/// <summary>
/// Force the recomputation of the cluster job list.
/// </summary>
/// <param name="virtualCluster">Virtual cluster to use (defined only for some cluster types).</param>
/// <param name="manager">Communication manager.</param>
protected override void RecomputeClusterJobList(string virtualCluster, CommManager manager)
{
this.clusterJobs = new Dictionary <string, ClusterJobInformation>();
var jobs = this.config.AzureClient.ExpandFileOrDirectory(AzureDfsFile.UriFromPath(this.config, "")).ToList();
int done = 0;
foreach (var job in jobs)
{
manager.Token.ThrowIfCancellationRequested();
string jobRootFolder = AzureDfsFile.PathFromUri(this.config, job);
ClusterJobInformation info = this.GetJobInfo(jobRootFolder);
if (info != null)
{
// ReSharper disable once AssignNullToNotNullAttribute
this.clusterJobs.Add(job.AbsolutePath, info);
}
manager.Progress(100 * done++ / jobs.Count);
}
manager.Progress(100);
}
/// <summary>
/// For a folder object, returns the contained file with the specified name.
/// </summary>
/// <param name="filename">File name within the folder.</param>
/// <returns>The file within the folder.</returns>
public override IClusterResidentObject GetFile(string filename)
{
this.PopulateCache();
string filepath;
if (this.client != null)
{
filepath = this.client.Combine(this.path, filename);
}
else
{
filepath = Path.Combine(this.path, filename);
}
bool isFolder = false;
bool isDfsStream = false;
long sz = -1;
if (blocks.ContainsKey(filepath))
{
isDfsStream = true;
sz = blocks[filepath];
}
else if (pages.ContainsKey(filepath))
{
isDfsStream = false;
sz = pages[filepath];
}
else if (this.client != null)
{
// if the client is null the information may be incorrect
isFolder = true;
}
var file = new AzureDfsFile(this.Config, this.Job, this.client, filepath, this.ShouldCacheLocally, isFolder);
file.IsDfsStream = isDfsStream;
file.size = sz;
return(file);
}
/// <summary>
/// If the current object is a folder, it returns the contained objects.
/// </summary>
/// <returns>An iterator over all contained objects that match the specified string.</returns>
/// <param name="match">A shell expression (similar to the argument of Directory.GetFiles()).</param>
public override IEnumerable <IClusterResidentObject> GetFilesAndFolders(string match)
{
this.PopulateCache();
long length = -1;
foreach (var child in this.client.EnumerateDirectory(this.path))
{
Regex re = Utilities.RegexFromSearchPattern(match);
if (!re.IsMatch(child))
{
continue;
}
bool isFolder = false;
bool isDfsStream = false;
if (blocks.ContainsKey(child))
{
isDfsStream = true;
length = blocks[child];
}
else if (pages.ContainsKey(child))
{
isDfsStream = false;
length = pages[child];
}
else if (this.client != null)
{
// otherwise this information may be incorrect
isFolder = true;
}
var file = new AzureDfsFile(this.Config, this.Job, this.client, child, this.ShouldCacheLocally, isFolder);
file.IsDfsStream = isDfsStream;
file.size = length;
yield return(file);
}
}
/// <summary>
/// Extract the job information from a folder with logs on the local machine.
/// </summary>
/// <param name="jobRootFolder">Folder with logs for the specified job.</param>
/// <returns>The job information, or null if not found.</returns>
private ClusterJobInformation GetJobInfo(string jobRootFolder)
{
DateTime date = DateTime.MinValue;
DateTime lastHeartBeat = DateTime.MinValue;
ClusterJobInformation.ClusterJobStatus status = ClusterJobInformation.ClusterJobStatus.Unknown;
bool found = false;
Uri uri = AzureDfsFile.UriFromPath(this.config, jobRootFolder);
var jobsFolders = this.config.AzureClient.ExpandFileOrDirectory(uri).ToList();
jobRootFolder = GetBlobName(this.config.Container, jobRootFolder);
string jobName = jobRootFolder;
foreach (var file in jobsFolders)
{
if (file.AbsolutePath.EndsWith("heartbeat"))
{
string blobName = GetBlobName(this.config.Container, file.AbsolutePath);
var blob = this.config.AzureClient.Container.GetPageBlobReference(blobName);
blob.FetchAttributes();
var props = blob.Metadata;
if (props.ContainsKey("status"))
{
var st = props["status"];
switch (st)
{
case "failure":
status = ClusterJobInformation.ClusterJobStatus.Failed;
break;
case "success":
status = ClusterJobInformation.ClusterJobStatus.Succeeded;
break;
case "running":
status = ClusterJobInformation.ClusterJobStatus.Running;
break;
case "killed":
status = ClusterJobInformation.ClusterJobStatus.Cancelled;
break;
default:
Console.WriteLine("Unknown status " + st);
break;
}
}
if (props.ContainsKey("heartbeat"))
{
var hb = props["heartbeat"];
if (DateTime.TryParse(hb, out lastHeartBeat))
{
lastHeartBeat = lastHeartBeat.ToLocalTime();
if (status == ClusterJobInformation.ClusterJobStatus.Running &&
DateTime.Now - lastHeartBeat > TimeSpan.FromSeconds(40))
{
// job has in fact crashed
status = ClusterJobInformation.ClusterJobStatus.Failed;
}
}
}
if (props.ContainsKey("jobname"))
{
jobName = props["jobname"];
}
if (props.ContainsKey("starttime"))
{
var t = props["starttime"];
if (DateTime.TryParse(t, out date))
{
date = date.ToLocalTime();
}
}
found = true;
}
else if (file.AbsolutePath.Contains("DryadLinqProgram__") &&
// newer heartbeats contain the date
date != DateTime.MinValue)
{
var blob = this.config.AzureClient.Container.GetBlockBlobReference(AzureDfsFile.PathFromUri(this.config, file));
blob.FetchAttributes();
var props = blob.Properties;
if (props.LastModified.HasValue)
{
date = props.LastModified.Value.DateTime;
date = date.ToLocalTime();
}
}
}
if (!found)
{
return(null);
}
TimeSpan running = TimeSpan.Zero;
if (date != DateTime.MinValue && lastHeartBeat != DateTime.MinValue)
{
running = lastHeartBeat - date;
}
var info = new ClusterJobInformation(this.config.Name, "", jobRootFolder, jobName, Environment.UserName, date, running, status);
return(info);
}
/// <summary>
/// For a folder object, returns the contained file with the specified name.
/// </summary>
/// <param name="filename">File name within the folder.</param>
/// <returns>The file within the folder.</returns>
public override IClusterResidentObject GetFile(string filename)
{
this.PopulateCache();
string filepath;
if (this.client != null)
filepath = this.client.Combine(this.path, filename);
else
filepath = Path.Combine(this.path, filename);
bool isFolder = false;
bool isDfsStream = false;
long sz = -1;
if (blocks.ContainsKey(filepath))
{
isDfsStream = true;
sz = blocks[filepath];
}
else if (pages.ContainsKey(filepath))
{
isDfsStream = false;
sz = pages[filepath];
}
else if (this.client != null)
// if the client is null the information may be incorrect
isFolder = true;
var file = new AzureDfsFile(this.Config, this.Job, this.client, filepath, this.ShouldCacheLocally, isFolder);
file.IsDfsStream = isDfsStream;
file.size = sz;
return file;
}
/// <summary>
/// If the current object is a folder, it returns the contained objects.
/// </summary>
/// <returns>An iterator over all contained objects that match the specified string.</returns>
/// <param name="match">A shell expression (similar to the argument of Directory.GetFiles()).</param>
public override IEnumerable<IClusterResidentObject> GetFilesAndFolders(string match)
{
this.PopulateCache();
long length = -1;
foreach (var child in this.client.EnumerateDirectory(this.path))
{
Regex re = Utilities.RegexFromSearchPattern(match);
if (!re.IsMatch(child)) continue;
bool isFolder = false;
bool isDfsStream = false;
if (blocks.ContainsKey(child))
{
isDfsStream = true;
length = blocks[child];
}
else if (pages.ContainsKey(child))
{
isDfsStream = false;
length = pages[child];
}
else if (this.client != null)
// otherwise this information may be incorrect
isFolder = true;
var file = new AzureDfsFile(this.Config, this.Job, this.client, child, this.ShouldCacheLocally, isFolder);
file.IsDfsStream = isDfsStream;
file.size = length;
yield return file;
}
}
/// <summary>
/// For a folder object, returns the contained file with the specified name.
/// </summary>
/// <param name="filename">File name within the folder.</param>
/// <returns>The file within the folder.</returns>
public override IClusterResidentObject GetFile(string filename)
{
this.PopulateCache();
string combined = Path.Combine(this.path, filename);
Uri filepath = UriFromPath(this.Config as AzureDfsClusterConfiguration, combined);
bool isFolder = false;
bool isDfsStream = false;
long sz = -1;
if (this.blocks.ContainsKey(combined))
{
isDfsStream = true;
sz = this.blocks[filepath.AbsolutePath];
}
else if (this.pages.ContainsKey(combined))
{
isDfsStream = false;
sz = this.pages[filepath.AbsolutePath];
}
var file = new AzureDfsFile(this.Config, this.Job, this.client, combined, this.ShouldCacheLocally, isFolder);
file.IsDfsStream = isDfsStream;
file.size = sz;
return file;
}
/// <summary>
/// If the current object is a folder, it returns the contained objects.
/// </summary>
/// <returns>An iterator over all contained objects that match the specified string.</returns>
/// <param name="match">A shell expression (similar to the argument of Directory.GetFiles()).</param>
public override IEnumerable<IClusterResidentObject> GetFilesAndFolders(string match)
{
this.PopulateCache();
long length = -1;
Uri uri = UriFromPath(this.Config as AzureDfsClusterConfiguration, this.path);
Console.WriteLine("AzureDfsFile.GetFileAndFolders({0}) -> {1}", this.path, uri);
foreach (var child in this.client.ExpandFileOrDirectory(uri))
{
Regex re = Utilities.RegexFromSearchPattern(match);
if (!re.IsMatch(child.AbsolutePath)) continue;
bool isFolder = false;
bool isDfsStream = false;
if (this.blocks.ContainsKey(child.AbsolutePath))
{
isDfsStream = true;
length = this.blocks[child.AbsolutePath];
}
else if (this.pages.ContainsKey(child.AbsolutePath))
{
isDfsStream = false;
length = this.pages[child.AbsolutePath];
}
var file = new AzureDfsFile(this.Config, this.Job, this.client, PathFromUri(this.Config as AzureDfsClusterConfiguration, child), this.ShouldCacheLocally, isFolder);
file.IsDfsStream = isDfsStream;
file.size = length;
yield return file;
}
}
/// <summary>
/// The directory where a specific process is created on the cluster.
/// </summary>
/// <param name="identifier">Process identifier</param>
/// <param name="machine">Machine where process ran.</param>
/// <returns>Home directory containing the process information (not working directory of vertex).</returns>
/// <param name="job">Job where the process belongs.</param>
/// <param name="terminated">True if vertex is terminated.</param>
public override IClusterResidentObject ProcessDirectory(DryadProcessIdentifier identifier, bool terminated, string machine, DryadLinqJobSummary job)
{
if (identifier.ToString() == "jm")
{
// The job manager process is special
var result = new AzureDfsFile(this, job, this.AzureClient, job.ClusterJobId, terminated, true);
return result;
}
// vertices not supported
return null;
}
