virtual void init()
{
HashDistributeMasterBase::init();
// JCSMORE should common up some with indexread
IHThorKeyedDistributeArg *helper = (IHThorKeyedDistributeArg *)queryHelper();
StringBuffer scoped;
OwnedRoxieString indexFileName(helper->getIndexFileName());
queryThorFileManager().addScope(container.queryJob(), indexFileName, scoped);
file.setown(queryThorFileManager().lookup(container.queryJob(), indexFileName));
if (!file)
throw MakeActivityException(this, 0, "KeyedDistribute: Failed to find key: %s", scoped.str());
if (0 == file->numParts())
throw MakeActivityException(this, 0, "KeyedDistribute: Can't distribute based on an empty key: %s", scoped.str());
checkFormatCrc(this, file, helper->getFormatCrc(), true);
Owned<IFileDescriptor> fileDesc = file->getFileDescriptor();
Owned<IPartDescriptor> tlkDesc = fileDesc->getPart(fileDesc->numParts()-1);
if (!tlkDesc->queryProperties().hasProp("@kind") || 0 != stricmp("topLevelKey", tlkDesc->queryProperties().queryProp("@kind")))
throw MakeActivityException(this, 0, "Cannot distribute using a non-distributed key: '%s'", scoped.str());
unsigned location;
OwnedIFile iFile;
StringBuffer filePath;
if (!getBestFilePart(this, *tlkDesc, iFile, location, filePath))
throw MakeThorException(TE_FileNotFound, "Top level key part does not exist, for key: %s", file->queryLogicalName());
OwnedIFileIO iFileIO = iFile->open(IFOread);
assertex(iFileIO);
tlkMb.append(iFileIO->size());
::read(iFileIO, 0, (size32_t)iFileIO->size(), tlkMb);
queryThorFileManager().noteFileRead(container.queryJob(), file);
}
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 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
