/// <summary>
/// Gets the metric values.
/// </summary>
/// <param name="dataLakeClient">The data lake client.</param>
/// <param name="testContent">Content of the test.</param>
/// <param name="sparkClient">The spark client.</param>
/// <param name="sparkClientSettings">The spark client settings.</param>
/// <returns>MetricValues.</returns>
private static MetricValues GetMetricValues(
DataLakeClient dataLakeClient,
SparkMetricsTestContent testContent,
SparkClient sparkClient,
SparkClientSettings sparkClientSettings,
CancellationToken cancellationToken)
{
string currentPath = testContent.GetCurrentStreamPath();
string previousPath = testContent.GetPreviousStreamPath();
var current = GetMetricValue(
testContent,
sparkClient,
sparkClientSettings,
currentPath,
dataLakeClient,
cancellationToken);
var previous = GetMetricValue(
testContent,
sparkClient,
sparkClientSettings,
previousPath,
dataLakeClient,
cancellationToken);
var metricValues = new MetricValues
{
Baseline = previous,
Current = current
};
return(metricValues);
}
public Configuration(
ISecretProvider secretProvider = null,
DatabricksClientWrapper databricksClientWrapper = null,
DataLakeClient dataLakeClient = null)
{
this.DataLakeClient = dataLakeClient;
this.DatabricksClientWrapper = databricksClientWrapper;
}
/// <summary>
/// Processes the metrics correctness test.
/// </summary>
/// <param name="dataLakeClient">The data lake client.</param>
/// <param name="testContent">Content of the test.</param>
/// <param name="sparkClient">The spark client.</param>
/// <param name="sparkClientSettings">The spark client settings.</param>
/// <returns>SparkMetricsTestResult.</returns>
/// <exception cref="NotSupportedException">ComparisonType {testContent.ComparisonType}</exception>
private static List <MetricsTestResult> ProcessMetricsCorrectnessTest(
DataLakeClient dataLakeClient,
SparkMetricsTestContent testContent,
SparkClient sparkClient,
SparkClientSettings sparkClientSettings,
CancellationToken cancellationToken)
{
MetricValues metricValues = null;
switch (testContent.ComparisonType)
{
case ComparisonType.DayOverDay:
case ComparisonType.WeekOverWeek:
case ComparisonType.MonthOverMonth:
case ComparisonType.YearOverYear:
metricValues = GetMetricValues(
dataLakeClient,
testContent,
sparkClient,
sparkClientSettings,
cancellationToken);
break;
case ComparisonType.VarianceToTarget:
var metricValue = GetMetricValue(
testContent,
sparkClient,
sparkClientSettings,
testContent.GetCurrentStreamPath(),
dataLakeClient,
cancellationToken);
testContent.NotebookParameters["cmdText"] = testContent.NotebookParameters["targetCmdText"];
var targetValue = GetMetricValue(
testContent,
sparkClient,
sparkClientSettings,
testContent.TargetStreamPath != null ? testContent.GetCurrentTargetStreamPath() : testContent.GetCurrentStreamPath(),
dataLakeClient,
cancellationToken);
metricValues = new MetricValues
{
Baseline = targetValue,
Current = metricValue
};
break;
default:
throw new NotSupportedException($"ComparisonType {testContent.ComparisonType} not supported for SparkWorker");
}
var results = new List <MetricsTestResult>();
foreach (var threshold in testContent.Thresholds)
{
var result = new MetricsTestResult
{
ComparisonType = testContent.ComparisonType,
Date = testContent.Date,
BaselineMetricValue = metricValues.Baseline[threshold.Name],
MetricValue = metricValues.Current[threshold.Name],
LowerBoundThreshold = threshold.LowerBound,
UpperBoundThreshold = threshold.UpperBound,
PercentDiff = MetricValues.ComputePercentDiff(metricValues, threshold.Name),
PreviousDate = testContent.GetPreviousDate(),
TestName = testRunName,
TestRunId = testRunId,
MetricName = threshold.Name
};
results.Add(result);
}
return(results);
}
/// <summary>
/// Gets the metric value.
/// </summary>
/// <param name="testContent">Content of the test.</param>
/// <param name="sparkClient">The spark client.</param>
/// <param name="sparkClientSettings">The spark client settings.</param>
/// <param name="stream">The stream.</param>
/// <param name="dataLakeClient">The data lake client.</param>
/// <returns>System.Nullable<System.Double>.</returns>
/// <exception cref="InvalidOperationException">Stream does not exist : {stream}</exception>
private static IDictionary <string, double> GetMetricValue(
SparkMetricsTestContent testContent,
SparkClient sparkClient,
SparkClientSettings sparkClientSettings,
string stream,
DataLakeClient dataLakeClient,
CancellationToken cancellationToken)
{
if (!dataLakeClient.CheckExists(testContent.GetDatalakeStore(), stream))
{
throw new FileNotFoundException($"Stream does not exist : {stream}");
}
var sparkRequest = new SparkClientRequest
{
NodeType = sparkClientSettings.NodeType,
NumWorkersMin = sparkClientSettings.NumWorkersMin,
NumWorkersMax = sparkClientSettings.NumWorkersMax,
CostPerNode = GetCostPerNode(sparkClientSettings.NodeTypes, sparkClientSettings.NodeType),
Libraries = sparkClientSettings.Libraries,
NotebookPath = testContent.NotebookPath,
NotebookParameters = testContent.NotebookParameters ?? new Dictionary <string, string>(),
TestRunId = testRunId,
TimeoutSeconds = sparkClientSettings.TimeoutSeconds
};
Console.WriteLine($"Running notebook={testContent.NotebookPath} for DataLakeStore={testContent.GetDatalakeStore()}, Path={stream}");
var mountPoint = GetMountPoint(testContent.GetDatalakeStore(), stream);
Console.WriteLine($"Running notebook={testContent.NotebookPath} for mountPoint={mountPoint}");
string streamPath = mountPoint;
// Disable this function
//if (testContent.ConvertToParquet)
//{
// var parquetFile = messageTag.MountPointToParquetFile[mountPoint];
// Console.WriteLine($"Running notebook={testContent.NotebookPath} using parquetFile={parquetFile}");
// streamPath = parquetFile;
//}
//else
//{
// streamPath = mountPoint;
//}
sparkRequest.NotebookParameters["streamPath"] = streamPath;
Console.WriteLine($"Notebook parameters : {string.Join(", ", sparkRequest.NotebookParameters.Select(t => t.Key + "=" + t.Value))}");
// Log request to OMS
var response = sparkClient.RunNotebook(sparkRequest, cancellationToken);
response.TestRunId = testRunId;
if (response.IsRunSuccess())
{
// For format reference see:
// https://pandas.pydata.org/pandas-docs/version/0.24.2/reference/api/pandas.DataFrame.to_json.html
var resultDataFrame = JsonConvert.DeserializeObject <Dictionary <string, Dictionary <string, double> > >(response.RunOutput);
var resultDictionary = new Dictionary <string, double>();
foreach (var pair in resultDataFrame)
{
// pair.Value is the column name
if (pair.Value == null || pair.Value.Count == 0)
{
throw new InvalidOperationException("Result does not contain any rows");
}
// We take the first row only
resultDictionary.Add(pair.Key, pair.Value.First().Value);
}
return(resultDictionary);
}
else
{
Console.WriteLine("Error getting metric.");
Console.WriteLine(JObject.Parse(JsonConvert.SerializeObject(response)));
throw new Exception($"Error getting metric. TestRun = {testRunId}, Spark job {response?.Run?.RunId} failed");
}
}
public static async Task Run(
[TimerTrigger("%RedisDataPumpSchedule%", RunOnStartup = true)] TimerInfo myTimer,
ILogger log)
{
// Get startTicks
long startTicks = DateTime.UtcNow.Ticks;
using (log.BeginScope(logProperties))
// This DataLake client is used for writing exceptions that occur inside RedisDataPump. Do NOT remove it.
using (DataLakeClient dataLake = new DataLakeClient(log))
{
try
{
List <Task> tasks = new List <Task>();
// Pump and dump Redis lists (default: res-datalake)
tasks.Add(new DataLakeClient(log).FlushRedis(startTicks, 55000));
// Named
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.AccountsPayableInvoiceInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.AccountsPayableInvoiceOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.AccountsPayablePayInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.AdvanceShippingNoticeInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.Adyen)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.Coupon)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.CouponDeliveryOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.CustomerInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.CustomerOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.DropshipOrderInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.EdiPurchaseOrder)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.EmployeeInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.FraudOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.GeneralLedgerOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.GenericApi)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.GenericInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.GenericOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.Invalid)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.InventoryInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.InventoryOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.Loyalty)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.LoyaltyInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.LoyaltyOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.MapleLake)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.PricingOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.ProductInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.ProductOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.PurchaseAgreementOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.PurchaseOrderConfirmInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.PurchaseOrderInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SalesOrderConfirmOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SalesOrderDropshipOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SalesOrderInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SalesOrderOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SalesOrderReturnInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SalesOrderReturnOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SalesOrderSurveyOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SapReturnOrderInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SapSalesOrderUpdateInbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SupportImport)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.SvsGiftCardOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.TaxwareOutbound)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.TradeAgreementInbound_Map)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.TradeAgreementInbound_Purch)).FlushRedis(startTicks, 55000));
tasks.Add(new DataLakeClient(log, Extension.ToString(Res.ServiceName.TradeAgreementOutbound)).FlushRedis(startTicks, 55000));
// Wait for all Tasks
Task.WaitAll(tasks.ToArray());
}
catch (Exception ex)
{
await RecoveryMessage.SendAsync(new RecoveryMessage()
{
ServiceName = ServiceName.DataLake,
FamilyName = Extension.ToString(FamilyName.DataLake),
FunctionName = nameof(RedisDataPump),
MethodName = nameof(Run),
ResourceGroup = Extension.ToString(Res.ServiceName.DataLake),
Exception = ex,
}, dataLake, log);
}
}
}
