/**
* Verify that all THdfsPartitions included in the descriptor table are referenced by at least one
* scan range or part of an inserted table. PrintScanRangeLocations will implicitly verify the
* converse (it'll fail if a scan range references a table/partition descriptor that is not
* present).
*/
private void testHdfsPartitionsReferenced(
TQueryExecRequest execRequest, String query, StringBuilder errorLog) {
long insertTableId = -1;
// Collect all partitions that are referenced by a scan range.
Set<THdfsPartition> scanRangePartitions = Sets.newHashSet();
if (execRequest.per_node_scan_ranges != null) {
for (Map.Entry<Integer, List<TScanRangeLocations>> entry :
execRequest.per_node_scan_ranges.entrySet()) {
if (entry.getValue() == null) {
continue;
}
for (TScanRangeLocations locations : entry.getValue()) {
if (locations.scan_range.isSetHdfs_file_split()) {
THdfsFileSplit split = locations.scan_range.getHdfs_file_split();
THdfsPartition partition = findPartition(entry.getKey(), split);
scanRangePartitions.add(partition);
}
}
}
}
if (execRequest.isSetFinalize_params()) {
insertTableId = execRequest.getFinalize_params().getTable_id();
}
boolean first = true;
// Iterate through all partitions of the descriptor table and verify all partitions
// are referenced.
if (execRequest.isSetDesc_tbl() && execRequest.desc_tbl.isSetTableDescriptors()) {
for (TTableDescriptor tableDesc : execRequest.desc_tbl.tableDescriptors) {
// All partitions of insertTableId are okay.
if (tableDesc.getId() == insertTableId) continue;
if (!tableDesc.isSetHdfsTable()) continue;
THdfsTable hdfsTable = tableDesc.getHdfsTable();
for (Map.Entry<Long, THdfsPartition> e : hdfsTable.getPartitions().entrySet()) {
THdfsPartition partition = e.getValue();
if (!scanRangePartitions.contains(partition)) {
if (first) errorLog.append("query:\n" + query + "\n");
errorLog.append(
" unreferenced partition: HdfsTable: "
+ tableDesc.getId()
+ " HdfsPartition: "
+ partition.getId()
+ "\n");
first = false;
}
}
}
}
}
private THdfsTable getHdfsTable() {
Map<Long, THdfsPartition> idToPartition = Maps.newHashMap();
for (HdfsPartition partition : partitions_) {
idToPartition.put(partition.getId(), partition.toThrift(true));
}
List<String> colNames = new ArrayList<String>();
for (int i = 0; i < colsByPos_.size(); ++i) {
colNames.add(colsByPos_.get(i).getName());
}
THdfsTable hdfsTable =
new THdfsTable(
hdfsBaseDir_, colNames, nullPartitionKeyValue_, nullColumnValue_, idToPartition);
hdfsTable.setAvroSchema(avroSchema_);
return hdfsTable;
}
/**
* Look up the partition corresponding to the plan node (identified by nodeId) and a file split.
*/
private THdfsPartition findPartition(int nodeId, THdfsFileSplit split) {
TPlanNode node = planMap_.get(nodeId);
Preconditions.checkNotNull(node);
Preconditions.checkState(node.node_id == nodeId && node.isSetHdfs_scan_node());
THdfsScanNode scanNode = node.getHdfs_scan_node();
int tupleId = scanNode.getTuple_id();
TTupleDescriptor tupleDesc = tupleMap_.get(tupleId);
Preconditions.checkNotNull(tupleDesc);
Preconditions.checkState(tupleDesc.id == tupleId);
TTableDescriptor tableDesc = tableMap_.get(tupleDesc.tableId);
Preconditions.checkNotNull(tableDesc);
Preconditions.checkState(tableDesc.id == tupleDesc.tableId && tableDesc.isSetHdfsTable());
THdfsTable hdfsTable = tableDesc.getHdfsTable();
THdfsPartition partition = hdfsTable.getPartitions().get(split.partition_id);
Preconditions.checkNotNull(partition);
Preconditions.checkState(partition.id == split.partition_id);
return partition;
}
@Override
protected void loadFromThrift(TTable thriftTable) throws TableLoadingException {
super.loadFromThrift(thriftTable);
THdfsTable hdfsTable = thriftTable.getHdfs_table();
hdfsBaseDir_ = hdfsTable.getHdfsBaseDir();
nullColumnValue_ = hdfsTable.nullColumnValue;
nullPartitionKeyValue_ = hdfsTable.nullPartitionKeyValue;
numHdfsFiles_ = 0;
totalHdfsBytes_ = 0;
for (Map.Entry<Long, THdfsPartition> part : hdfsTable.getPartitions().entrySet()) {
HdfsPartition hdfsPart = HdfsPartition.fromThrift(this, part.getKey(), part.getValue());
numHdfsFiles_ += hdfsPart.getFileDescriptors().size();
totalHdfsBytes_ += hdfsPart.getSize();
partitions_.add(hdfsPart);
}
uniqueHostPortsCount_ = countUniqueDataNetworkLocations(partitions_);
avroSchema_ = hdfsTable.isSetAvroSchema() ? hdfsTable.getAvroSchema() : null;
}
