hydra

Class FixedPartitionPrms

java.lang.Object

hydra.BasePrms

hydra.FixedPartitionPrms

public class FixedPartitionPrms
extends BasePrms

A class used to store keys for region configuration settings related to fixed partitioning. The settings are used to create instances of PartitionDescription, and can be referenced from a partition configuration via PartitionPrms.fixedPartitionName.

The number of description instances is gated by names. For other parameters, if fewer values than names are given, the remaining instances will use the last value in the list. See $JTESTS/hydra/hydra.txt for more details.

Unused parameters default to null, except where noted. This uses the product default, except where noted.

Values, fields, and subfields of a parameter can be set to BasePrms.DEFAULT, except where noted. This uses the product default, except where noted.

Values, fields, and subfields of a parameter can use oneof, range, or robing except where noted, but each description created will use a fixed value chosen at test configuration time. Use as a task attribute is illegal.

Field Summary

Fields

Modifier and Type	Field and Description
static java.lang.Long	datastores (int(s)) Number of datastores that will create regions using each fixed partition description.
static java.lang.Long	mappingAlgorithm (Comma-separated String pair(s)) Algorithm used to map primaries and secondaries to nodes for fixed partitions.
static java.lang.Long	names (String(s)) Logical names of the fixed partition descriptions.
static java.lang.Long	partitionBuckets (Comma-separated String(s)) Number of buckets in the fixed partitions for each fixed partition description.
static java.lang.Long	partitionNames (Comma-separated String(s)) Names of fixed partitions for each fixed partition description.

Fields inherited from class hydra.BasePrms

DASH, DEFAULT, NONE

Constructor Summary

Constructors

Constructor and Description
FixedPartitionPrms()

Method Summary

Modifier and Type	Method and Description
static void	main(java.lang.String[] args)

Methods inherited from class hydra.BasePrms

dumpKeys, keyForName, nameForKey, setValues, tab, tasktab

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

names

public static java.lang.Long names

(String(s)) Logical names of the fixed partition descriptions. Each name must be unique. Defaults to null. Not for use with oneof, range, or robing.

datastores

public static java.lang.Long datastores

(int(s)) Number of datastores that will create regions using each fixed partition description. This is a required parameter.

mappingAlgorithm

public static java.lang.Long mappingAlgorithm

(Comma-separated String pair(s)) Algorithm used to map primaries and secondaries to nodes for fixed partitions. An algorithm consists of a classname followed by a method name.

The method must have signature public static List and take three arguments: a String with the region name, the logical FixedPartitionDescription, and an int giving the number of redundant copies. It must return the list of fixed partition attributes for the calling hydra client JVM.

Algorithms can use instances of a hydra blackboard such as FixedPartitionBlackboard to coordinate the mapping across multiple hydra client JVMs. It is up to the user to ensure that all fixed partitions are legally assigned, including redundant copies.

If an algorithm intends to support multiple regions being created from the same logical region description, it can use the region name argument to distinguish invocations of the algorithm on behalf of different regions. If an algorithm intends to support creating the same region in multiple WAN sites, it can use the logical distributed system name obtained through DistributedSystemHelper to distinguish invocations of the algorithm on behalf of different WAN sites.

Defaults to FixedPartitionHelper.assignRoundRobin(java.lang.String, hydra.FixedPartitionDescription, int), which assigns primaries and secondaries in a round robin fashion, while ensuring that no node contains multiple FixedPartitionAttributes for the same partition.

partitionNames

public static java.lang.Long partitionNames

(Comma-separated String(s)) Names of fixed partitions for each fixed partition description. This is a required parameter. EXAMPLE: To configure a logical partition configuration named "quarters" with fixed partitions representing 4 yearly quarters with the default of 1 bucket per quarter, and another one for "halves" with 5 and 10 buckets each, with each configuration mapping partitions to 2 datastores, specify: hydra.FixedPartitionPrms-names = quarters halves; hydra.FixedPartitionPrms-partitionNames = Q1 Q2 Q3 Q4, H1 H2; hydra.FixedPartitionPrms-partitionBuckets = default, 5 10; hydra.FixedPartitionPrms-datastores = 2;

partitionBuckets

public static java.lang.Long partitionBuckets

(Comma-separated String(s)) Number of buckets in the fixed partitions for each fixed partition description. Defaults to 1 bucket per partition. EXAMPLE: To configure a logical fixed partition configuration "quarters" with partitions representing 4 yearly quarters with 10 buckets for Q1, 12 buckets for Q2, and 42 buckets for Q3 and Q4, specify: hydra.PartitionPrms-names = datastore accessor; hydra.PartitionPrms-fixedPartitionNames = quarters none; hydra.FixedPartitionPrms-names = quarters; hydra.FixedPartitionPrms-partitionNames = Q1 Q2 Q3 Q4; hydra.FixedPartitionPrms-partitionBuckets = 10 12 42;

Constructor Detail

FixedPartitionPrms

public FixedPartitionPrms()

Method Detail

main

public static void main(java.lang.String[] args)