在 HDInsight 中使用 Azure PowerShell 运行 Apache Sqoop 作业

项目
06/17/2024

了解如何使用 Azure PowerShell 运行 Azure HDInsight 中的 Apache Sqoop 作业，以便在 HDInsight 群集与 Azure SQL 数据库或 SQL Server 之间导入和导出数据。本文是在 HDInsight 中将 Apache Sqoop 与 Hadoop 配合使用一文的延续。

先决条件

一个安装有 Azure PowerShell AZ 模块的工作站。
已完成在 HDInsight 中将 Apache Sqoop 与 Hadoop 配合使用一文中的设置测试环境步骤。
熟悉 Sqoop。有关详细信息，请参阅 Sqoop 用户指南。

Sqoop 导出

从 Hive 到 SQL。

此示例将数据从 Hive hivesampletable 表导出到 SQL 中的 mobiledata 表。设置以下变量的值，然后执行命令。

$hdinsightClusterName = ""
$httpPassword = ''
$sqlDatabasePassword = ''

# These values only need to be changed if the template was not followed.
$httpUserName = "admin"
$sqlServerLogin = "sqluser"
$sqlServerName = $hdinsightClusterName + "dbserver"
$sqlDatabaseName = $hdinsightClusterName + "db"

$pw = ConvertTo-SecureString -String $httpPassword -AsPlainText -Force
$httpCredential = New-Object System.Management.Automation.PSCredential($httpUserName,$pw)

# Connection string
$connectionString = "jdbc:sqlserver://$sqlServerName.database.chinacloudapi.cn;user=$sqlServerLogin@$sqlServerName;password=$sqlDatabasePassword;database=$sqlDatabaseName"

# start export
New-AzHDInsightSqoopJobDefinition `
    -Command "export --connect $connectionString --table mobiledata --hcatalog-table hivesampletable" `
    | Start-AzHDInsightJob `
        -ClusterName $hdinsightClusterName `
        -HttpCredential $httpCredential

替代执行

以下代码执行相同的导出；不过，它提供了一种方法来读取输出日志。执行代码，开始导出。

$sqoopCommand = "export --connect $connectionString --table mobiledata --hcatalog-table hivesampletable"

$sqoopDef = New-AzHDInsightSqoopJobDefinition `
    -Command $sqoopCommand

$sqoopJob = Start-AzHDInsightJob `
                -ClusterName $hdinsightClusterName `
                -HttpCredential $httpCredential `
                -JobDefinition $sqoopDef

下面的代码显示输出日志。执行下面的代码：

Get-AzHDInsightJobOutput `
    -ClusterName $hdinsightClusterName `
    -HttpCredential $httpCredential `
    -JobId $sqoopJob.JobId `
    -DisplayOutputType StandardError

Get-AzHDInsightJobOutput `
    -ClusterName $hdinsightClusterName `
    -HttpCredential $httpCredential `
    -JobId $sqoopJob.JobId `
    -DisplayOutputType StandardOutput

如果收到错误消息“The specified blob does not exist.”，请在几分钟后重试。

Sqoop 导入

从 SQL 到 Azure 存储。此示例将数据从 SQL 中的 mobiledata 表导入 HDInsight 上的 wasb:///tutorials/usesqoop/importeddata 目录。数据中的字段将通过制表符分隔，并且相关行由换行符终止。此示例假定你已完成上一示例。

$sqoopCommand = "import --connect $connectionString --table mobiledata --target-dir wasb:///tutorials/usesqoop/importeddata --fields-terminated-by '\t' --lines-terminated-by '\n' -m 1"

$sqoopDef = New-AzHDInsightSqoopJobDefinition `
    -Command $sqoopCommand

$sqoopJob = Start-AzHDInsightJob `
                -ClusterName $hdinsightClusterName `
                -HttpCredential $httpCredential `
                -JobDefinition $sqoopDef

Get-AzHDInsightJobOutput `
    -ClusterName $hdinsightClusterName `
    -HttpCredential $httpCredential `
    -JobId $sqoopJob.JobId `
    -DisplayOutputType StandardError

Get-AzHDInsightJobOutput `
    -ClusterName $hdinsightClusterName `
    -HttpCredential $httpCredential `
    -JobId $sqoopJob.JobId `
    -DisplayOutputType StandardOutput

其他 Sqoop 导出示例

这是一个说服力很强的示例，它从默认存储帐户中的 /tutorials/usesqoop/data/sample.log 导出数据，然后将其导入到 SQL Server 数据库中名为 log4jlogs 的表中。此示例不依赖于前面的示例。

下面的 PowerShell 脚本预处理源文件，然后将其导出到 log4jlogs 表。请将 CLUSTERNAME、CLUSTERPASSWORD 和 SQLPASSWORD 替换为在先决条件部分使用的值。

<#------ BEGIN USER INPUT ------#>
$hdinsightClusterName = "CLUSTERNAME"
$httpUserName = "admin"  #default is admin, update as needed
$httpPassword = 'CLUSTERPASSWORD'
$sqlDatabasePassword = 'SQLPASSWORD'
<#------- END USER INPUT -------#>

# Other fixed variable that should be used as is
$sqlServerName = $hdinsightClusterName + "dbserver"
$sqlDatabaseName = $hdinsightClusterName + "db"
$tableName_log4j = "log4jlogs"
$exportDir_log4j = "/tutorials/usesqoop/data"
$sourceBlobName = "example/data/sample.log"
$destBlobName = "tutorials/usesqoop/data/sample.log"
$sqljdbcdriver = "/user/oozie/share/lib/sqoop/mssql-jdbc-7.0.0.jre8.jar"

$cluster = Get-AzHDInsightCluster -ClusterName $hdinsightClusterName
$defaultStorageAccountName = $cluster.DefaultStorageAccount -replace '.blob.core.chinacloudapi.cn'
$defaultStorageContainer = $cluster.DefaultStorageContainer
$resourceGroup = $cluster.ResourceGroup

$sqlServer = Get-AzSqlServer -ResourceGroupName $resourceGroup -ServerName $sqlServerName
$sqlServerLogin = $sqlServer.SqlAdministratorLogin
$sqlServerFQDN = $sqlServer.FullyQualifiedDomainName

#Connect to Azure subscription
Write-Host "`nConnecting to your Azure subscription ..." -ForegroundColor Green
try{Get-AzContext}
catch{Connect-AzAccount -Environment AzureChinaCloud}

#pre-process the source file
Write-Host "`nPreprocessing the source file ..." -ForegroundColor Green

# This procedure creates a new file with $destBlobName
# Define the connection string
$defaultStorageAccountKey = (Get-AzStorageAccountKey `
                                -ResourceGroupName $resourceGroup `
                                -Name $defaultStorageAccountName)[0].Value

# Create block blob objects referencing the source and destination blob.
$storageAccount = Get-AzStorageAccount `
    -ResourceGroupName $resourceGroup `
    -Name $defaultStorageAccountName

$storageContainer = ($storageAccount |Get-AzStorageContainer -Name $defaultStorageContainer).CloudBlobContainer

$sourceBlob = $storageContainer.GetBlockBlobReference($sourceBlobName)
$destBlob = $storageContainer.GetBlockBlobReference($destBlobName)

# Define a MemoryStream and a StreamReader for reading from the source file
$stream = New-Object System.IO.MemoryStream
$stream = $sourceBlob.OpenRead()
$sReader = New-Object System.IO.StreamReader($stream)

# Define a MemoryStream and a StreamWriter for writing into the destination file
$memStream = New-Object System.IO.MemoryStream
$writeStream = New-Object System.IO.StreamWriter $memStream

# Pre-process the source blob
$exString = "java.lang.Exception:"
while(-Not $sReader.EndOfStream){
    $line = $sReader.ReadLine()
    $split = $line.Split(" ")

    # remove the "java.lang.Exception" from the first element of the array
    # for example: java.lang.Exception: 2012-02-03 19:11:02 SampleClass8 [WARN] problem finding id 153454612
    if ($split[0] -eq $exString){
        #create a new ArrayList to remove $split[0]
        $newArray = [System.Collections.ArrayList] $split
        $newArray.Remove($exString)

        # update $split and $line
        $split = $newArray
        $line = $newArray -join(" ")
    }

    # remove the lines that has less than 7 elements
    if ($split.count -ge 7){
        write-host $line
        $writeStream.WriteLine($line)
    }
}

# Write to the destination blob
$writeStream.Flush()
$memStream.Seek(0, "Begin")
$destBlob.UploadFromStream($memStream)

#export the log file from the cluster to SQL
Write-Host "Exporting the log file ..." -ForegroundColor Green

$pw = ConvertTo-SecureString -String $httpPassword -AsPlainText -Force
$httpCredential = New-Object System.Management.Automation.PSCredential($httpUserName,$pw)

# Connection string
$connectionString = "jdbc:sqlserver://$sqlServerFQDN;user=$sqlServerLogin@$sqlServerName;password=$sqlDatabasePassword;database=$sqlDatabaseName"

# Submit a Sqoop job
$sqoopDef = New-AzHDInsightSqoopJobDefinition `
    -Command "export --connect $connectionString --table $tableName_log4j --export-dir $exportDir_log4j --input-fields-terminated-by \0x20 -m 1" `
    -Files $sqljdbcdriver

$sqoopJob = Start-AzHDInsightJob `
                -ClusterName $hdinsightClusterName `
                -HttpCredential $httpCredential `
                -JobDefinition $sqoopDef

Wait-AzHDInsightJob `
    -ResourceGroupName $resourceGroup `
    -ClusterName $hdinsightClusterName `
    -HttpCredential $httpCredential `
    -JobId $sqoopJob.JobId

Write-Host "Standard Error" -BackgroundColor Green
Get-AzHDInsightJobOutput `
    -ResourceGroupName $resourceGroup `
    -ClusterName $hdinsightClusterName `
    -DefaultStorageAccountName $defaultStorageAccountName `
    -DefaultStorageAccountKey $defaultStorageAccountKey `
    -DefaultContainer $defaultStorageContainer `
    -HttpCredential $httpCredential `
    -JobId $sqoopJob.JobId `
    -DisplayOutputType StandardError

Write-Host "Standard Output" -BackgroundColor Green
Get-AzHDInsightJobOutput `
    -ResourceGroupName $resourceGroupName `
    -ClusterName $hdinsightClusterName `
    -DefaultStorageAccountName $defaultStorageAccountName `
    -DefaultStorageAccountKey $defaultStorageAccountKey `
    -DefaultContainer $defaultStorageContainer `
    -HttpCredential $httpCredential `
    -JobId $sqoopJob.JobId `
    -DisplayOutputType StandardOutput

限制

基于 Linux 的 HDInsight 存在以下限制：

批量导出：用于将数据导出到 SQL 的 Sqoop 连接器目前不支持批量插入。
批处理：如果在执行插入时使用 -batch 开关，Sqoop 将执行多次插入而不是批处理插入操作。

后续步骤

现在，你已了解了如何使用 Sqoop。若要了解更多信息，请参阅以下文章：

将 Apache Oozie 和 HDInsight 配合使用：在 Oozie 工作流中使用 Sqoop 操作。
将数据上传到 HDInsight：了解将数据上传到 HDInsight 或 Azure Blob 存储的其他方法。

通过