D-Node failure due to an exception
When working with Data Nodes (D-Nodes), you might experience unexpected technical issues that require detailed investigation. You can debug the issues if the log level for D-Nodes com.pega.dsm.dnode.impl.cassandra.Cassandra was set to DEBUG. For more information, see System Operations.
The most common problems and exceptions for D-Nodes are:
Cassandra starts with corrupted SSTables; the log shows the following information statement:
Failed to start Cassandra. Corrupted SSTables: <CorruptSSTableException>
If the D-Node was added through the Services landing page, the following information statement is displayed for nodes with an exception at startup:
Unable to enable as DNode
The D-Node fails due to an exception; the log shows the following information statement:
Cannot initialize DNode context and start Cassandra, PRPC continuing as non-DNode
Data ownership is incorrect after adding a node. Data ownership is not evenly distributed across the number of nodes; the values do not add up to 100%.
Snappy Compression support is unavailable in Pega® Platform; the log shows the following information statement:
java.lang.NoClassDefFoundError: org.xerial.snappy.Snappy (initialization failure)
Solution - Information statements 1 to 4
- Decommission a node by clicking Decommission in the Action column.
- Stop the node.
- Remove the Cassandra data files. The location of the data files is defined by the data_files_directories parameter in the prconfig.xml file. For more information, see Configuration settings for Pega Platform nodes.
- Restart the node.
- Assign the node to the applicable service.
For more information, see Managing decision data nodes.
Solution - Snappy Compression support is unavailable in the platform
Install Snappy compression library or disable compression through the dnode/yaml/internode_compression setting in the prconfig.xml file. For more information, see Configuration settings for Pega Platform nodes.
Published January 17, 2017 — Updated June 30, 2017