Hadoop FS

The Hadoop FS destination writes data to Hadoop Distributed File System (HDFS). You can also use the destination to write to Azure Blob storage. For information about supported versions, see Supported Systems and Versions in the Data Collector documentation.

You can write data as flat files or Hadoop sequence files. You can also use the whole file data format to write whole files to HDFS.

When you configure a Hadoop FS destination, you can define a directory template and time basis to determine the output directories that the destination creates and the files where records are written.

As part of the Drift Synchronization Solution for Hive, you can alternatively use record header attributes to perform record-based writes. You can write records to the specified directory, use the defined Avro schema, and roll files based on record header attributes. For more information, see Record Header Attributes for Record-Based Writes.

You can define a file prefix and suffix, the data time zone, and properties that define when the destination closes a file. You can specify the amount of time that a record can be written to its associated directory and what happens to late records.

The destination can generate events for an event stream. For more information about the event framework, see Dataflow Triggers Overview.

When necessary, you can enable Kerberos authentication. You can also specify a Hadoop user to impersonate, define a Hadoop configuration file directory, and add Hadoop configuration properties as needed.

You can use Gzip, Bzip2, Snappy, LZ4, and other compression formats to write output files.

Directory Templates

By default, the Hadoop FS destination uses directory templates to create output and late record directories. Hadoop FS writes records to the directories based on the configured time basis.

You can alternatively write records to directories based on the targetDirectory record header attribute. Using the targetDirectory attribute disables the ability to define directory templates.

When you define a directory template, you can use a mix of constants, field values, and datetime variables. You can use the every function to create new directories at regular intervals based on hours, minutes, or seconds, starting on the hour. You can also use the record:valueOrDefault function to create new directories from field values or a default in the directory template.

For example, the following directory template creates output directories for event data based on the state and timestamp of a record with hours as the smallest unit of measure, creating a new directory every hour:
 /outputfiles/${record:valueOrDefault("/State", "unknown")}/${YY()}-${MM()}-${DD()}-${hh()}
You can use the following elements in a directory template:
Constants
You can use any constant, such as output.
Datetime Variables
You can use datetime variables, such as ${YYYY()} or ${DD()}. The destination creates directories as needed, based on the smallest datetime variable that you use. For example, if the smallest variable is hours, then the directories are created for every hour of the day that receives output records.
When you use datetime variables in an expression, use all of the datetime variables between one of the year variables and the smallest variable that you want to use. Do not skip a variable within the progression. For example, to create directories on a daily basis, use a year variable, a month variable, and then a day variable. You might use one of the following datetime variable progressions:
${YYYY()}-${MM()}-${DD()}
${YY()}_${MM()}_${DD()}
For details about datetime variables, see Datetime Variables.
every function
You can use the every function in a directory template to create directories at regular intervals based on hours, minutes, or seconds, beginning on the hour. The intervals should be a submultiple or integer factor of 60. For example, you can create directories every 15 minutes or 30 seconds.
Use the every function to replace the smallest datetime variable used in the template.
For example, the following directory template creates directories every 5 minutes, starting on the hour:
/HDFS_output/${YYYY()}-${MM()}-${DD()}-${hh()}-${every(5,mm())}
For details about the every function, see Miscellaneous Functions.
record:valueOrDefault function
You can use the record:valueOrDefault function in a directory template to create directories with the value of a field or the specified default value if the field does not exist or if the field is null:
${record:valueOrDefault(<field path>, <default value>)}
For example, the following directory template creates a directory based on the Product field every day, and if the Product field is empty or null, uses Misc in the directory path:
/${record:valueOrDefault("/Product", "Misc")}/${YY()}-${MM()}-${DD()}
This template might create the following paths:
/Shirts/2015-07-31 
/Misc/2015-07-31

Time Basis

When using directory templates, the time basis helps determine when directories are created. It also determines the directory Hadoop FS uses when writing a record, and whether a record is late.

When using the targetDirectory record header attribute to write records, the time basis determines only whether a record is late.

You can use the following times as the time basis:

Processing Time
When you use processing time as the time basis, the destination creates directories based on the processing time and the directory template, and writes records to the directories based on when they are processed.
For example, say a directory template creates directories every minute and the time basis is the time of processing. Then, directories are created for every minute that the destination writes output records. And the output records are written to the directory for that minute of processing.
To use the processing time as the time basis, use the following expression: ${time:now()}. This is the default time basis.
Record Time
When you use the time associated with a record as the time basis, you specify a Date field in the record. The destination creates directories based on the datetimes associated with the records and writes the records to the appropriate directories.
For example, say a directory template creates directories every hour and the time basis is based on the record. Then, directories are created for every hour associated with output records and the destination writes the records to the related output directory.
To use a time associated with the record, use an expression that calls a field and resolves to a datetime value, such as ${record:value("/Timestamp")}.

Late Records and Late Record Handling

When you use a record time as the time basis, you can define a time limit for records to be written to their associated output file. When the destination creates a new output file in a new directory, the previous output file is kept open for the specified late record time limit. When records that belong in that file arrive within the time limit, the destination writes the records to the open output file. When the late record time limit is reached, the output file is closed and any record that arrives past this limit is considered late.

Tip: The late records properties are not applicable if you use processing time as the time basis. If you use processing time, set the late record time limit to one second.

You can send late records to a late records file or to the stage for error handling. When you send records to a late records file, you define a late records directory template.

For example, you use a record time as the time basis, configure a one-hour late record time limit, configure late records to be sent to the stage for error handling, and use the default directory template value:
/tmp/out/${YYYY()}-${MM()}-${DD()}-${hh()} 

The first records that arrive have a datetime between the hours of 02:00 and 02:59, and so are written to an output file in the 02 directory. When records with a datetime between the hours of 03:00 and 03:59 arrive, the destination creates a new file in an 03 directory. The destination keeps the file in the 02 directory open for another hour.

If a record with a datetime between the hours of 02:00 and 02:59 arrives before the hour time limit, the destination writes the record to the open file in the 02 directory. After one hour, the destination closes the output file in the 02 directory. Any records with a datetime between the hours of 02:00 and 02:59 that arrive after the one-hour time limit are considered late. The late records are sent to the stage for error handling.

Timeout to Close Idle Files

You can configure the maximum time that an open output file can remain idle. After no records are written to an output file for the specified amount of time, the Hadoop FS destination closes the file.

You might want to configure an idle timeout when output files remain open and idle for too long, thus delaying another system from processing the files.

Output files might remain idle for too long for the following reasons:
  • You configured the maximum number of records to be written to output files or the maximum size of output files, but records have stopped arriving. An output file that has not reached the maximum number of records or the maximum file size stays open until more records arrive.
  • You configured a date field in the record as the time basis and have configured a late record time limit, but records arrive in chronological order. When a new directory is created, the output file in the previous directory remains open for the configured late record time limit. However, no records are ever written to the open file in the previous directory.

    For example, when a record with a datetime of 03:00 arrives, the destination creates a new file in a new 03 directory. The previous file in the 02 directory is kept open for the late record time limit, which is an hour by default. However, when records arrive in chronological order, no records that belong in the 02 directory arrive after the 03 directory is created.

In either situation, configure an idle timeout so that other systems can process the files sooner, instead of waiting for the configured maximum records, maximum file size, or late records conditions to occur.

Recovery

The Hadoop FS destination supports recovery after an unexpected stop of the pipeline by renaming temporary files when the pipeline restarts.

The destination names temporary open output files using the following format:
_tmp_<prefix>_<runnerId>
Where <prefix> is the file prefix defined for the destination and <runnerId> is the ID of the pipeline runner performing the pipeline processing. For example, when the destination prefix is defined as sdc and the destination runs from a single-threaded pipeline, the temporary file is named like so: _tmp_sdc_0.
When the destination closes the file, either after it is fully written, after the idle timeout expires, or when you deliberately stop the pipeline, it renames the file to remove the _tmp_ string and to replace the pipeline runner ID with a random unique identifier like so:
<prefix>_e7ce67c5-013d-47a7-9496-8c882ddb28a0

However, when the pipeline stops unexpectedly, the temporary files remain. When the pipeline restarts, the destination scans all subdirectories of the defined directory template to rename any temporary files that match the defined prefix for the destination. After the destination renames the temporary files, it continues writing to new output files.

Note: The destination renames all temporary files that match the defined prefix in all subdirectories of the defined directory template, even if those files were not written by that pipeline. So if you happen to have another file whose name begins with the same pattern - _tmp_<prefix> - the destination renames that file also.
The destination might not rename all temporary files in the following situations:
The directory template includes an expression with the record:value or record:valueOrDefault function.
If the record:value or record:valueOrDefault function evaluates to an empty string or to a subdirectory, the destination cannot determine those locations when the pipeline restarts. As a result, the destination cannot rename any temporary files written to those locations.
For example, let’s assume that the directory template is defined as follows:
/tmp/out/${YY()}-${MM()}-${DD()}/${sdc:hostname()}/${record:value('/a')}/${record:value('/b')}
If the expression ${record:value('/b')} evaluates to an empty string or to a subdirectory such as /folder1/folder2, then the destination cannot determine those locations when the pipeline restarts.
The directory is defined in the targetDirectory record header attribute.
When the directory is defined in the targetDirectory record header attribute, the destination cannot determine where to look for temporary files when the pipeline restarts. As a result, it cannot rename the temporary files.

In either of these situations, you must manually rename the temporary files.

File recovery can slow down the pipeline as it restarts. If needed, you can configure the destination to skip file recovery.

Data Formats

Hadoop FS writes data to HDFS based on the data format that you select. You can use the following data formats:
Avro
The destination writes records based on the Avro schema. You can use one of the following methods to specify the location of the Avro schema definition:
  • In Pipeline Configuration - Use the schema that you provide in the stage configuration.
  • In Record Header - Use the schema included in the avroSchema record header attribute.
  • Confluent Schema Registry - Retrieve the schema from Confluent Schema Registry. Confluent Schema Registry is a distributed storage layer for Avro schemas. You can configure the destination to look up the schema in Confluent Schema Registry by the schema ID or subject.

    If using the Avro schema in the stage or in the record header attribute, you can optionally configure the destination to register the Avro schema with Confluent Schema Registry.

The destination includes the schema definition in each file.
You can compress data with an Avro-supported compression codec. When using Avro compression, avoid using other compression properties in the destination.
Binary
The stage writes binary data to a single field in the record.
Delimited
The destination writes records as delimited data. When you use this data format, the root field must be list or list-map.
You can use the following delimited format types:
  • Default CSV - File that includes comma-separated values. Ignores empty lines in the file.
  • RFC4180 CSV - Comma-separated file that strictly follows RFC4180 guidelines.
  • MS Excel CSV - Microsoft Excel comma-separated file.
  • MySQL CSV - MySQL comma-separated file.
  • Tab-Separated Values - File that includes tab-separated values.
  • PostgreSQL CSV - PostgreSQL comma-separated file.
  • PostgreSQL Text - PostgreSQL text file.
  • Custom - File that uses user-defined delimiter, escape, and quote characters.
  • Multi Character Delimited - File that uses multiple user-defined characters to delimit fields and lines, and single user-defined escape and quote characters.
JSON
The destination writes records as JSON data. You can use one of the following formats:
  • Array - Each file includes a single array. In the array, each element is a JSON representation of each record.
  • Multiple objects - Each file includes multiple JSON objects. Each object is a JSON representation of a record.
Parquet
The destination writes a Parquet file for each partition and includes the Parquet schema in every file.
Output files use the following naming convention:
part-<multipart partition number>.snappy.parquet
For more information about the Parquet data format, see Parquet Data Format.
Protobuf
Writes a batch of messages in each file.
Uses the user-defined message type and the definition of the message type in the descriptor file to generate the messages in the file.
For information about generating the descriptor file, see Protobuf Data Format Prerequisites.
SDC Record
The destination writes records in the SDC Record data format.
Text
The destination writes data from a single text field to the destination system. When you configure the stage, you select the field to use.
You can configure the characters to use as record separators. By default, the destination uses a UNIX-style line ending (\n) to separate records.
When a record does not contain the selected text field, the destination can report the missing field as an error or to ignore the missing field. By default, the destination reports an error.
When configured to ignore a missing text field, the destination can discard the record or write the record separator characters to create an empty line for the record. By default, the destination discards the record.
Whole File
Streams whole files to the destination system. The destination writes the data to the file and location defined in the stage. If a file of the same name already exists, you can configure the destination to overwrite the existing file or send the current file to error.
By default, written files use the default access permissions for the destination system. You can specify an expression that defines access permissions.
You can configure the destination to generate a checksum for the written file and pass checksum information to the destination system in an event record.
Using this data format requires setting the File Type property to Whole File.
For more information about the whole file data format, see Whole File Data Format.

Writing to Azure Blob Storage

Using the HDP stage library, the Hadoop FS destination can access Azure Blob storage with the WASB protocol. To write to Azure Blob storage, configure the Hadoop FS destination to use the HDP stage library and to connect with the appropriate URI and Azure credentials.

  1. On the General tab, set the Stage Library property to the HDP stage library version 2.4 or later.
  2. On the Connection tab, set the File System URI property as follows:
    <wasb[s]>://<container name>@<storage account name>.blob.core.windows.net

    In the URI, <container name> is the Azure container name, and <storage account name> is the Azure storage account name.

  3. In the Additional Configuration property, enter the Azure credentials:
    1. If necessary, click the Add icon to add an additional configuration property.
    2. In the Name property, enter the Azure storage account.
    3. In the Value property, enter the plain-text key for the storage account.
    Note: You can also use bulk edit mode to add configuration properties.

Writing to Azure Blob Storage with HDInsight

If you installed Data Collector on Azure HDInsight, then you can write to Azure Blob storage with HDInsight.

To write to Azure Blob storage with HDInsight, configure the Hadoop FS destination to use the HDP stage library and to connect with the appropriate Azure credentials and URI.
  1. On the General tab of the Hadoop FS destination, set the Stage Library property to the HDP stage library version 2.4 or later.
  2. Configure Azure credentials in one of the following ways:
    • If the Azure credentials are defined in the HDFS configuration file core-site.xml, configure the destination to access the file.
      1. On the Connection tab, configure the Configuration Files Directory property to point to the directory that includes the file.
    • If the credentials are not defined in the core-site.xml file, configure the destination to pass an additional configuration property that contains the Azure credentials:
      1. On the Connection tab, configure the Additional Configuration property to pass the Azure credentials.

        You can use simple or bulk edit mode to add configuration properties.

      2. If necessary, click the Add icon to add a new additional configuration property.
      3. In the Name property, enter the Azure storage account name as follows:
        fs.azure.account.key.<storage account name>.blob.core.windows.net
        For example, if the storage account name is sdchd, then enter the following name for the property:
        fs.azure.account.key.sdchd.blob.core.windows.net
        Tip: You can find the Azure storage account name on the Access Keys page in the Microsoft Azure portal. To view the page in the Microsoft Azure portal, click All Resources > Storage Account > Access Keys. A page like the following appears, with the storage account name and access keys:

      4. In the Value property, enter an access key value for the Azure storage account. You can use any valid key.
        Tip: The account key value also displays on the Access Keys page. For example, on the image above, you could use either the key1 or key2 value.
  3. On the Connection tab, configure the File System URI property as follows:
    <wasb[s]>://<container name>@<storage account name>.blob.core.windows.net/<path to files>

    In the URI, <container name> is the Azure container name, and <storage account name> is the Azure storage account name.

    For example, for a sdc-hd container in a storage account named sdchd, with all files in a files directory, you would define the file system URI as follows:
    wasbs://sdc-hd@sdchd.blob.core.windows.net/files
    Tip: You can find the container name and storage account name on the Essentials page in the Microsoft Azure portal. For a standard storage account, in the Microsoft Azure portal, click All Resources > Storage Account > Overview > Blobs. For a blob storage account, click All Resources > Storage Account > Overview.

    The page shows the container name and storage account name:

    Though the host name for the file system URI is <storage account name>.blob.core.windows.net, you can alternatively use the host name of the Azure blob service endpoint as the host name for the file system URI.

Example

The following image shows how to configure the Hadoop FS destination to write to Azure Blob storage with HDInsight using the Azure account information in the examples above:

Event Generation

The Hadoop FS destination can generate events that you can use in an event stream. When you enable event generation, the destination generates event records each time the destination closes a file or completes streaming a whole file.

Hadoop FS events can be used in any logical way. For example:

For more information about dataflow triggers and the event framework, see Dataflow Triggers Overview.

Event Records

Hadoop FS event records include the following event-related record header attributes. Record header attributes are stored as String values:

Record Header Attribute Description
sdc.event.type Event type. Uses one of the following types:
  • file-closed - Generated when the destination closes a file.
  • wholeFileProcessed - Generated when the destination completes streaming a whole file.
sdc.event.version Integer that indicates the version of the event record type.
sdc.event.creation_timestamp Epoch timestamp when the stage created the event.
The destination can generate the following types of event records:
File closure
The destination generates a file closure event record when it closes an output file.
File closure event records have the sdc.event.type record header attribute set to file-closed and include the following fields:
Field Description
filepath Absolute path to the closed file.
filename File name of the closed file.
length Size of the closed file in bytes.
Whole file processed
The destination generates an event record when it completes streaming a whole file. Whole file event records have the sdc.event.type record header attribute set to wholeFileProcessed and have the following fields:
Field Description
sourceFileInfo A map of attributes about the original whole file that was processed. The attributes include:
  • size - Size of the whole file in bytes.

Additional attributes depend on the information provided by the origin system.

targetFileInfo A map of attributes about the whole file written to the destination. The attributes include:
  • path - An absolute path to the processed whole file.
checksum Checksum generated for the written file.

Included only when you configure the destination to include checksums in the event record.

checksumAlgorithm Algorithm used to generate the checksum.

Included only when you configure the destination to include checksums in the event record.

Kerberos Authentication

You can use Kerberos authentication to connect to HDFS. When you use Kerberos authentication, Data Collector uses the Kerberos principal and keytab to connect to HDFS. By default, Data Collector uses the user account who started it to connect.

The Kerberos principal and keytab are defined in the Data Collector configuration file, $SDC_CONF/sdc.properties. To use Kerberos authentication, configure all Kerberos properties in the Data Collector configuration file, and then enable Kerberos in the Hadoop FS destination.

For more information about enabling Kerberos authentication for Data Collector, see Kerberos Authentication in the Data Collector documentation.

Impersonation User

Data Collector can either use the currently logged in Data Collector user or a user configured in the destination to write to HDFS.

A Data Collector configuration property can be set that requires using the currently logged in Data Collector user. When this property is not set, you can specify a user in the origin. For more information about Hadoop impersonation and the Data Collector property, see Hadoop Impersonation Mode in the Data Collector documentation.

Note that the destination uses a different user account to connect to HDFS. By default, Data Collector uses the user account who started it to connect to external systems. When using Kerberos, Data Collector uses the Kerberos principal.

To configure a user in the destination to write to HDFS, perform the following tasks:
  1. On Hadoop, configure the user as a proxy user and authorize the user to impersonate a Hadoop user.

    For more information, see the Hadoop documentation.

  2. In the Hadoop FS destination, on the Connection tab, configure the Impersonation User property.

HDFS Properties and Configuration Files

You can configure the Hadoop FS destination to use individual HDFS properties or HDFS configuration files:
HDFS configuration files
You can use the following HDFS configuration files with the Hadoop FS destination:
  • core-site.xml
  • hdfs-site.xml
To use HDFS configuration files:
  1. Store the files or a symlink to the files in the Data Collector resources directory.
  2. In the Hadoop FS destination, configure the Configuration Files Directory property to specify the location of the files.
Note: For a Cloudera Manager installation, Data Collector automatically creates a symlink to the files named hadoop-conf. Enter hadoop-conf for the location of the files in the Hadoop FS destination.
Individual properties
You can configure individual HDFS properties in the destination. To add an HDFS property, you specify the exact property name and the value. The Hadoop FS destination does not validate the property names or values.
Note: Individual properties override properties defined in the HDFS configuration file.

Configuring a Hadoop FS Destination

Configure a Hadoop FS destination to write data to HDFS.

  1. In the Properties panel, on the General tab, configure the following properties:
    General Property Description
    Name Stage name.
    Description Optional description.
    Stage Library Library version that you want to use.
    Produce Events Generates event records when events occur. Use for event handling.
    Required Fields Fields that must include data for the record to be passed into the stage.
    Tip: You might include fields that the stage uses.

    Records that do not include all required fields are processed based on the error handling configured for the pipeline.

    Preconditions Conditions that must evaluate to TRUE to allow a record to enter the stage for processing. Click Add to create additional preconditions.

    Records that do not meet all preconditions are processed based on the error handling configured for the stage.

    On Record Error Error record handling for the stage:
    • Discard - Discards the record.
    • Send to Error - Sends the record to the pipeline for error handling.
    • Stop Pipeline - Stops the pipeline.
  2. On the Connection tab, configure the following properties:
    Connection Property Description
    File System URI Optional HDFS URI.

    When not configured, the stage uses the URI defined by the fs.defaultFS property in the core-site.xml file.

    Impersonation User The HDFS user to impersonate to access HDFS. When using this property, make sure HDFS is configured appropriately.

    When not configured, the pipeline uses the currently logged in Data Collector user.

    Not configurable when Data Collector is configured to use the currently logged in Data Collector user. For more information, see Hadoop Impersonation Mode in the Data Collector documentation.

    Kerberos Authentication Uses Kerberos credentials to connect to HDFS.

    When selected, uses the Kerberos principal and keytab defined in the Data Collector configuration file, $SDC_CONF/sdc.properties.

    Configuration Files Directory Location of the HDFS configuration files.

    For a Cloudera Manager installation, enter hadoop-conf. For all other installations, use a directory or symlink within the Data Collector resources directory.

    You can use the following configuration files:
    • core-site.xml
    • hdfs-site.xml
    Note: Properties in the configuration files are overridden by individual properties defined in the stage.
    Additional Configuration Additional HDFS properties to pass to HDFS. These properties take precedence over properties defined in Hadoop configuration files.

    To add properties, click Add and define the property name and value. Use the property names and values as expected by HDFS.

  3. On the Output Files tab, configure the following options:
    Output Files Property Description
    Directory Template Template for creating output directories. You can use constants, field values, and datetime variables.

    Output directories are created based on the smallest datetime variable in the template.

    File Type Output file type:
    • Text files
    • Sequence files
    • Whole files - Select when using the whole file data format.
    Files Prefix Prefix to use for output files. Use when writing to a directory that receives files from other sources.

    Uses the prefix sdc-${sdc:id()} by default. The prefix evaluates to sdc-<Data Collector ID>.

    The Data Collector ID is stored in the following file: $SDC_DATA/sdc.id. For more information about environment variables, see Data Collector Environment Configuration in the Data Collector documentation.

    Files Suffix Suffix to use for output files, such as txt or json. When used, the destination adds a period and the configured suffix as follows: <filename>.<suffix>.

    You can include periods within the suffix, but do not start the suffix with a period. Forward slashes are not allowed.

    Not available for the whole file data format.

    Directory in Header Indicates that the target directory is defined in record headers. Use only when the targetDirectory header attribute is defined for all records.
    Data Time Zone Time zone for the destination system. Used to resolve datetimes in the directory template and evaluate where records are written.
    Time Basis Time basis to use for creating output directories and writing records to the directories. Use one of the following expressions:
    • ${time:now()} - Uses the processing time as the time basis.
    • ${record:value(<date field path>)} - Uses the time associated with the record as the time basis.
    Max Records in File Maximum number of records written to an output file. Additional records are written to a new file.

    Use 0 to opt out of this property.

    Not available when using the whole file data format.

    Max File Size (MB) Maximum size of an output file. Additional records are written to a new file.

    Use 0 to opt out of this property.

    Not available when using the whole file data format.

    Idle Timeout (secs) Maximum time that an output file can remain idle. After no records are written to a file for this amount of time, the destination closes the file. Enter a time in seconds or use the MINUTES or HOURS constant in an expression to define the time increment.

    Use -1 to set no limit. Default is 1 hour, defined as follows: ${1 * HOURS}.

    Not available when using the whole file data format.

    Compression Codec Compression type for output files:
    • None
    • gzip
    • bzip2
    • Snappy
    • LZ4
    • Other
    Note: Do not use with Avro data. To compress Avro data, use the Avro Compression Codec property on the Data Formats tab.
    Compression Codec Class Full class name of the other compression codec that you want to use.
    Sequence File Key Record key for creating sequence files. Use one of the following options:
    • ${record:value(<field path>)}
    • ${uuid()}
    Compression Type Compression type for sequence files when using a compression codec:
    • Block Compression
    • Record Compression
    Use Roll Attribute Checks the record header for the roll header attribute and closes the current file when the roll attribute exists.

    Can be used with Max Records in a File and Max File Size to close files.

    Roll Attribute Name Name of the roll header attribute.

    Default is roll.

    Validate Permissions When you start the pipeline, the destination tries writing to the configured directory template to validate permissions. The pipeline does not start if validation fails.
    Note: Do not use this option when the directory template uses expressions to represent the entire directory.
    Skip File Recovery Determines whether the destination performs file recovery after an unexpected stop of the pipeline.
  4. On the Late Records tab, configure the following properties:
    Tip: These properties are relevant for a time basis based on the time of a record. If you use processing time as the time basis, set the late record time limit to one second.
    Late Records Property Description
    Late Record Time Limit (secs) Time limit for output directories to accept data.

    You can enter a time in seconds, or use the expression to enter a time in hours. You can also use MINUTES in the default expression to define the time in minutes.

    Late Record Handling Determines how to handle late records:
    • Send to error - Sends the record to the stage for error handling.
    • Send to late records file - Sends the record to a late records file.
    Late Record Directory Template Template for creating late record directories. You can use constants, field values, and datetime variables.

    Output directories are created based on the smallest datetime variable in the template.

  5. On the Data Format tab, configure the following property:
    Data Format Property Description
    Data Format Format of data to be written. Use one of the following options:
  6. For Avro data, on the Data Format tab, configure the following properties:
    Avro Property Description
    Avro Schema Location Location of the Avro schema definition to use when writing data:
    • In Pipeline Configuration - Use the schema that you provide in the stage configuration.
    • In Record Header - Use the schema in the avroSchema record header attribute. Use only when the avroSchema attribute is defined for all records.
    • Confluent Schema Registry - Retrieve the schema from Confluent Schema Registry.
    Avro Schema Avro schema definition used to write the data.

    You can optionally use the runtime:loadResource function to load a schema definition stored in a runtime resource file.

    Register Schema Registers a new Avro schema with Confluent Schema Registry.
    Schema Registry URLs Confluent Schema Registry URLs used to look up the schema or to register a new schema. To add a URL, click Add and then enter the URL in the following format:
    http://<host name>:<port number>
    Basic Auth User Info User information needed to connect to Confluent Schema Registry when using basic authentication.

    Enter the key and secret from the schema.registry.basic.auth.user.info setting in Schema Registry using the following format:

    <key>:<secret>
    Tip: To secure sensitive information such as user names and passwords, you can use runtime resources or credential stores. For more information about credential stores, see Credential Stores in the Data Collector documentation.
    Look Up Schema By Method used to look up the schema in Confluent Schema Registry:
    • Subject - Look up the specified Avro schema subject.
    • Schema ID - Look up the specified Avro schema ID.
    Schema Subject Avro schema subject to look up or to register in Confluent Schema Registry.

    If the specified subject to look up has multiple schema versions, the destination uses the latest schema version for that subject. To use an older version, find the corresponding schema ID, and then set the Look Up Schema By property to Schema ID.

    Schema ID Avro schema ID to look up in Confluent Schema Registry.
    Include Schema Includes the schema in each file.
    Note: Omitting the schema definition can improve performance, but requires the appropriate schema management to avoid losing track of the schema associated with the data.
    Avro Compression Codec The Avro compression type to use.

    When using Avro compression, do not enable other compression available in the destination.

  7. For binary data, on the Data Format tab, configure the following property:
    Binary Property Description
    Binary Field Path Field that contains the binary data.
  8. For delimited data, on the Data Format tab, configure the following properties:
    Delimited Property Description
    Delimiter Format Format for delimited data:
    • Default CSV - File that includes comma-separated values. Ignores empty lines in the file.
    • RFC4180 CSV - Comma-separated file that strictly follows RFC4180 guidelines.
    • MS Excel CSV - Microsoft Excel comma-separated file.
    • MySQL CSV - MySQL comma-separated file.
    • Tab-Separated Values - File that includes tab-separated values.
    • PostgreSQL CSV - PostgreSQL comma-separated file.
    • PostgreSQL Text - PostgreSQL text file.
    • Custom - File that uses user-defined delimiter, escape, and quote characters.
    Header Line Indicates whether to create a header line.
    Delimiter Character Delimiter character for a custom delimiter format. Select one of the available options or use Other to enter a custom character.

    You can enter a Unicode control character using the format \uNNNN, where ​N is a hexadecimal digit from the numbers 0-9 or the letters A-F. For example, enter \u0000 to use the null character as the delimiter or \u2028 to use a line separator as the delimiter.

    Default is the pipe character ( | ).

    Record Separator String Characters to use to separate records. Use any valid Java string literal. For example, when writing to Windows, you might use \r\n to separate records.

    Available when using a custom delimiter format.

    Escape Character Escape character for a custom delimiter format. Select one of the available options or use Other to enter a custom character.

    Default is the backslash character ( \ ).

    Quote Character Quote character for a custom delimiter format. Select one of the available options or use Other to enter a custom character.

    Default is the quotation mark character ( " ).

    Replace New Line Characters Replaces new line characters with the configured string.

    Recommended when writing data as a single line of text.

    New Line Character Replacement String to replace each new line character. For example, enter a space to replace each new line character with a space.

    Leave empty to remove the new line characters.

    Charset Character set to use when writing data.
  9. For JSON data, on the Data Format tab, configure the following property:
    JSON Property Description
    JSON Content Method to write JSON data:
    • JSON Array of Objects - Each file includes a single array. In the array, each element is a JSON representation of each record.
    • Multiple JSON Objects - Each file includes multiple JSON objects. Each object is a JSON representation of a record.
    Charset Character set to use when writing data.
  10. For Parquet data, on the Data Format tab, configure the following properties:
    Parquet Property Description
    Parquet Schema Location Location to load the Parquet schema from:
    • Parquet Schema in Record Header - Uses the schema in the parquetSchema record header attribute.
    • In Pipeline Configuration - Uses the schema you provide in the stage configuration.
    • Infer from records - The destination infers the schema from the first record of the first batch of data.
    Parquet Compression Codec Compression type for output files:
    • Uncompressed
    • GZIP
    • LZO
    • Snappy
    Parquet Format Version Parquet file format version to use.
    Parquet Row Group Size (bytes) Maximum size in bytes of a row group being buffered in memory.

    The size of a row group is defined by the batch size, but you can limit it to restrict the memory usage while writingfor the write.

    Parquet Page Size (bytes) Page size in bytes for compressed pages.
    Parquet Max Padding Size (bytes) Maximum size in bytes of allowed padding to align rows.
    Enable Parquet Dictionary Encoding Enables Parquet dictionary encoding.

    There is one dictionary page per column per group when dictionary encoding is enabled.

    Parquet Dictionary Page Size (bytes) Page size in bytes for dictionary pages.
  11. For protobuf data, on the Data Format tab, configure the following properties:
    Protobuf Property Description
    Protobuf Descriptor File Descriptor file (.desc) to use. The descriptor file must be in the Data Collector resources directory, $SDC_RESOURCES.

    For more information about environment variables, see Data Collector Environment Configuration in the Data Collector documentation. For information about generating the descriptor file, see Protobuf Data Format Prerequisites.

    Message Type Fully-qualified name for the message type to use when writing data.

    Use the following format: <package name>.<message type>.

    Use a message type defined in the descriptor file.
  12. For text data, on the Data Format tab, configure the following properties:
    Text Property Description
    Text Field Path Field that contains the text data to be written. All data must be incorporated into the specified field.
    Record Separator Characters to use to separate records. Use any valid Java string literal. For example, when writing to Windows, you might use \r\n to separate records.

    By default, the destination uses \n.

    On Missing Field When a record does not include the text field, determines whether the destination reports the missing field as an error or ignores the missing field.
    Insert Record Separator if No Text When configured to ignore a missing text field, inserts the configured record separator string to create an empty line.

    When not selected, discards records without the text field.

    Charset Character set to use when writing data.
  13. For whole files, on the Data Format tab, configure the following properties:
    Whole File Property Description
    File Name Expression

    Expression to use for the file names.

    For tips on how to name files based on input file names, see Writing Whole Files.

    Permissions Expression Expression that defines the access permissions for output files. Expressions should evaluate to a symbolic or numeric/octal representation of the permissions you want to use.

    By default, with no specified expression, files use the default permissions of the destination system.

    To use the original source file access permissions, use the following expression:
    ${record:value('/fileInfo/permissions')}
    File Exists Action to take when a file of the same name already exists in the output directory. Use one of the following options:
    • Send to Error - Handles the record based on stage error record handling.
    • Overwrite - Overwrites the existing file.
    Include Checksum in Events Includes checksum information in whole file event records.

    Use only when the destination generates event records.

    Checksum Algorithm Algorithm to generate the checksum.