Universal Deployer

---

Author	XAP Version	Last Updated	Reference	Download
Shay Hassidim	8.0.1	April 2011

Overview

The GigaSpaces Universal Deployer (GUD) allows deploying the different supported processing units (Space PU, Regular PU, Elastic PU, memCache PU or Web PU via a simple configuration file. The GUD support dependency based deployment allowing multiple processing units to be deployed as one atomic process (e.g. composite application).

Dependency Based Deployment

The GUD allows you to specify for each deployed PU its dependency PU list. The dependent PU will be deployed after all its PUs dependencies will be fully deployed. The mechanism is recursive; if any of these PUs depends on other PU(s), these will be deployed before the top level dependent PU is deployed. This ensures the deploy order will have the entire PUs deployed in the correct order.

To describe the above deploy plan the GUD should have the following config file:

A [B,C] -file myA_PU.jar
B [D,E] -file myB_PU.jar
C [F,G] -file myC_PU.jar
D [] -file myD_PU.jar
E [] -file myE_PU.jar
F [] -file myF_PU.jar
G [] -file myG_PU.jar

With the above config the PUs will be deployed in the following order: F, D, E, G, C, B, A.

Abort Strategy

In case of a deployment failure, you may specify an abort strategy. In case of unsuccessful deployment you may instruct the GUD to undeploy the specific PU that could not be deployed or undeploy the entire PUs that already been deployed.

GUD Command Line Options

The GUD support the following command line arguments:

Switch	Description	Options	Default
-command	deploy or undeploy PU listed at the config file location	deploy , undeploy	deploy
-config	GUD config file location
-locators	lookup service locators
-groups	lookup service groups
-abortDeployOnFailure	In case of a failure with the deploy of any of the PUs, undeploy all the PUs listed at the GUD config file	true,false	false

Running the GUD

Install the GUD

• Download the GigaSpaces Universal Deployer and extract it into an empty folder.
• Edit the runGSUniversalDeployer.bat or runGSUniversalDeployer.sh to include the correct parameters.

Here an example how you should run the GUD:

java com.gigaspaces.admin.GSUniversalDeployer -config c:\\puList.txt -locators 127.0.0.1 -abortDeployOnFailure true

GUD Configuration File

The universal Deployer configuration file may include multiple lines with the following format:

PU Name [Dependent PU List] deploy options

Example:

mySpace [] -type space -cluster schema=partitioned-sync2backup total_members=2,1
myPU [] -type regular -file c:\myPU.zip
A [B,C] -elastic manualCapacity  -user xxx -password yyyy -undeployOnFailure true -file mywar.war
B [D,E] -cluster schema=partitioned total_members=2,1 -user xxx -password yyyy  -file myjar.jar
C [F,E] -elastic eagerCapacity -user xxx -password yyyy -properties embed://prop2=value2;prop3=value3 -file myzip.zip
D [] -cluster total_members=2 -user xxx -password yyyy -properties file://d:/temp/context.properties
E [] -cluster total_members=2 -user xxx -password yyyy -zones zone1,zone2

GUD Configuration File Options

Each GUD configuration file line supports the following options:

Switch	Description	Options	Default
-file	PU file. You may have .zip , .jar , .war file deployed.
-type	PU Type	Regular,Space,memcache	Regular
-timeout	Timeout to abort the deploy process. If the deploy process is not completed within this timeframe, undeploy process will be initiated.		120 seconds
-undeployOnFailure	undeploy the PU in case of a failure or when the deploy could not be completed within the timeout specified	false , true	false
-spaceurl	Used with memcache PU type.	The url of the space, can be embedded, e.g.: /./myMemcached, or remote e.g.: jini://*/*/myMemcached	/./memcached
-sla	Location of an optional xml file holding the SLA element	sla file location
-cluster	Set cluster parameters	schema=partitioned-sync2backup : The cluster schema total_members=1,1 : The number of instances and number of backups
-user xxx -password yyyy	Deploys a secured processing unit propagated with the supplied user and password
-secured	Deploys a secured processing unit (implicit when using -user/-password)	false,true	false
-properties	Location of context level properties file or list of properties. Examples: -properties embed://prop2=value2;prop3=value3 or -properties file://d:/context.properties
-max-instances-per-vm number	Set the SLA number of instances per VM	Numeric value
-max-instances-per-machine number	Set the SLA number of instances per machine	Numeric value
-max-instances-per-zone	Set the SLA number of instances per zone	zone/number
-zones	Set the SLA zone requirements
-elastic	Deploy the PU as an Elastic PU. You should have the ESM started when using this option.	manualCapacity,eagerCapacity
-highlyAvailable	Used with a Stale-full Elastic PU. Start backup instances. Specifies if the space should duplicate each information on two different machines.	true,false	true
-memoryCapacityPerContainer	Used with an Elastic PU. Specifies the the heap size per container (operating system process).	Numeric value with a Memory Unit. Memory units supported: m for MB,g for GB	32m
-maxMemoryCapacity	Used with an Elastic PU. Specifies the estimated maximum total memory capacity used with this processing unit.	Numeric value with a Memory Unit. Memory units supported: m for MB,g for GB	256m
-memoryCapacity	Used with an Elastic PU. Specifies the estimated initial total memory capacity used with this processing unit.	Numeric value with a Memory Unit. Memory units supported: m for MB,g for GB	128m
-singleMachineDeployment	Used with an Elastic PU. Allows deployment of the processing unit on a single machine, by lifting the limitation for primary and backup processing unit instances from the same partition to be deployed on different machines.	false,true	false

Commenting lines

You may use / or # to comment out a line within the configuration file.

Logging

When deploying the GUD will display information about started/terminated containers, started/terminated PU instances.