemporary|TDXjobtemp) & helper->getFlags()), 0 != (TDRoptional & helper->getFlags()), true));
if (file)
{
if (file->numParts() > 1)
fileDesc.setown(getConfiguredFileDescriptor(*file));
else
fileDesc.setown(file->getFileDescriptor());
reInit = 0 != (helper->getFlags() & (TDXvarfilename|TDXdynamicfilename));
if (container.queryLocal() || helper->canMatchAny()) // if local, assume may match
{
bool local;
if (0 == (TDXtemporary & helper->getFlags())) // don't add temp files
{
queryThorFileManager().noteFileRead(container.queryJob(), file);
local = container.queryLocal();
}
else
local = false;
mapping.setown(getFileSlaveMaps(file->queryLogicalName(), *fileDesc, container.queryJob().queryUserDescriptor(), container.queryJob().querySlaveGroup(), local, false, hash, file->querySuperFile()));
}
if (0 != (helper->getFlags() & TDRfilenamecallback)) // only get/serialize if using virtual file name fields
{
IDistributedSuperFile *super = file->querySuperFile();
if (super)
{
unsigned numsubs = super->numSubFiles(true);
unsigned s=0;
for (; s<numsubs; s++)
{
IDistributedFile &subfile = super->querySubFile(s, true);
subfileLogicalFilenames.append(subfile.queryLogicalName());
}
}
}
validateFile(file);
void *ekey;
size32_t ekeylen;
helper->getEncryptKey(ekeylen,ekey);
bool encrypted = fileDesc->queryProperties().getPropBool("@encrypted");
if (0 != ekeylen)
{
memset(ekey,0,ekeylen);
free(ekey);
if (!encrypted)
{
Owned<IException> e = MakeActivityWarning(&container, TE_EncryptionMismatch, "Ignoring encryption key provided as file '%s' was not published as encrypted", fileName.get());
container.queryJob().fireException(e);
}
}
else if (encrypted)
throw MakeActivityException(this, 0, "File '%s' was published as encrypted but no encryption key provided", fileName.get());
}
}
void doReplicate(CActivityBase *activity, IPartDescriptor &partDesc, ICopyFileProgress *iProgress)
{
try
{
_doReplicate(activity, partDesc, iProgress);
}
catch (IException *e)
{
Owned<IThorException> e2 = MakeActivityWarning(activity, e, "doReplicate");
e->Release();
activity->fireException(e2);
}
}
void prepareKey(IDistributedFile *index)
{
IDistributedFile *f = index;
IDistributedSuperFile *super = f->querySuperFile();
unsigned nparts = f->numParts(); // includes tlks if any, but unused in array
performPartLookup.ensure(nparts);
bool checkTLKConsistency = (nullptr != super) && !localKey && (0 != (TIRsorted & indexBaseHelper->getFlags()));
if (nofilter)
{
while (nparts--) performPartLookup.append(true);
if (!checkTLKConsistency) return;
}
else
{
while (nparts--) performPartLookup.append(false); // parts to perform lookup set later
}
Owned<IDistributedFileIterator> iter;
if (super)
{
iter.setown(super->getSubFileIterator(true));
verifyex(iter->first());
f = &iter->query();
}
unsigned width = f->numParts();
if (!localKey)
--width;
assertex(width);
unsigned tlkCrc = 0;
bool first = true;
unsigned superSubIndex=0;
bool fileCrc = false, rowCrc = false;
for (;;)
{
Owned<IDistributedFilePart> part = f->getPart(width);
if (checkTLKConsistency)
{
unsigned _tlkCrc;
if (part->getCrc(_tlkCrc))
fileCrc = true;
else if (part->queryAttributes().hasProp("@crc")) // NB: key "@crc" is not a crc on the file, but data within.
{
_tlkCrc = part->queryAttributes().getPropInt("@crc");
rowCrc = true;
}
else if (part->queryAttributes().hasProp("@tlkCrc")) // backward compat.
{
_tlkCrc = part->queryAttributes().getPropInt("@tlkCrc");
rowCrc = true;
}
else
{
if (rowCrc || fileCrc)
{
checkTLKConsistency = false;
Owned<IException> e = MakeActivityWarning(&container, 0, "Cannot validate that tlks in superfile %s match, some crc attributes are missing", super->queryLogicalName());
queryJobChannel().fireException(e);
}
}
if (rowCrc && fileCrc)
{
checkTLKConsistency = false;
Owned<IException> e = MakeActivityWarning(&container, 0, "Cannot validate that tlks in superfile %s match, due to mixed crc types.", super->queryLogicalName());
queryJobChannel().fireException(e);
}
if (checkTLKConsistency)
{
if (first)
{
tlkCrc = _tlkCrc;
first = false;
}
else if (tlkCrc != _tlkCrc)
throw MakeActivityException(this, 0, "Sorted output on super files comprising of non coparitioned sub keys is not supported (TLK's do not match)");
}
}
if (!nofilter)
{
Owned<IKeyIndex> keyIndex;
unsigned copy;
for (copy=0; copy<part->numCopies(); copy++)
{
RemoteFilename rfn;
OwnedIFile ifile = createIFile(part->getFilename(rfn,copy));
if (ifile->exists())
{
StringBuffer remotePath;
rfn.getRemotePath(remotePath);
unsigned crc = 0;
part->getCrc(crc);
keyIndex.setown(createKeyIndex(remotePath.str(), crc, false, false));
break;
}
}
if (!keyIndex)
throw MakeThorException(TE_FileNotFound, "Top level key part does not exist, for key: %s", index->queryLogicalName());
unsigned fixedSize = indexBaseHelper->queryDiskRecordSize()->querySerializedDiskMeta()->getFixedSize(); // used only if fixed
Owned <IKeyManager> tlk = createLocalKeyManager(keyIndex, fixedSize, NULL);
indexBaseHelper->createSegmentMonitors(tlk);
tlk->finishSegmentMonitors();
tlk->reset();
while (tlk->lookup(false))
{
if (tlk->queryFpos())
performPartLookup.replace(true, (aindex_t)(super?super->numSubFiles(true)*(tlk->queryFpos()-1)+superSubIndex:tlk->queryFpos()-1));
}
}
if (!super||!iter->next())
break;
superSubIndex++;
f = &iter->query();
if (width != f->numParts()-1)
throw MakeActivityException(this, 0, "Super key %s, with mixture of sub key width are not supported.", f->queryLogicalName());
}
}
static void _doReplicate(CActivityBase *activity, IPartDescriptor &partDesc, ICopyFileProgress *iProgress)
{
StringBuffer primaryName;
getPartFilename(partDesc, 0, primaryName);;
RemoteFilename rfn;
IFileDescriptor &fileDesc = partDesc.queryOwner();
unsigned copies = partDesc.numCopies();
unsigned c=1;
for (; c<copies; c++)
{
unsigned replicateCopy;
unsigned clusterNum = partDesc.copyClusterNum(c, &replicateCopy);
rfn.clear();
partDesc.getFilename(c, rfn);
StringBuffer dstName;
rfn.getPath(dstName);
assertex(dstName.length());
if (replicateCopy>0 )
{
try
{
queryThor().queryBackup().backup(dstName.str(), primaryName.str());
}
catch (IException *e)
{
Owned<IThorException> re = MakeActivityWarning(activity, e, "Failed to create replicate file '%s'", dstName.str());
e->Release();
activity->fireException(re);
}
}
else // another primary
{
ActPrintLog(activity, "Copying to primary %s", dstName.str());
StringBuffer tmpName(dstName.str());
tmpName.append(".tmp");
OwnedIFile tmpIFile = createIFile(tmpName.str());
OwnedIFile srcFile = createIFile(primaryName.str());
CFIPScope fipScope(tmpName.str());
try
{
try
{
ensureDirectoryForFile(dstName.str());
::copyFile(tmpIFile, srcFile, 0x100000, iProgress);
}
catch (IException *e)
{
IThorException *re = MakeActivityException(activity, e, "Failed to copy to tmp file '%s' from source file '%s'", tmpIFile->queryFilename(), srcFile->queryFilename());
e->Release();
throw re;
}
try
{
OwnedIFile dstIFile = createIFile(dstName.str());
dstIFile->remove();
tmpIFile->rename(pathTail(dstName.str()));
}
catch (IException *e)
{
IThorException *re = ThorWrapException(e, "Failed to rename '%s' to '%s'", tmpName.str(), dstName.str());
e->Release();
throw re;
}
}
catch (IException *)
{
try { tmpIFile->remove(); }
catch (IException *e) { ActPrintLog(&activity->queryContainer(), e, NULL); e->Release(); }
throw;
}
}
}
}
virtual void init()
{
CMasterActivity::init();
OwnedRoxieString indexFileName(helper->getIndexFileName());
Owned<IDistributedFile> dataFile;
Owned<IDistributedFile> indexFile = queryThorFileManager().lookup(container.queryJob(), indexFileName, false, 0 != (helper->getJoinFlags() & JFindexoptional), true);
unsigned keyReadWidth = (unsigned)container.queryJob().getWorkUnitValueInt("KJKRR", 0);
if (!keyReadWidth || keyReadWidth>container.queryJob().querySlaves())
keyReadWidth = container.queryJob().querySlaves();
initMb.clear();
initMb.append(indexFileName.get());
if (helper->diskAccessRequired())
numTags += 2;
initMb.append(numTags);
unsigned t=0;
for (; t<numTags; t++)
{
tags[t] = container.queryJob().allocateMPTag();
initMb.append(tags[t]);
}
bool keyHasTlk = false;
if (indexFile)
{
unsigned numParts = 0;
localKey = indexFile->queryAttributes().getPropBool("@local");
if (container.queryLocalData() && !localKey)
throw MakeActivityException(this, 0, "Keyed Join cannot be LOCAL unless supplied index is local");
checkFormatCrc(this, indexFile, helper->getIndexFormatCrc(), true);
Owned<IFileDescriptor> indexFileDesc = indexFile->getFileDescriptor();
IDistributedSuperFile *superIndex = indexFile->querySuperFile();
unsigned superIndexWidth = 0;
unsigned numSuperIndexSubs = 0;
if (superIndex)
{
numSuperIndexSubs = superIndex->numSubFiles(true);
bool first=true;
// consistency check
Owned<IDistributedFileIterator> iter = superIndex->getSubFileIterator(true);
ForEach(*iter)
{
IDistributedFile &f = iter->query();
unsigned np = f.numParts()-1;
IDistributedFilePart &part = f.queryPart(np);
const char *kind = part.queryAttributes().queryProp("@kind");
bool hasTlk = NULL != kind && 0 == stricmp("topLevelKey", kind); // if last part not tlk, then deemed local (might be singlePartKey)
if (first)
{
first = false;
keyHasTlk = hasTlk;
superIndexWidth = f.numParts();
if (keyHasTlk)
--superIndexWidth;
}
else
{
if (hasTlk != keyHasTlk)
throw MakeActivityException(this, 0, "Local/Single part keys cannot be mixed with distributed(tlk) keys in keyedjoin");
if (keyHasTlk && superIndexWidth != f.numParts()-1)
throw MakeActivityException(this, 0, "Super sub keys of different width cannot be mixed with distributed(tlk) keys in keyedjoin");
}
}
if (keyHasTlk)
numParts = superIndexWidth * numSuperIndexSubs;
else
numParts = superIndex->numParts();
}
else
{
numParts = indexFile->numParts();
if (numParts)
{
const char *kind = indexFile->queryPart(indexFile->numParts()-1).queryAttributes().queryProp("@kind");
keyHasTlk = NULL != kind && 0 == stricmp("topLevelKey", kind);
if (keyHasTlk)
--numParts;
}
}
if (numParts)
{
initMb.append(numParts);
initMb.append(superIndexWidth); // 0 if not superIndex
initMb.append((superIndex && superIndex->isInterleaved()) ? numSuperIndexSubs : 0);
unsigned p=0;
UnsignedArray parts;
for (; p<numParts; p++)
parts.append(p);
indexFileDesc->serializeParts(initMb, parts);
if (localKey)
keyHasTlk = false; // not used
initMb.append(keyHasTlk);
if (keyHasTlk)
{
if (numSuperIndexSubs)
initMb.append(numSuperIndexSubs);
else
initMb.append((unsigned)1);
Owned<IDistributedFileIterator> iter;
IDistributedFile *f;
if (superIndex)
{
iter.setown(superIndex->getSubFileIterator(true));
f = &iter->query();
}
else
f = indexFile;
loop
{
unsigned location;
OwnedIFile iFile;
StringBuffer filePath;
Owned<IFileDescriptor> fileDesc = f->getFileDescriptor();
Owned<IPartDescriptor> tlkDesc = fileDesc->getPart(fileDesc->numParts()-1);
if (!getBestFilePart(this, *tlkDesc, iFile, location, filePath))
throw MakeThorException(TE_FileNotFound, "Top level key part does not exist, for key: %s", f->queryLogicalName());
OwnedIFileIO iFileIO = iFile->open(IFOread);
assertex(iFileIO);
size32_t tlkSz = (size32_t)iFileIO->size();
initMb.append(tlkSz);
::read(iFileIO, 0, tlkSz, initMb);
if (!iter || !iter->next())
break;
f = &iter->query();
}
}
if (helper->diskAccessRequired())
{
OwnedRoxieString fetchFilename(helper->getFileName());
if (fetchFilename)
{
dataFile.setown(queryThorFileManager().lookup(container.queryJob(), fetchFilename, false, 0 != (helper->getFetchFlags() & FFdatafileoptional), true));
if (dataFile)
{
if (superIndex)
throw MakeActivityException(this, 0, "Superkeys and full keyed joins are not supported");
Owned<IFileDescriptor> dataFileDesc = getConfiguredFileDescriptor(*dataFile);
void *ekey;
size32_t ekeylen;
helper->getFileEncryptKey(ekeylen,ekey);
bool encrypted = dataFileDesc->queryProperties().getPropBool("@encrypted");
if (0 != ekeylen)
{
memset(ekey,0,ekeylen);
free(ekey);
if (!encrypted)
{
Owned<IException> e = MakeActivityWarning(&container, TE_EncryptionMismatch, "Ignoring encryption key provided as file '%s' was not published as encrypted", helper->getFileName());
container.queryJob().fireException(e);
}
}
else if (encrypted)
throw MakeActivityException(this, 0, "File '%s' was published as encrypted but no encryption key provided", fetchFilename.get());
unsigned dataReadWidth = (unsigned)container.queryJob().getWorkUnitValueInt("KJDRR", 0);
if (!dataReadWidth || dataReadWidth>container.queryJob().querySlaves())
dataReadWidth = container.queryJob().querySlaves();
Owned<IGroup> grp = container.queryJob().querySlaveGroup().subset((unsigned)0, dataReadWidth);
dataFileMapping.setown(getFileSlaveMaps(dataFile->queryLogicalName(), *dataFileDesc, container.queryJob().queryUserDescriptor(), *grp, false, false, NULL));
dataFileMapping->serializeFileOffsetMap(offsetMapMb.clear());
}
else
indexFile.clear();
}
}
}
else