com.mesosphere.sdk.curator

Class CuratorUtils

java.lang.Object

com.mesosphere.sdk.curator.CuratorUtils

public class CuratorUtils
extends java.lang.Object

A set of common utilites for managing Curator/Zookeeper paths and data.

Method Summary

Modifier and Type	Method and Description
(package private) static RetryPolicy	getDefaultRetry() Returns a reasonable default retry policy for querying ZK.
static java.lang.String	getServiceRootPath(java.lang.String frameworkName) Returns the root node to store all scheduler ZK data inside.
(package private) static void	initServiceName(Persister persister, java.lang.String serviceName) Compares the service name to the previously stored name in zookeeper, or creates a new node containing this data if it isn't already present.

Methods inherited from class java.lang.Object

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

Method Detail

getServiceRootPath

public static java.lang.String getServiceRootPath(java.lang.String frameworkName)

Returns the root node to store all scheduler ZK data inside. For example:

• "your-name-here" => /dcos-service-your-name-here
• "/path/to/your-name-here" => /dcos-service-path__to__your-name-here

getDefaultRetry

static RetryPolicy getDefaultRetry()

Returns a reasonable default retry policy for querying ZK.

initServiceName

static void initServiceName(Persister persister,
                            java.lang.String serviceName)

Compares the service name to the previously stored name in zookeeper, or creates a new node containing this data if it isn't already present. This is useful for two situations where foldered service names may be confused with literal double-underscore delimited names:

• Protecting against collisions if someone uses periods in their service names. For example, a foldered service named "/myteam/kafka" could collide with a non-foldered service named "myteam.kafka" as the prior is converted to the latter where ZK is concerned.
• Allowing deterministic mapping of a given ZK node back to the originating service. For example, the data in "dcos-service-myteam.kafka" could be for a service named "myteam.kafka", or for a foldered service named "/myteam/kafka". Storing the name explicitly allows potential tooling to tell which one it is. This isn't currently needed but could be useful someday.

Parameters:

persister - the persister where the service name should be written

serviceName - the service name to check for equality, or to write if no service name node exists

See Also:

getServiceRootPath(String)