private DrVertexSet CreateVertexSet(DrGraph graph, DrInputStreamManager inputStream)
{
DrVertexSet result = new DrVertexSet();
List <DrStorageVertex> vertices = inputStream.GetVertices();
for (int i = 0; i < vertices.Count; i++)
{
DrVertex v = vertices[i];
result.Add(v);
}
return(result);
}
private DrInputStreamManager CreateInputNode(DryadLINQApp app, VertexInfo info, string inputName)
{
DrInputStreamManager s;
int err = 0;
DryadLogger.LogInformation("methodEntry: " + inputName);
if (info.ioType == VertexInfo.IOType.PARTITIONEDFILE)
{
DrPartitionInputStream input = new DrPartitionInputStream();
err = input.Open(app.GetUniverse(), new Uri(info.sources[0]).AbsolutePath);
if (!SUCCEEDED(err))
{
string msg = String.Format("Could not read Partitioned file input {0}", info.sources[0]);
throw new LinqToDryadException(msg, err);
}
DrManagerBase inputStage = new DrManagerBase(app.GetGraph(), inputName);
DrInputStreamManager inputManager = new DrInputStreamManager(input, inputStage);
s = inputManager;
}
//else if ( info.ioType == VertexInfo.IOType.STREAM )
//{
// DrDscInputStream input = new DrDscInputStream();
// DryadLogger.LogInformation("Create input node", "Opening DSC input fileset");
// err = input.Open(app.GetUniverse(), info.sources[0]);
// if (!SUCCEEDED(err))
// {
// string msg = String.Format("Could not read DSC input fileset {0}", info.sources[0]);
// throw new LinqToDryadException(msg, err);
// }
// DryadLogger.LogInformation("Create input node", "Opened DSC input fileset");
// DrManagerBase inputStage = new DrManagerBase(app.GetGraph(), inputName);
// DrInputStreamManager inputManager = new DrInputStreamManager(input, inputStage);
// s = inputManager;
//}
else if (info.ioType == VertexInfo.IOType.HDFS_STREAM)
{
DrHdfsInputStream input = new DrHdfsInputStream();
DryadLogger.LogInformation("Create input node", "Opening HDFS input fileset");
Uri srcUri = new Uri(info.sources[0]);
err = input.Open(app.GetUniverse(), srcUri.GetLeftPart(UriPartial.Path), info.recordType);
if (!SUCCEEDED(err))
{
string msg = String.Format("Could not read HDFS input fileset {0}: {1}", info.sources[0], input.GetError());
throw new LinqToDryadException(msg, err);
}
DryadLogger.LogInformation("Create input node", "Opened HDFS input fileset");
DrManagerBase inputStage = new DrManagerBase(app.GetGraph(), inputName);
DrInputStreamManager inputManager = new DrInputStreamManager(input, inputStage);
s = inputManager;
}
else if (info.ioType == VertexInfo.IOType.AZUREBLOB)
{
DrAzureInputStream input = new DrAzureInputStream();
DryadLogger.LogInformation("Create input node", "Opening Azure input fileset");
try
{
input.Open(info.sources[0]);
}
catch (Exception e)
{
string msg = String.Format("Could not read Azure input fileset {0}: {1}", info.sources[0], e.ToString());
throw new LinqToDryadException(msg);
}
DryadLogger.LogInformation("Create input node", "Opened Azure input fileset");
DrManagerBase inputStage = new DrManagerBase(app.GetGraph(), inputName);
DrInputStreamManager inputManager = new DrInputStreamManager(input, inputStage);
s = inputManager;
}
else
{
string msg = String.Format("Unknown input type {0}", info.ioType);
throw new LinqToDryadException(msg);
}
DryadLogger.LogInformation("methodExit");
return(s);
}
private void CreateVertexSet(Vertex v, DryadLINQApp app, Query query, Dictionary <int, GraphStageInfo> graphStageMap)
{
SortedDictionary <int, Vertex> queryPlan = query.queryPlan;
DrVertexSet nodes = null;
DrStageManager newManager = null;
app.GetGraph().GetParameters();
DrGraphParameters parameters = app.GetGraph().GetParameters();
string stdVertexName = "MW";
if (v.type == Vertex.Type.INPUTTABLE)
{
DrInputStreamManager input = CreateInputNode(app, v.info, v.name);
newManager = input.GetStageManager();
nodes = CreateVertexSet(app.GetGraph(), input);
}
else if (v.type == Vertex.Type.OUTPUTTABLE)
{
DrOutputStreamManager output = CreateOutputNode(app, v.info, v.name);
newManager = output.GetStageManager();
nodes = CreateVertexSet(app.GetGraph(), output, v.partitions);
}
else if (v.type == Vertex.Type.CONCAT)
{
newManager = new DrManagerBase(app.GetGraph(), v.name);
// the set of nodes in a concat is just the set of nodes in the predecessor stages concatenated
nodes = new DrVertexSet();
foreach (Predecessor p in v.info.predecessors)
{
GraphStageInfo value = null;
if (graphStageMap.TryGetValue(p.uniqueId, out value))
{
nodes.InsertRange(nodes.Count, value.members);
}
else
{
throw new LinqToDryadException(String.Format("Concat: Failed to find predecessor {0} in graph stage map", p.uniqueId));
}
}
}
else
{
newManager = new DrManagerBase(app.GetGraph(), v.name);
DrVertex vertex;
if (v.type == Vertex.Type.TEE)
{
DrTeeVertex teeVertex = new DrTeeVertex(newManager);
vertex = teeVertex;
}
else
{
DrActiveVertex activeVertex = new DrActiveVertex(newManager, parameters.m_defaultProcessTemplate, parameters.m_defaultVertexTemplate);
activeVertex.AddArgument(stdVertexName);
activeVertex.AddArgument(v.info.assemblyName);
activeVertex.AddArgument(v.info.className);
activeVertex.AddArgument(v.info.methodName);
vertex = activeVertex;
}
nodes = CreateVertexSet(app.GetGraph(), vertex, v.partitions);
if (v.machines != null && v.machines.Length != 0 && v.type != Vertex.Type.TEE)
{
for (int i = 0; i < v.partitions; i++)
{
DrResource r = app.GetUniverse().LookUpResource(v.machines[i]);
if (r != null)
{
DrVertex baseVertex = nodes[i];
DrActiveVertex activeVertex = baseVertex as DrActiveVertex;
activeVertex.GetAffinity().SetHardConstraint(true);
activeVertex.GetAffinity().AddLocality(r);
}
}
}
FileStream mapfile = app.GetIdentityMapFile();
if (mapfile != null)
{
for (int i = 0; i < v.partitions; i++)
{
DrVertex jmv = nodes[i];
string message = String.Format("Mapping {0}[{1}] to {2}\n", v.uniqueId, i, jmv.GetId());
System.Text.ASCIIEncoding encoding = new System.Text.ASCIIEncoding();
mapfile.Seek(0, SeekOrigin.End);
mapfile.Write(encoding.GetBytes(message), 0, message.Length);
}
}
}
graphStageMap.Add(v.uniqueId, new GraphStageInfo(nodes, v, newManager));
}
private DrInputStreamManager CreateInputNode(DryadLINQApp app, VertexInfo info, string inputName)
{
DrInputStreamManager s;
int err = 0;
DryadLogger.LogMethodEntry(inputName);
if (info.ioType == VertexInfo.IOType.PARTITIONEDFILE)
{
DrPartitionInputStream input = new DrPartitionInputStream();
err = input.Open(app.GetUniverse(), info.sources[0]);
if (!SUCCEEDED(err))
{
string msg = String.Format("Could not read DSC input file {0}", info.sources[0]);
throw new LinqToDryadException(msg, err);
}
DrManagerBase inputStage = new DrManagerBase(app.GetGraph(), inputName);
DrInputStreamManager inputManager = new DrInputStreamManager(input, inputStage);
s = inputManager;
}
//else if ( info.ioType == VertexInfo.IOType.STREAM )
//{
// DrDscInputStream input = new DrDscInputStream();
// DryadLogger.LogInformation("Create input node", "Opening DSC input fileset");
// err = input.Open(app.GetUniverse(), info.sources[0]);
// if (!SUCCEEDED(err))
// {
// string msg = String.Format("Could not read DSC input fileset {0}", info.sources[0]);
// throw new LinqToDryadException(msg, err);
// }
// DryadLogger.LogInformation("Create input node", "Opened DSC input fileset");
// DrManagerBase inputStage = new DrManagerBase(app.GetGraph(), inputName);
// DrInputStreamManager inputManager = new DrInputStreamManager(input, inputStage);
// s = inputManager;
//}
else if (info.ioType == VertexInfo.IOType.HDFS_STREAM)
{
DrHdfsInputStream input = new DrHdfsInputStream();
DryadLogger.LogInformation("Create input node", "Opening HDFS input fileset");
err = input.Open(app.GetUniverse(), info.sources[0]);
if (!SUCCEEDED(err))
{
string msg = String.Format("Could not read HDFS input fileset {0}", info.sources[0]);
throw new LinqToDryadException(msg, err);
}
DryadLogger.LogInformation("Create input node", "Opened HDFS input fileset");
DrManagerBase inputStage = new DrManagerBase(app.GetGraph(), inputName);
DrInputStreamManager inputManager = new DrInputStreamManager(input, inputStage);
s = inputManager;
}
else
{
string msg = String.Format("Unknown input type {0}", info.ioType);
throw new LinqToDryadException(msg);
}
DryadLogger.LogMethodExit();
return(s);
}
