Errors

A flow can meet two kinds of failure from the Chunking Connector.

Errors the operation raises happen when read-chunked runs, before any chunk is read. They carry the connector's CHUNKING error types, and a flow handles them by type.

Failures while the chunks are read happen after read-chunked has returned, while foreach or an expression reads the chunks. Mule reports them with its own error types, and a flow handles them with ANY.

Errors the operation raises

These are raised by read-chunked itself, before any chunk is read, so a flow can handle them by type:

Error	When
CHUNKING:INVALID_PARAMETER	A value is not valid. Parent of the three errors below, so handling it catches all of them. Raised itself for an encoding the runtime does not recognise and for a streaming strategy setting that is not positive (maxInMemoryChunks, readers).
CHUNKING:INVALID_CHUNK_SIZE	chunkSize or maxChunkSize is not a positive number of bytes
CHUNKING:INVALID_RECORDS_PER_CHUNK	recordsPerChunk is not positive
CHUNKING:INVALID_DELIMITER	The delimiter is empty. An empty delimiter written as text is rejected when the application loads, so in practice this comes from an expression that gives empty text.
CHUNKING:READ_ERROR	There is no content to read: the content is null

Each message says what was wrong and ends with the value received, for example recordsPerChunk must be a positive number of records (got: 0).

Handling CHUNKING errors

A handler for CHUNKING:INVALID_PARAMETER catches it and its three children:

<try>
    <chunking:read-chunked>
        <chunking:chunking-strategy>
            <chunking:by-delimiter recordsPerChunk="0"/>
        </chunking:chunking-strategy>
    </chunking:read-chunked>
    <error-handler>
        <on-error-continue type="CHUNKING:INVALID_PARAMETER">
            <logger level="ERROR" message="#[error.errorType.identifier ++ ': ' ++ error.description]"/>
        </on-error-continue>
    </error-handler>
</try>

Or handle one type alone. This is the sample application's error-invalid-chunk-size flow, without its HTTP listener and summary:

<try doc:name="Try">
    <chunking:read-chunked doc:name="chunkSize 0">
        <chunking:chunking-strategy>
            <chunking:by-size chunkSize="0"/>
        </chunking:chunking-strategy>
    </chunking:read-chunked>
    <set-variable variableName="caught" value="nothing" doc:name="No error"/>
    <error-handler>
        <on-error-continue type="CHUNKING:INVALID_CHUNK_SIZE" doc:name="On CHUNKING:INVALID_CHUNK_SIZE">
            <set-variable variableName="caught" value="#[error.errorType.namespace ++ ':' ++ error.errorType.identifier]" doc:name="Error type"/>
            <set-variable variableName="description" value="#[error.description]" doc:name="Description"/>
        </on-error-continue>
    </error-handler>
</try>

The sample application has a flow for each of these errors; see Set Up.

Failures while the chunks are read

The content is read while the flow reads the chunks, after read-chunked has returned. A failure then is raised to the flow at once, after the chunks before it, and the content is never reported as having ended early. Mule reports these failures with its own error types:

• inside foreach, MULE:UNKNOWN;
• inside a DataWeave expression that reads the chunks, MULE:EXPRESSION.

Handle failures while reading with ANY

An error handler cannot name MULE:UNKNOWN: an application that does fails to build. Handle these failures with ANY, as in the example below, and use error.description to tell them apart.

The error's description ends with the connector's message, which says what happened. The failures are:

• the content cannot be read, for example the source stream fails part-way;
• a record is larger than maxChunkSize (by delimiter; see A record larger than the maximum);
• one of the three sliding-window failures under Failures and what to change;
• in-memory reaches its limit;
• file-store cannot write or read its temporary file;
• the chunks are read after they were released (see When the chunks are released).

<chunking:read-chunked>
    <chunking:chunking-strategy>
        <chunking:by-delimiter recordsPerChunk="1000" maxChunkSize="65536"/>
    </chunking:chunking-strategy>
</chunking:read-chunked>
<try>
    <foreach>
        <logger level="INFO" message="#['Chunk $(payload.index) holds $(payload.recordCount) records']"/>
    </foreach>
    <error-handler>
        <on-error-continue type="ANY">
            <logger level="ERROR" message="#['Reading the chunks failed: ' ++ error.description]"/>
        </on-error-continue>
    </error-handler>
</try>

The sample application's error-failure-while-reading flow shows this with a record larger than maxChunkSize.

Logging

Failures a flow must handle are raised as errors and are not logged. What the connector logs is described under Logging.