PXF - Invalid chunk header error when querying from Oracle external table

Products

• VMware Tanzu Greenplum

Issue/Introduction

Receiving the following error when selecting from an external table located in the Oracle database:

Sep 24, 2020 11:23:30 AM org.apache.catalina.core.StandardWrapperValve invoke
SEVERE: Servlet.service() for servlet [PXF REST Service] in context with path [/pxf] threw exception [javax.servlet.ServletException: java.io.IOException: Invalid chunk header] with root cause
java.io.IOException: Invalid chunk header
    at org.apache.coyote.http11.filters.ChunkedInputFilter.throwIOException(ChunkedInputFilter.java:619)
    at org.apache.coyote.http11.filters.ChunkedInputFilter.doRead(ChunkedInputFilter.java:192)
    at org.apache.coyote.http11.AbstractInputBuffer.doRead(AbstractInputBuffer.java:341)
    at org.apache.coyote.Request.doRead(Request.java:431)
    at org.apache.catalina.connector.InputBuffer.realReadBytes(InputBuffer.java:290)
    at org.apache.tomcat.util.buf.ByteChunk.substract(ByteChunk.java:390)
    at org.apache.catalina.connector.InputBuffer.readByte(InputBuffer.java:304)
    at org.apache.catalina.connector.CoyoteInputStream.read(CoyoteInputStream.java:106)
    at java.io.DataInputStream.readInt(DataInputStream.java:387)
    at org.greenplum.pxf.api.io.GPDBWritable.readPktLen(GPDBWritable.java:158)

The table definition is using the JDBC profile and Oracle Driver:

PROFILE=JDBC&JDBC_DRIVER=oracle.jdbc.driver.OracleDriver&DB_URL=jdbc:oracle

Environment

Product Version: 5.22

Resolution

The connection pool is throwing an error because one of the processes cannot acquire a connection to the Oracle database. These properties can be tuned in the jdbc-site.xml file.  For that, the first requirement is to have a server configuration for PXF. Create the server configuration for oracle, as an example, we use “oracle” as the server name:

1. mkdir -p $PXF_CONF/servers/oracle
2. cp $PXF_CONF/templates/jdbc-site.xml $PXF_CONF/servers/oracle

Then edit the file $PXF_CONF/servers/oracle/jdbc-site.xml, set the jdbc.driver, jdbc.url, jdbc.user, and jdbc.password properties.

Next, increase the jdbc.pool.property.maximumPoolSize property to be number of segments per host +2 connections. So if you have 8 segments per host, set the jdbc.pool.property.maximumPoolSize property to 10.

Finally, the external table will need to be recreated specifying the above server configuration.