/// <summary>
/// Process a single dataset and fetch the info about it.
/// </summary>
protected override void ProcessRecord()
{
var trimmedDSName = DatasetName.Trim();
if (_resultsCache.Value[trimmedDSName] is PSGRIDDatasetInfo cHit)
{
WriteObject(cHit);
}
else
{
// Setup for verbosity if we need it.
var listener = new PSListener(this);
Trace.Listeners.Add(listener);
try
{
// Where are we going to be doing query on - we need a machine.
var sm = JobParser.GetSubmissionMachine();
// Get the remove environment configured if it needs to be
if (_connection == null)
{
_connection = new SSHConnection(sm.MachineName, sm.UserName);
_connection
.Apply(() => DisplayStatus("Setting up ATLAS"))
.setupATLAS()
.Apply(() => DisplayStatus("Setting up Rucio"))
.setupRucio(_gridCredentials.Username)
.Apply(() => DisplayStatus("Acquiring GRID credentials"))
.VomsProxyInit("atlas", failNow: () => Stopping);
}
// Great - get the info on this dataset.
var fileInfo = _connection
.Apply(() => DisplayStatus($"Checking for info on {trimmedDSName}."))
.FileInfoFromGRID(trimmedDSName, failNow: () => Stopping);
// Next, build the resulting thingy.
var r = new PSGRIDDatasetInfo()
{
DatasetName = trimmedDSName,
nFiles = fileInfo.Count,
TotalSizeMB = (int)fileInfo.Sum(fi => fi.size),
FileInfo = fileInfo.ToArray()
};
_resultsCache.Value[trimmedDSName] = r;
using (var pp = listener.PauseListening())
{
WriteObject(r);
}
}
finally
{
Trace.Listeners.Remove(listener);
}
}
}
/// <summary>
/// Create the connection, and setup the appropriate stuff for using AMI.
/// </summary>
/// <returns></returns>
private SSHConnection GetConnection()
{
if (_connection != null)
{
return(_connection);
}
// Establish the connection
var sm = JobParser.GetSubmissionMachine();
var connection = new SSHConnection(sm.MachineName, sm.UserName);
// Setup for using ami.
connection.Apply(() => DisplayStatus("Setting up ATLAS"))
.setupATLAS()
.Apply(() => DisplayStatus("Setting up pyAMI"))
.ExecuteLinuxCommand("lsetup pyami")
.Apply(() => DisplayStatus("Acquiring GRID credentials"))
.VomsProxyInit("atlas", failNow: () => Stopping);
// And the connection is now ready for use!
_connection = connection;
return(_connection);
}
/// <summary>
/// Load up the job requested. Fail, obviously, if we can't.
/// </summary>
protected override void ProcessRecord()
{
// Setup for verbosity if we need it.
var listener = new PSListener(this);
Trace.Listeners.Add(listener);
try
{
// Get the job
var job = JobParser.FindJob(JobName, JobVersion);
// Get the expected resulting dataset name. Since this will be a personal
// dataset, we need to get the GRID info.
var originalDatasetName = DatasetName.Trim();
string resultDatasetName = job.ResultingDataSetName(originalDatasetName, _gridCredentials, JobIteration);
// See if there is already a job defined that will produce this
var pandaJob = (resultDatasetName + "/").FindPandaJobWithTaskName(useCacheIfPossible: !DoNotUsePandaTaskCache.IsPresent);
if (pandaJob == null)
{
// Where are we going to be doing the submission on?
var sm = JobParser.GetSubmissionMachine();
// Get the remove environment configured if it needs to be
var firstJob = false;
if (_connection == null)
{
firstJob = true;
_connection = new SSHConnection(sm.MachineName, sm.UserName);
_connection
.Apply(() => DisplayStatus("Setting up ATLAS"))
.setupATLAS(dumpOnly: WhatIf.IsPresent)
.Apply(() => DisplayStatus("Setting up Rucio"))
.setupRucio(_gridCredentials.Username, dumpOnly: WhatIf.IsPresent)
.Apply(() => DisplayStatus("Acquiring GRID credentials"))
.VomsProxyInit("atlas", failNow: () => Stopping, dumpOnly: WhatIf.IsPresent);
}
// Check to see if the original dataset exists. We will use the location known as Local for doing the
// setup, I suppose.
var files = _connection
.Apply(() => DisplayStatus("Checking dataset exists on the GRID"))
.FilelistFromGRID(originalDatasetName, failNow: () => Stopping, dumpOnly: WhatIf.IsPresent);
if (files.Length == 0 && !WhatIf.IsPresent)
{
throw new ArgumentException($"Dataset '{originalDatasetName}' has zero files - won't submit a job against it!");
}
// Submit the job
_connection
.SubmitJobAsync(job, originalDatasetName, resultDatasetName, DisplayStatus, failNow: () => Stopping, sameJobAsLastTime: !firstJob, dumpOnly: WhatIf.IsPresent)
.WaitAndUnwrapException();
// Try to find the job again if requested. The submission can take a very long time to show up in
// big panda, so skip unless requested.
if (WaitForPandaRegistration)
{
var pandJobResult = Policy
.Handle <InvalidOperationException>()
.WaitAndRetryForever(nthRetry => TimeSpan.FromMinutes(1),
(e, ts) => { WriteWarning($"Failed to find the submitted panda job on bigpanda: {e.Message}. Will wait one minute and try again."); })
.ExecuteAndCapture(() => FindPandaJobForDS(resultDatasetName));
if (pandJobResult.Outcome != OutcomeType.Successful)
{
throw pandJobResult.FinalException;
}
pandaJob = pandJobResult.Result;
}
}
// Return a helper obj that contains the info about this job that can be used by other commands.
if (pandaJob != null)
{
var r = new AtlasPandaTaskID()
{
ID = pandaJob.jeditaskid, Name = pandaJob.taskname
};
using (var pp = listener.PauseListening())
{
WriteObject(r);
}
}
} finally
{
Trace.Listeners.Remove(listener);
}
}
