Installation on z/OS

Java 8+, tested with IBM J9 VM.

Please refer to configuration pages to configure your GridGain nodes. The configuration file provided with GridGain distribution will start an in-memory node with large data region, which should form a single-node cluster after start-up. You may need to tune the size of default data region to adjust resource usage.

Most text files are provided in native EBCDIC 1047 code page. However, some files are required to be in ASCII-derived encoding, notably, configuration XML files. If you need to view or edit any of those, consider tagging them using chtag command, using the following command as reference:

chtag -t -R -c ISO8859-1 path/to/file.xml

Set the MEMLIMIT parameter. The parameter defines the limit on memory for a single process. Set it to a value that is larger than the RAM memory required for your GridGain nodes, including Heap and Off-Heap data regions.

We recommend to always specify -Dfile.encoding=UTF-8 JVM argument. You do not need to specify it if you start GridGain by running ignite.sh. However, if you start nodes from custom code, add it to the startup parameters of your application.

If you are going to use the native persistence feature, set the IGNITE_WAL_MMAP system property (or environment variable) to false, for example, by adding -DIGNITE_WAL_MMAP=false JVM argument. You do not need to specify it if you start GridGain by running ignite.sh.

Configure static discovery instead of the default multicast option. Specify the TcpDiscoverySpi.localAddress property and the IgniteConfiguration.localHost property with the preferred IP address for the current host. You do not need to specify it if you start GridGain by using provided config/default-config.xml.

If you are going to use log4j for logging, you will need to specify charset/encoding of file appenders as IBM-1047 in Log2J/Log4J2 configuration. Please use provided config/ignite-log4j.xml and config/ignite-log4j2.xml as reference.

To run GridGain normally, please execute the following bash command:

bash $IGNITE_HOME/bin/ignite.sh -v $IGNITE_HOME/config/default-config.xml

Note that running ignite.sh blocks the execution flow, so you can also start a node in the nohup mode by executing the command below:

nohup bash $IGNITE_HOME/bin/ignite.sh -v $IGNITE_HOME/config/default-config.xml 1> log.txt 2>&1 &

By default, the GridGain logs are located inside $IGNITE_HOME/work/logs. If you start the node with nohup, you can redirect STDOUT and STDERR to a file as it was done in the command above. If the node starts successfully, you will see the Topology snapshot message indicating that the cluster is created and it contains one server node:

cat <LOG_FILE> |grep "Topology snapshot"

To verify normal operation of the cluster, use the SQLLine tools:

bash $IGNITE_HOME/bin/sqlline.sh --verbose=true -u jdbc:ignite:thin://127.0.0.1/

CREATE TABLE Person (id LONG, name VARCHAR, city_id LONG, PRIMARY KEY (id, city_id))WITH "backups=1, affinityKey=city_id";

INSERT INTO Person (id, name, city_id) VALUES (1, 'John', 1);

SELECT * FROM Person;

You can also check that your Java process was started successfully by using the following command:

ps -ef |grep -i ignite |grep -iv grep

RWPAREC   50335493   83889905  - 15:22:38 ttyp0000  0:00 bash /GridGain/gridgain-zos-enterprise-8.8.5/bin/ignite.sh -v /GridGain/gridga...

For nodes with a persistent storage, increase the dataStorage.checkpointThreads value. The default value is 4, but we recommend you set it to a value between 16 and 32 on z/OS nodes for best checkpointing speed.

If your load profile includes intensive network usage, consider the following TcpCommunicationSpi parameters: socketWriteTimeout=5000 (ms), usePairedConnections=true, pick the connectionsPerNode value from between 2 and 8.

Nodes may benefit from a larger striped pool size, increase it by setting IgniteConfiguration.stripedPoolSize to 32.