http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/markdown/tutorial_user_def_rebalancer.md ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/markdown/tutorial_user_def_rebalancer.md b/website/0.6.8/src/site/markdown/tutorial_user_def_rebalancer.md new file mode 100644 index 0000000..2149739 --- /dev/null +++ b/website/0.6.8/src/site/markdown/tutorial_user_def_rebalancer.md @@ -0,0 +1,172 @@ +<!--- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. +--> + +<head> + <title>Tutorial - User-Defined Rebalancing</title> +</head> + +## [Helix Tutorial](./Tutorial.html): User-Defined Rebalancing + +Even though Helix can compute both the location and the state of replicas internally using a default fully-automatic rebalancer, specific applications may require rebalancing strategies that optimize for different requirements. Thus, Helix allows applications to plug in arbitrary rebalancer algorithms that implement a provided interface. One of the main design goals of Helix is to provide maximum flexibility to any distributed application. Thus, it allows applications to fully implement the rebalancer, which is the core constraint solver in the system, if the application developer so chooses. + +Whenever the state of the cluster changes, as is the case when participants join or leave the cluster, Helix automatically calls the rebalancer to compute a new mapping of all the replicas in the resource. When using a pluggable rebalancer, the only required step is to register it with Helix. Subsequently, no additional bootstrapping steps are necessary. Helix uses reflection to look up and load the class dynamically at runtime. As a result, it is also technically possible to change the rebalancing strategy used at any time. + +The Rebalancer interface is as follows: + +``` +void init(HelixManager manager); + +IdealState computeNewIdealState(String resourceName, IdealState currentIdealState, + final CurrentStateOutput currentStateOutput, final ClusterDataCache clusterData); +``` +The first parameter is the resource to rebalance, the second is pre-existing ideal mappings, the third is a snapshot of the actual placements and state assignments, and the fourth is a full cache of all of the cluster data available to Helix. Internally, Helix implements the same interface for its own rebalancing routines, so a user-defined rebalancer will be cognizant of the same information about the cluster as an internal implementation. Helix strives to provide applications the ability to implement algorithms that may require a large portion of the entire state of the cluster to make the best placement and state assignment decisions possible. + +An IdealState is a full representation of the location of each replica of each partition of a given resource. This is a simple representation of the placement that the algorithm believes is the best possible. If the placement meets all defined constraints, this is what will become the actual state of the distributed system. + +### Specifying a Rebalancer +For implementations that set up the cluster through existing code, the following HelixAdmin calls will update the Rebalancer class: + +``` +IdealState idealState = helixAdmin.getResourceIdealState(clusterName, resourceName); +idealState.setRebalanceMode(RebalanceMode.USER_DEFINED); +idealState.setRebalancerClassName(className); +helixAdmin.setResourceIdealState(clusterName, resourceName, idealState); +``` + +There are two key fields to set to specify that a pluggable rebalancer should be used. First, the rebalance mode should be set to USER_DEFINED, and second the rebalancer class name should be set to a class that implements Rebalancer and is within the scope of the project. The class name is a fully-qualified class name consisting of its package and its name. Without specification of the USER_DEFINED mode, the user-defined rebalancer class will not be used even if specified. Furthermore, Helix will not attempt to rebalance the resources through its standard routines if its mode is USER_DEFINED, regardless of whether or not a rebalancer class is registered. + +### Example + +In the next release (0.7.0), we will provide a full recipe of a user-defined rebalancer in action. + +Consider the case where partitions are locks in a lock manager and 6 locks are to be distributed evenly to a set of participants, and only one participant can hold each lock. We can define a rebalancing algorithm that simply takes the modulus of the lock number and the number of participants to evenly distribute the locks across participants. Helix allows capping the number of partitions a participant can accept, but since locks are lightweight, we do not need to define a restriction in this case. The following is a succinct implementation of this algorithm. + +``` +@Override +IdealState computeNewIdealState(String resourceName, IdealState currentIdealState, + final CurrentStateOutput currentStateOutput, final ClusterDataCache clusterData) { + // Get the list of live participants in the cluster + List<String> liveParticipants = new ArrayList<String>(clusterData.getLiveInstances().keySet()); + + // Count the number of participants allowed to lock each lock (in this example, this is 1) + int lockHolders = Integer.parseInt(currentIdealState.getReplicas()); + + // Fairly assign the lock state to the participants using a simple mod-based sequential + // assignment. For instance, if each lock can be held by 3 participants, lock 0 would be held + // by participants (0, 1, 2), lock 1 would be held by (1, 2, 3), and so on, wrapping around the + // number of participants as necessary. + int i = 0; + for (String partition : currentIdealState.getPartitionSet()) { + List<String> preferenceList = new ArrayList<String>(); + for (int j = i; j < i + lockHolders; j++) { + int participantIndex = j % liveParticipants.size(); + String participant = liveParticipants.get(participantIndex); + // enforce that a participant can only have one instance of a given lock + if (!preferenceList.contains(participant)) { + preferenceList.add(participant); + } + } + currentIdealState.setPreferenceList(partition, preferenceList); + i++; + } + return assignment; +} +``` + +Here are the IdealState preference lists emitted by the user-defined rebalancer for a 3-participant system whenever there is a change to the set of participants. + +* Participant_A joins + +``` +{ + "lock_0": ["Participant_A"], + "lock_1": ["Participant_A"], + "lock_2": ["Participant_A"], + "lock_3": ["Participant_A"], + "lock_4": ["Participant_A"], + "lock_5": ["Participant_A"], +} +``` + +A preference list is a mapping for each resource of partition to the participants serving each replica. The state model is a simple LOCKED/RELEASED model, so participant A holds all lock partitions in the LOCKED state. + +* Participant_B joins + +``` +{ + "lock_0": ["Participant_A"], + "lock_1": ["Participant_B"], + "lock_2": ["Participant_A"], + "lock_3": ["Participant_B"], + "lock_4": ["Participant_A"], + "lock_5": ["Participant_B"], +} +``` + +Now that there are two participants, the simple mod-based function assigns every other lock to the second participant. On any system change, the rebalancer is invoked so that the application can define how to redistribute its resources. + +* Participant_C joins (steady state) + +``` +{ + "lock_0": ["Participant_A"], + "lock_1": ["Participant_B"], + "lock_2": ["Participant_C"], + "lock_3": ["Participant_A"], + "lock_4": ["Participant_B"], + "lock_5": ["Participant_C"], +} +``` + +This is the steady state of the system. Notice that four of the six locks now have a different owner. That is because of the naïve modulus-based assignmemt approach used by the user-defined rebalancer. However, the interface is flexible enough to allow you to employ consistent hashing or any other scheme if minimal movement is a system requirement. + +* Participant_B fails + +``` +{ + "lock_0": ["Participant_A"], + "lock_1": ["Participant_C"], + "lock_2": ["Participant_A"], + "lock_3": ["Participant_C"], + "lock_4": ["Participant_A"], + "lock_5": ["Participant_C"], +} +``` + +On any node failure, as in the case of node addition, the rebalancer is invoked automatically so that it can generate a new mapping as a response to the change. Helix ensures that the Rebalancer has the opportunity to reassign locks as required by the application. + +* Participant_B (or the replacement for the original Participant_B) rejoins + +``` +{ + "lock_0": ["Participant_A"], + "lock_1": ["Participant_B"], + "lock_2": ["Participant_C"], + "lock_3": ["Participant_A"], + "lock_4": ["Participant_B"], + "lock_5": ["Participant_C"], +} +``` + +The rebalancer was invoked once again and the resulting IdealState preference lists reflect the steady state. + +### Caveats +- The rebalancer class must be available at runtime, or else Helix will not attempt to rebalance at all +- The Helix controller will only take into account the preference lists in the new IdealState for this release. In 0.7.0, Helix rebalancers will be able to compute the full resource assignment, including the states. +- Helix does not currently persist the new IdealState computed by the user-defined rebalancer. However, the Helix property store is available for saving any computed state. In 0.7.0, Helix will persist the result of running the rebalancer.
http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/markdown/tutorial_yaml.md ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/markdown/tutorial_yaml.md b/website/0.6.8/src/site/markdown/tutorial_yaml.md new file mode 100644 index 0000000..1e4772e --- /dev/null +++ b/website/0.6.8/src/site/markdown/tutorial_yaml.md @@ -0,0 +1,102 @@ +<!--- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. +--> + +<head> + <title>Tutorial - YAML Cluster Setup</title> +</head> + +## [Helix Tutorial](./Tutorial.html): YAML Cluster Setup + +As an alternative to using Helix Admin to set up the cluster, its resources, constraints, and the state model, Helix supports bootstrapping a cluster configuration based on a YAML file. Below is an annotated example of such a file for a simple distributed lock manager where a lock can only be LOCKED or RELEASED, and each lock only allows a single participant to hold it in the LOCKED state. + +``` +clusterName: lock-manager-custom-rebalancer # unique name for the cluster (required) +resources: + - name: lock-group # unique resource name (required) + rebalancer: # required + mode: USER_DEFINED # required - USER_DEFINED means we will provide our own rebalancer + class: org.apache.helix.userdefinedrebalancer.LockManagerRebalancer # required for USER_DEFINED + partitions: + count: 12 # number of partitions for the resource (default is 1) + replicas: 1 # number of replicas per partition (default is 1) + stateModel: + name: lock-unlock # model name (required) + states: [LOCKED, RELEASED, DROPPED] # the list of possible states (required if model not built-in) + transitions: # the list of possible transitions (required if model not built-in) + - name: Unlock + from: LOCKED + to: RELEASED + - name: Lock + from: RELEASED + to: LOCKED + - name: DropLock + from: LOCKED + to: DROPPED + - name: DropUnlock + from: RELEASED + to: DROPPED + - name: Undrop + from: DROPPED + to: RELEASED + initialState: RELEASED # (required if model not built-in) + constraints: + state: + counts: # maximum number of replicas of a partition that can be in each state (required if model not built-in) + - name: LOCKED + count: "1" + - name: RELEASED + count: "-1" + - name: DROPPED + count: "-1" + priorityList: [LOCKED, RELEASED, DROPPED] # states in order of priority (all priorities equal if not specified) + transition: # transitions priority to enforce order that transitions occur + priorityList: [Unlock, Lock, Undrop, DropUnlock, DropLock] # all priorities equal if not specified +participants: # list of nodes that can serve replicas (optional if dynamic joining is active, required otherwise) + - name: localhost_12001 + host: localhost + port: 12001 + - name: localhost_12002 + host: localhost + port: 12002 + - name: localhost_12003 + host: localhost + port: 12003 +``` + +Using a file like the one above, the cluster can be set up either with the command line: + +``` +helix/helix-core/target/helix-core/pkg/bin/YAMLClusterSetup.sh localhost:2199 lock-manager-config.yaml +``` + +or with code: + +``` +YAMLClusterSetup setup = new YAMLClusterSetup(zkAddress); +InputStream input = + Thread.currentThread().getContextClassLoader() + .getResourceAsStream("lock-manager-config.yaml"); +YAMLClusterSetup.YAMLClusterConfig config = setup.setupCluster(input); +``` + +Some notes: + +- A rebalancer class is only required for the USER_DEFINED mode. It is ignored otherwise. + +- Built-in state models, like OnlineOffline, LeaderStandby, and MasterSlave, or state models that have already been added only require a name for stateModel. If partition and/or replica counts are not provided, a value of 1 is assumed. http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/resources/.htaccess ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/resources/.htaccess b/website/0.6.8/src/site/resources/.htaccess new file mode 100644 index 0000000..d5c7bf3 --- /dev/null +++ b/website/0.6.8/src/site/resources/.htaccess @@ -0,0 +1,20 @@ +# +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +# + +Redirect /download.html /download.cgi http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/resources/download.cgi ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/resources/download.cgi b/website/0.6.8/src/site/resources/download.cgi new file mode 100755 index 0000000..f9a0e30 --- /dev/null +++ b/website/0.6.8/src/site/resources/download.cgi @@ -0,0 +1,22 @@ +#!/bin/sh +# Just call the standard mirrors.cgi script. It will use download.html +# as the input template. +# +# Licensed to the Apache Software Foundation (ASF) under one +# or more contributor license agreements. See the NOTICE file +# distributed with this work for additional information +# regarding copyright ownership. The ASF licenses this file +# to you under the Apache License, Version 2.0 (the +# "License"); you may not use this file except in compliance +# with the License. You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, +# software distributed under the License is distributed on an +# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +# KIND, either express or implied. See the License for the +# specific language governing permissions and limitations +# under the License. +# +exec /www/www.apache.org/dyn/mirrors/mirrors.cgi $* http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/resources/images/PFS-Generic.png ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/resources/images/PFS-Generic.png b/website/0.6.8/src/site/resources/images/PFS-Generic.png new file mode 100644 index 0000000..7eea3a0 Binary files /dev/null and b/website/0.6.8/src/site/resources/images/PFS-Generic.png differ http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/resources/images/RSYNC_BASED_PFS.png ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/resources/images/RSYNC_BASED_PFS.png b/website/0.6.8/src/site/resources/images/RSYNC_BASED_PFS.png new file mode 100644 index 0000000..0cc55ae Binary files /dev/null and b/website/0.6.8/src/site/resources/images/RSYNC_BASED_PFS.png differ http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/resources/images/TaskFrameworkLayers.png ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/resources/images/TaskFrameworkLayers.png b/website/0.6.8/src/site/resources/images/TaskFrameworkLayers.png new file mode 100644 index 0000000..4ee24a8 Binary files /dev/null and b/website/0.6.8/src/site/resources/images/TaskFrameworkLayers.png differ http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/site.xml ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/site.xml b/website/0.6.8/src/site/site.xml new file mode 100644 index 0000000..c0e9013 --- /dev/null +++ b/website/0.6.8/src/site/site.xml @@ -0,0 +1,140 @@ +<?xml version="1.0" encoding="ISO-8859-1"?> +<!-- + Licensed to the Apache Software Foundation (ASF) under one or more + contributor license agreements. See the NOTICE file distributed with + this work for additional information regarding copyright ownership. + The ASF licenses this file to You under the Apache License, Version 2.0 + (the "License"); you may not use this file except in compliance with + the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + + Unless required by applicable law or agreed to in writing, software + distributed under the License is distributed on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + See the License for the specific language governing permissions and + limitations under the License. +--> +<project name="Apache Helix"> + <bannerLeft> + <src>../images/helix-logo.jpg</src> + <href>http://helix.apache.org/</href> + </bannerLeft> + <bannerRight> + <src>../images/feather_small.gif</src> + <href>http://www.apache.org/</href> + </bannerRight> + <version position="none"/> + + <publishDate position="right"/> + + <skin> + <groupId>lt.velykis.maven.skins</groupId> + <artifactId>reflow-maven-skin</artifactId> + <version>1.0.0</version> + </skin> + + <body> + + <head> + <script type="text/javascript"> + + var _gaq = _gaq || []; + _gaq.push(['_setAccount', 'UA-3211522-12']); + _gaq.push(['_trackPageview']); + + (function() { + var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; + ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; + var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); + })(); + + </script> + + </head> + + <breadcrumbs position="left"> + <item name="Apache Helix" href="http://helix.apache.org/"/> + <item name="Release 0.6.8" href="http://helix.apache.org/0.6.8-docs/"/> + </breadcrumbs> + + <links> + <item name="Helix 0.6.8" href="./index.html"/> + </links> + + <menu name="Get Helix"> + <item name="Download" href="./download.html"/> + <item name="Building" href="./Building.html"/> + <item name="Release Notes" href="./releasenotes/release-0.6.8.html"/> + </menu> + + <menu name="Hands-On"> + <item name="Quick Start" href="./Quickstart.html"/> + <item name="Tutorial" href="./Tutorial.html"/> + <item name="Javadocs" href="http://helix.apache.org/javadocs/0.6.8"/> + </menu> + + <menu name="Recipes"> + <item name="Distributed lock manager" href="./recipes/lock_manager.html"/> + <item name="Rabbit MQ consumer group" href="./recipes/rabbitmq_consumer_group.html"/> + <item name="Rsync replicated file store" href="./recipes/rsync_replicated_file_store.html"/> + <item name="Service discovery" href="./recipes/service_discovery.html"/> + <item name="Distributed task DAG execution" href="./recipes/task_dag_execution.html"/> + </menu> +<!-- + <menu ref="reports" inherit="bottom"/> + <menu ref="modules" inherit="bottom"/> + + + <menu name="ASF"> + <item name="How Apache Works" href="http://www.apache.org/foundation/how-it-works.html"/> + <item name="Foundation" href="http://www.apache.org/foundation/"/> + <item name="Sponsoring Apache" href="http://www.apache.org/foundation/sponsorship.html"/> + <item name="Thanks" href="http://www.apache.org/foundation/thanks.html"/> + </menu> +--> + <footer> + <div class="row span16"><div>Apache Helix, Apache, the Apache feather logo, and the Apache Helix project logos are trademarks of The Apache Software Foundation. + All other marks mentioned may be trademarks or registered trademarks of their respective owners.</div> + <a href="${project.url}/privacy-policy.html">Privacy Policy</a> + </div> + </footer> + + + </body> + + <custom> + <reflowSkin> + <theme>default</theme> + <highlightJs>false</highlightJs> + <brand> + <name>Apache Helix</name> + <href>http://helix.apache.org</href> + </brand> + <slogan>A cluster management framework for partitioned and replicated distributed resources</slogan> + <bottomNav> + <column>Get Helix</column> + <column>Hands-On</column> + <column>Recipes</column> + </bottomNav> + <pages> + <index> + <sections> + <columns>3</columns> + </sections> + </index> + </pages> + </reflowSkin> + <!--fluidoSkin> + <topBarEnabled>true</topBarEnabled> + <sideBarEnabled>true</sideBarEnabled> + <googleSearch></googleSearch> + <twitter> + <user>ApacheHelix</user> + <showUser>true</showUser> + <showFollowers>false</showFollowers> + </twitter> + </fluidoSkin--> + </custom> + +</project> http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/site/xdoc/download.xml.vm ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/site/xdoc/download.xml.vm b/website/0.6.8/src/site/xdoc/download.xml.vm new file mode 100644 index 0000000..1e8d0a7 --- /dev/null +++ b/website/0.6.8/src/site/xdoc/download.xml.vm @@ -0,0 +1,214 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!-- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. + +--> +#set( $releaseName = "0.6.8" ) +#set( $releaseDate = "01/20/2017" ) +<document xmlns="http://maven.apache.org/XDOC/2.0"; xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"; + xsi:schemaLocation="http://maven.apache.org/XDOC/2.0 http://maven.apache.org/xsd/xdoc-2.0.xsd";> + + <properties> + <title>Apache Helix Downloads</title> + <author email="[email protected]">Apache Helix Documentation Team</author> + </properties> + + <body> + <div class="toc_container"> + <macro name="toc"> + <param name="class" value="toc"/> + </macro> + </div> + + <section name="Introduction"> + <p>Apache Helix artifacts are distributed in source and binary form under the terms of the + <a href="http://www.apache.org/licenses/LICENSE-2.0";>Apache License, Version 2.0</a>. + See the included <tt>LICENSE</tt> and <tt>NOTICE</tt> files included in each artifact for additional license + information. + </p> + <p>Use the links below to download a source distribution of Apache Helix. + It is good practice to <a href="#Verifying_Releases">verify the integrity</a> of the distribution files.</p> + </section> + + <section name="Release"> + <p>Release date: ${releaseDate} </p> + <p><a href="releasenotes/release-${releaseName}.html">${releaseName} Release notes</a></p> + <a name="mirror"/> + <subsection name="Mirror"> + + <p> + [if-any logo] + <a href="[link]"> + <img align="right" src="[logo]" border="0" + alt="logo"/> + </a> + [end] + The currently selected mirror is + <b>[preferred]</b>. + If you encounter a problem with this mirror, + please select another mirror. + If all mirrors are failing, there are + <i>backup</i> + mirrors + (at the end of the mirrors list) that should be available. + </p> + + <form action="[location]" method="get" id="SelectMirror" class="form-inline"> + Other mirrors: + <select name="Preferred" class="input-xlarge"> + [if-any http] + [for http] + <option value="[http]">[http]</option> + [end] + [end] + [if-any ftp] + [for ftp] + <option value="[ftp]">[ftp]</option> + [end] + [end] + [if-any backup] + [for backup] + <option value="[backup]">[backup] (backup)</option> + [end] + [end] + </select> + <input type="submit" value="Change" class="btn"/> + </form> + + <p> + You may also consult the + <a href="http://www.apache.org/mirrors/";>complete list of mirrors.</a> + </p> + + </subsection> + <subsection name="${releaseName} Sources"> + <table> + <thead> + <tr> + <th>Artifact</th> + <th>Signatures</th> + <th>Hashes</th> + </tr> + </thead> + <tbody> + <tr> + <td> + <a href="[preferred]helix/${releaseName}/src/helix-${releaseName}-src.zip">helix-${releaseName}-src.zip</a> + </td> + <td> + <a href="http://www.apache.org/dist/helix/${releaseName}/src/helix-${releaseName}-src.zip.asc";>asc</a> + </td> + <td> + <a href="http://www.apache.org/dist/helix/${releaseName}/src/helix-${releaseName}-src.zip.md5";>md5</a> + <a href="http://www.apache.org/dist/helix/${releaseName}/src/helix-${releaseName}-src.zip.sha1";>sha1</a> + </td> + </tr> + </tbody> + </table> + </subsection> + <subsection name="${releaseName} Binaries"> + <table> + <thead> + <tr> + <th>Artifact</th> + <th>Signatures</th> + <th>Hashes</th> + </tr> + </thead> + <tbody> + <tr> + <td> + <a href="[preferred]helix/${releaseName}/binaries/helix-core-${releaseName}-pkg.tar">helix-core-${releaseName}-pkg.tar</a> + </td> + <td> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-core-${releaseName}-pkg.tar.asc";>asc</a> + </td> + <td> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-core-${releaseName}-pkg.tar.md5";>md5</a> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-core-${releaseName}-pkg.tar.sha1";>sha1</a> + </td> + </tr> + <tr> + <td> + <a href="[preferred]helix/${releaseName}/binaries/helix-admin-webapp-${releaseName}-pkg.tar">helix-admin-webapp-${releaseName}-pkg.tar</a> + </td> + <td> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-admin-webapp-${releaseName}-pkg.tar.asc";>asc</a> + </td> + <td> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-admin-webapp-${releaseName}-pkg.tar.md5";>md5</a> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-admin-webapp-${releaseName}-pkg.tar.sha1";>sha1</a> + </td> + </tr> + <tr> + <td> + <a href="[preferred]helix/${releaseName}/binaries/helix-agent-${releaseName}-pkg.tar">helix-agent-${releaseName}-pkg.tar</a> + </td> + <td> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-agent-${releaseName}-pkg.tar.asc";>asc</a> + </td> + <td> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-agent-${releaseName}-pkg.tar.md5";>md5</a> + <a href="http://www.apache.org/dist/helix/${releaseName}/binaries/helix-agent-${releaseName}-pkg.tar.sha1";>sha1</a> + </td> + </tr> + </tbody> + </table> + </subsection> + </section> + +<!-- <section name="Older Releases"> + </section>--> + + <section name="Verifying Releases"> + <p>It is essential that you verify the integrity of the downloaded file using the PGP signature (<tt>.asc</tt> file) or a hash (<tt>.md5</tt> or <tt>.sha1</tt> file). Please read <a href="http://www.apache.org/info/verification.html";>Verifying Apache Software Foundation Releases</a> for more information on why you should verify our releases.</p> + <p>The PGP signature can be verified using <a href="http://www.pgpi.org/";>PGP</a> or <a href="http://www.gnupg.org/";>GPG</a>. First download the <a href="http://www.apache.org/dist/helix/KEYS";>KEYS</a> as well as the <tt>*.asc</tt> signature files for the relevant distribution. Make sure you get these files from the main distribution site, rather than from a mirror. Then verify the signatures using one of the following sets of commands: + + <source>% pgpk -a KEYS +% pgpv downloaded_file.asc</source> + + or<br/> + + <source>% pgp -ka KEYS +% pgp downloaded_file.asc</source> + + or<br/> + + <source>% gpg --import KEYS +% gpg --verify downloaded_file.asc</source> + </p> + <p>Alternatively, you can verify the MD5 signature on the files. A Unix/Linux program called + <code>md5</code> or + <code>md5sum</code> is included in most distributions. It is also available as part of + <a href="http://www.gnu.org/software/textutils/textutils.html";>GNU Textutils</a>. + Windows users can get binary md5 programs from these (and likely other) places: + <ul> + <li> + <a href="http://www.md5summer.org/";>http://www.md5summer.org/</a> + </li> + <li> + <a href="http://www.fourmilab.ch/md5/";>http://www.fourmilab.ch/md5/</a> + </li> + <li> + <a href="http://www.pc-tools.net/win32/md5sums/";>http://www.pc-tools.net/win32/md5sums/</a> + </li> + </ul> + </p> + </section> + </body> +</document> http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.6.8/src/test/conf/testng.xml ---------------------------------------------------------------------- diff --git a/website/0.6.8/src/test/conf/testng.xml b/website/0.6.8/src/test/conf/testng.xml new file mode 100644 index 0000000..58f0803 --- /dev/null +++ b/website/0.6.8/src/test/conf/testng.xml @@ -0,0 +1,27 @@ +<?xml version="1.0" encoding="UTF-8"?> +<!-- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. +--> +<!DOCTYPE suite SYSTEM "http://testng.org/testng-1.0.dtd";> +<suite name="Suite" parallel="none"> + <test name="Test" preserve-order="false"> + <packages> + <package name="org.apache.helix"/> + </packages> + </test> +</suite> http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.7.0-incubating/pom.xml ---------------------------------------------------------------------- diff --git a/website/0.7.0-incubating/pom.xml b/website/0.7.0-incubating/pom.xml new file mode 100644 index 0000000..341aeba --- /dev/null +++ b/website/0.7.0-incubating/pom.xml @@ -0,0 +1,51 @@ +<?xml version="1.0" encoding="UTF-8" ?> +<!-- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. +--> +<project xmlns="http://maven.apache.org/POM/4.0.0"; xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"; xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/maven-v4_0_0.xsd";> + <modelVersion>4.0.0</modelVersion> + + <parent> + <groupId>org.apache.helix</groupId> + <artifactId>website</artifactId> + <version>0.7.2-SNAPSHOT</version> + </parent> + + <artifactId>0.7.0-incubating-docs</artifactId> + <packaging>bundle</packaging> + <name>Apache Helix :: Website :: 0.7.0-incubating</name> + + <properties> + </properties> + + <dependencies> + <dependency> + <groupId>org.testng</groupId> + <artifactId>testng</artifactId> + <version>6.0.1</version> + </dependency> + </dependencies> + <build> + <pluginManagement> + <plugins> + </plugins> + </pluginManagement> + <plugins> + </plugins> + </build> +</project> http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.7.0-incubating/src/site/apt/privacy-policy.apt ---------------------------------------------------------------------- diff --git a/website/0.7.0-incubating/src/site/apt/privacy-policy.apt b/website/0.7.0-incubating/src/site/apt/privacy-policy.apt new file mode 100644 index 0000000..ada9363 --- /dev/null +++ b/website/0.7.0-incubating/src/site/apt/privacy-policy.apt @@ -0,0 +1,52 @@ + ---- + Privacy Policy + ----- + Olivier Lamy + ----- + 2013-02-04 + ----- + +~~ Licensed to the Apache Software Foundation (ASF) under one +~~ or more contributor license agreements. See the NOTICE file +~~ distributed with this work for additional information +~~ regarding copyright ownership. The ASF licenses this file +~~ to you under the Apache License, Version 2.0 (the +~~ "License"); you may not use this file except in compliance +~~ with the License. You may obtain a copy of the License at +~~ +~~ http://www.apache.org/licenses/LICENSE-2.0 +~~ +~~ Unless required by applicable law or agreed to in writing, +~~ software distributed under the License is distributed on an +~~ "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +~~ KIND, either express or implied. See the License for the +~~ specific language governing permissions and limitations +~~ under the License. + +Privacy Policy + + Information about your use of this website is collected using server access logs and a tracking cookie. The + collected information consists of the following: + + [[1]] The IP address from which you access the website; + + [[2]] The type of browser and operating system you use to access our site; + + [[3]] The date and time you access our site; + + [[4]] The pages you visit; and + + [[5]] The addresses of pages from where you followed a link to our site. + + [] + + Part of this information is gathered using a tracking cookie set by the + {{{http://www.google.com/analytics/}Google Analytics}} service and handled by Google as described in their + {{{http://www.google.com/privacy.html}privacy policy}}. See your browser documentation for instructions on how to + disable the cookie if you prefer not to share this data with Google. + + We use the gathered information to help us make our site more useful to visitors and to better understand how and + when our site is used. We do not track or collect personally identifiable information or associate gathered data + with any personally identifying information from other sources. + + By using this website, you consent to the collection of this data in the manner and for the purpose described above. http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.7.0-incubating/src/site/apt/releasenotes/release-0.7.0-incubating.apt ---------------------------------------------------------------------- diff --git a/website/0.7.0-incubating/src/site/apt/releasenotes/release-0.7.0-incubating.apt b/website/0.7.0-incubating/src/site/apt/releasenotes/release-0.7.0-incubating.apt new file mode 100644 index 0000000..7661df0 --- /dev/null +++ b/website/0.7.0-incubating/src/site/apt/releasenotes/release-0.7.0-incubating.apt @@ -0,0 +1,174 @@ + ----- + Release Notes for Apache Helix 0.7.0-incubating + ----- + +~~ Licensed to the Apache Software Foundation (ASF) under one +~~ or more contributor license agreements. See the NOTICE file +~~ distributed with this work for additional information +~~ regarding copyright ownership. The ASF licenses this file +~~ to you under the Apache License, Version 2.0 (the +~~ "License"); you may not use this file except in compliance +~~ with the License. You may obtain a copy of the License at +~~ +~~ http://www.apache.org/licenses/LICENSE-2.0 +~~ +~~ Unless required by applicable law or agreed to in writing, +~~ software distributed under the License is distributed on an +~~ "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +~~ KIND, either express or implied. See the License for the +~~ specific language governing permissions and limitations +~~ under the License. + +~~ NOTE: For help with the syntax of this file, see: +~~ http://maven.apache.org/guides/mini/guide-apt-format.html + +Release Notes for Apache Helix 0.7.0-incubating + + The Apache Helix team would like to announce the release of Apache Helix 0.7.0-incubating + + This is the fourth release and second major release under the Apache umbrella. + + Helix is a generic cluster management framework used for the automatic management of partitioned, replicated and distributed resources hosted on a cluster of nodes. Helix provides the following features: + + * Automatic assignment of resource/partition to nodes + + * Node failure detection and recovery + + * Dynamic addition of Resources + + * Dynamic addition of nodes to the cluster + + * Pluggable distributed state machine to manage the state of a resource via state transitions + + * Automatic load balancing and throttling of transitions + + * Configurable, pluggable rebalancing + + [] + +* Changes + +** Sub-task + + * [HELIX-18] - Unify cluster setup and helixadmin + + * [HELIX-79] - consecutive GC may mess up helix session ids + + * [HELIX-83] - Add typed classes to denote helix ids + + * [HELIX-90] - Clean up Api's + + * [HELIX-98] - clean up setting constraint api + + * [HELIX-100] - Improve the helix config api + + * [HELIX-102] - Add new wrapper classes for Participant, Controller, Spectator, Administrator + + * [HELIX-104] - Add support to reuse zkclient + + * [HELIX-123] - ZkHelixManager.isLeader() should check session id in addition to instance name + + * [HELIX-139] - Need to double check the logic to prevent 2 controllers to control the same cluster + + * [HELIX-168] - separate HelixManager implementation for participant, controller, and distributed controller + + * [HELIX-176] - Need a list of tests that must pass to certify a release + + * [HELIX-224] - Move helix examples to separate module + + * [HELIX-233] - Ensure that website and wiki fully capture the updated changes in 0.7.0 + + * [HELIX-234] - Create concrete id classes for constructs, replacing strings + + * [HELIX-235] - Create a hierarchical logical model for the cluster + + * [HELIX-236] - Create a hierarchical cluster snapshot to replace ClusterDataCache + + * [HELIX-237] - Create helix-internal config classes for the hierarchical model + + * [HELIX-238] - Create accessors for the logical model + + * [HELIX-239] - List use cases for the logical model + + * [HELIX-240] - Write an example of the key use cases for the logical model + + * [HELIX-241] - Write the controller pipeline with the logical model + + * [HELIX-242] - Re-integrate the scheduler rebalancing into the new controller pipeline + + * [HELIX-243] - Fix failing tests related to helix model overhaul + + * [HELIX-244] - Redesign rebalancers using rebalancer-specific configs + + * [HELIX-246] - Refactor scheduler task config to comply with new rebalancer config and fix related scheduler task tests + + * [HELIX-248] - Resource logical model should be general enough to handle various resource types + + * [HELIX-268] - Atomic API + + * [HELIX-297] - Make 0.7.0 backward compatible for user-defined rebalancing + + +** Bug + + * [HELIX-40] - fix zkclient subscribe path leaking and zk callback-handler leaking in case of session expiry + + * [HELIX-46] - Add REST/cli admin command for message selection constraints + + * [HELIX-47] - when drop resource, remove resource-level config also + + * [HELIX-48] - use resource instead of db in output messages + + * [HELIX-50] - Ensure num replicas and preference list size in idealstate matches + + * [HELIX-59] - controller not cleaning dead external view generated from old sessions + + * [HELIX-136] - Write IdealState back to ZK when computed by custom Rebalancer + + * [HELIX-200] - helix controller send ERROR->DROPPED transition infinitely + + * [HELIX-214] - User-defined rebalancer should never use SEMI_AUTO code paths + + * [HELIX-225] - fix helix-example package build error + + * [HELIX-271] - ZkHelixAdmin#addResource() backward compatible problem + + * [HELIX-292] - ZNRecordStreamingSerializer should not assume id comes first + + * [HELIX-296] - HelixConnection in 0.7.0 does not remove LiveInstance znode + + * [HELIX-300] - Some files in 0.7.0 are missing license headers + + * [HELIX-302] - fix helix version compare bug + +** Improvement + + * [HELIX-37] - Cleanup CallbackHandler + + * [HELIX-202] - Ideal state should be a full mapping, not just a set of instance preferences + +** Task + + * [HELIX-109] - Review Helix model package + + * [HELIX-174] - Clean up ideal state calculators, move them to the controller rebalancer package + + * [HELIX-212] - Rebalancer interface should have 1 function to compute the entire ideal state + + * [HELIX-232] - Validation of 0.7.0 + + * [HELIX-290] - Ensure 0.7.0 can respond correctly to ideal state changes + + * [HELIX-295] - Upgrade or remove xstream dependency + + * [HELIX-301] - Update integration test utils for 0.7.0 + +** Test + + * [HELIX-286] - add a test for redefine state model definition + + [] + + Cheers, + -- + The Apache Helix Team http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.7.0-incubating/src/site/markdown/Building.md ---------------------------------------------------------------------- diff --git a/website/0.7.0-incubating/src/site/markdown/Building.md b/website/0.7.0-incubating/src/site/markdown/Building.md new file mode 100644 index 0000000..8e1ec84 --- /dev/null +++ b/website/0.7.0-incubating/src/site/markdown/Building.md @@ -0,0 +1,42 @@ +<!--- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. +--> + +Build Instructions +------------------ + +### From Source + +Requirements: JDK 1.6+, Maven 2.0.8+ + +``` +git clone https://git-wip-us.apache.org/repos/asf/helix.git +cd helix +git checkout tags/helix-0.7.0-incubating +mvn install package -DskipTests +``` + +### Maven Dependency + +``` +<dependency> + <groupId>org.apache.helix</groupId> + <artifactId>helix-core</artifactId> + <version>0.7.0-incubating</version> +</dependency> +``` http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.7.0-incubating/src/site/markdown/Features.md ---------------------------------------------------------------------- diff --git a/website/0.7.0-incubating/src/site/markdown/Features.md b/website/0.7.0-incubating/src/site/markdown/Features.md new file mode 100644 index 0000000..ba9d0e7 --- /dev/null +++ b/website/0.7.0-incubating/src/site/markdown/Features.md @@ -0,0 +1,313 @@ +<!--- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. +--> + +<head> + <title>Features</title> +</head> + +Features +---------------------------- + + +### CONFIGURING IDEALSTATE + + +Read concepts page for definition of Idealstate. + +The placement of partitions in a DDS is very critical for reliability and scalability of the system. +For example, when a node fails, it is important that the partitions hosted on that node are reallocated evenly among the remaining nodes. Consistent hashing is one such algorithm that can guarantee this. +Helix by default comes with a variant of consistent hashing based of the RUSH algorithm. + +This means given a number of partitions, replicas and number of nodes Helix does the automatic assignment of partition to nodes such that + +* Each node has the same number of partitions and replicas of the same partition do not stay on the same node. +* When a node fails, the partitions will be equally distributed among the remaining nodes +* When new nodes are added, the number of partitions moved will be minimized along with satisfying the above two criteria. + + +Helix provides multiple ways to control the placement and state of a replica. + +``` + + |AUTO REBALANCE| AUTO | CUSTOM | + ----------------------------------------- + LOCATION | HELIX | APP | APP | + ----------------------------------------- + STATE | HELIX | HELIX | APP | + ----------------------------------------- +``` + +#### HELIX EXECUTION MODE + + +Idealstate is defined as the state of the DDS when all nodes are up and running and healthy. +Helix uses this as the target state of the system and computes the appropriate transitions needed in the system to bring it to a stable state. + +Helix supports 3 different execution modes which allows application to explicitly control the placement and state of the replica. + +##### AUTO_REBALANCE + +When the idealstate mode is set to AUTO_REBALANCE, Helix controls both the location of the replica along with the state. This option is useful for applications where creation of a replica is not expensive. Example + +``` +{ + "id" : "MyResource", + "simpleFields" : { + "IDEAL_STATE_MODE" : "AUTO_REBALANCE", + "NUM_PARTITIONS" : "3", + "REPLICAS" : "2", + "STATE_MODEL_DEF_REF" : "MasterSlave", + } + "listFields" : { + "MyResource_0" : [], + "MyResource_1" : [], + "MyResource_2" : [] + }, + "mapFields" : { + } +} +``` + +If there are 3 nodes in the cluster, then Helix will internally compute the ideal state as + +``` +{ + "id" : "MyResource", + "simpleFields" : { + "NUM_PARTITIONS" : "3", + "REPLICAS" : "2", + "STATE_MODEL_DEF_REF" : "MasterSlave", + }, + "mapFields" : { + "MyResource_0" : { + "N1" : "MASTER", + "N2" : "SLAVE", + }, + "MyResource_1" : { + "N2" : "MASTER", + "N3" : "SLAVE", + }, + "MyResource_2" : { + "N3" : "MASTER", + "N1" : "SLAVE", + } + } +} +``` + +Another typical example is evenly distributing a group of tasks among the currently alive processes. For example, if there are 60 tasks and 4 nodes, Helix assigns 15 tasks to each node. +When one node fails Helix redistributes its 15 tasks to the remaining 3 nodes. Similarly, if a node is added, Helix re-allocates 3 tasks from each of the 4 nodes to the 5th node. + +#### AUTO + +When the idealstate mode is set to AUTO, Helix only controls STATE of the replicas where as the location of the partition is controlled by application. Example: The below idealstate indicates thats 'MyResource_0' must be only on node1 and node2. But gives the control of assigning the STATE to Helix. + +``` +{ + "id" : "MyResource", + "simpleFields" : { + "IDEAL_STATE_MODE" : "AUTO", + "NUM_PARTITIONS" : "3", + "REPLICAS" : "2", + "STATE_MODEL_DEF_REF" : "MasterSlave", + } + "listFields" : { + "MyResource_0" : [node1, node2], + "MyResource_1" : [node2, node3], + "MyResource_2" : [node3, node1] + }, + "mapFields" : { + } +} +``` +In this mode when node1 fails, unlike in AUTO-REBALANCE mode the partition is not moved from node1 to others nodes in the cluster. Instead, Helix will decide to change the state of MyResource_0 in N2 based on the system constraints. For example, if a system constraint specified that there should be 1 Master and if the Master failed, then node2 will be made the new master. + +#### CUSTOM + +Helix offers a third mode called CUSTOM, in which application can completely control the placement and state of each replica. Applications will have to implement an interface that Helix will invoke when the cluster state changes. +Within this callback, the application can recompute the idealstate. Helix will then issue appropriate transitions such that Idealstate and Currentstate converges. + +``` +{ + "id" : "MyResource", + "simpleFields" : { + "IDEAL_STATE_MODE" : "CUSTOM", + "NUM_PARTITIONS" : "3", + "REPLICAS" : "2", + "STATE_MODEL_DEF_REF" : "MasterSlave", + }, + "mapFields" : { + "MyResource_0" : { + "N1" : "MASTER", + "N2" : "SLAVE", + }, + "MyResource_1" : { + "N2" : "MASTER", + "N3" : "SLAVE", + }, + "MyResource_2" : { + "N3" : "MASTER", + "N1" : "SLAVE", + } + } +} +``` + +For example, the current state of the system might be 'MyResource_0' -> {N1:MASTER,N2:SLAVE} and the application changes the ideal state to 'MyResource_0' -> {N1:SLAVE,N2:MASTER}. Helix will not blindly issue MASTER-->SLAVE to N1 and SLAVE-->MASTER to N2 in parallel since it might result in a transient state where both N1 and N2 are masters. +Helix will first issue MASTER-->SLAVE to N1 and after its completed it will issue SLAVE-->MASTER to N2. + + +### State Machine Configuration + +Helix comes with 3 default state models that are most commonly used. Its possible to have multiple state models in a cluster. +Every resource that is added should have a reference to the state model. + +* MASTER-SLAVE: Has 3 states OFFLINE,SLAVE,MASTER. Max masters is 1. Slaves will be based on the replication factor. Replication factor can be specified while adding the resource +* ONLINE-OFFLINE: Has 2 states OFFLINE and ONLINE. Very simple state model and most applications start off with this state model. +* LEADER-STANDBY:1 Leader and many stand bys. In general the standby's are idle. + +Apart from providing the state machine configuration, one can specify the constraints of states and transitions. + +For example one can say +Master:1. Max number of replicas in Master state at any time is 1. +OFFLINE-SLAVE:5 Max number of Offline-Slave transitions that can happen concurrently in the system + +STATE PRIORITY +Helix uses greedy approach to satisfy the state constraints. For example if the state machine configuration says it needs 1 master and 2 slaves but only 1 node is active, Helix must promote it to master. This behavior is achieved by providing the state priority list as MASTER,SLAVE. + +STATE TRANSITION PRIORITY +Helix tries to fire as many transitions as possible in parallel to reach the stable state without violating constraints. By default Helix simply sorts the transitions alphabetically and fires as many as it can without violating the constraints. +One can control this by overriding the priority order. + +### Config management + +Helix allows applications to store application specific properties. The configuration can have different scopes. + +* Cluster +* Node specific +* Resource specific +* Partition specific + +Helix also provides notifications when any configs are changed. This allows applications to support dynamic configuration changes. + +See HelixManager.getConfigAccessor for more info + +### Intra cluster messaging api + +This is an interesting feature which is quite useful in practice. Often times, nodes in DDS requires a mechanism to interact with each other. One such requirement is a process of bootstrapping a replica. + +Consider a search system use case where the index replica starts up and it does not have an index. One of the commonly used solutions is to get the index from a common location or to copy the index from another replica. +Helix provides a messaging api, that can be used to talk to other nodes in the system. The value added that Helix provides here is, message recipient can be specified in terms of resource, +partition, state and Helix ensures that the message is delivered to all of the required recipients. In this particular use case, the instance can specify the recipient criteria as all replicas of P1. +Since Helix is aware of the global state of the system, it can send the message to appropriate nodes. Once the nodes respond Helix provides the bootstrapping replica with all the responses. + +This is a very generic api and can also be used to schedule various periodic tasks in the cluster like data backups etc. +System Admins can also perform adhoc tasks like on demand backup or execute a system command(like rm -rf ;-)) across all nodes. + +``` + ClusterMessagingService messagingService = manager.getMessagingService(); + //CONSTRUCT THE MESSAGE + Message requestBackupUriRequest = new Message( + MessageType.USER_DEFINE_MSG, UUID.randomUUID().toString()); + requestBackupUriRequest + .setMsgSubType(BootstrapProcess.REQUEST_BOOTSTRAP_URL); + requestBackupUriRequest.setMsgState(MessageState.NEW); + //SET THE RECIPIENT CRITERIA, All nodes that satisfy the criteria will receive the message + Criteria recipientCriteria = new Criteria(); + recipientCriteria.setInstanceName("%"); + recipientCriteria.setRecipientInstanceType(InstanceType.PARTICIPANT); + recipientCriteria.setResource("MyDB"); + recipientCriteria.setPartition(""); + //Should be processed only the process that is active at the time of sending the message. + //This means if the recipient is restarted after message is sent, it will not be processed. + recipientCriteria.setSessionSpecific(true); + // wait for 30 seconds + int timeout = 30000; + //The handler that will be invoked when any recipient responds to the message. + BootstrapReplyHandler responseHandler = new BootstrapReplyHandler(); + //This will return only after all recipients respond or after timeout. + int sentMessageCount = messagingService.sendAndWait(recipientCriteria, + requestBackupUriRequest, responseHandler, timeout); +``` + +See HelixManager.getMessagingService for more info. + + +### Application specific property storage + +There are several usecases where applications needs support for distributed data structures. Helix uses Zookeeper to store the application data and hence provides notifications when the data changes. +One value add Helix provides is the ability to specify cache the data and also write through cache. This is more efficient than reading from ZK every time. + +See HelixManager.getHelixPropertyStore + +### Throttling + +Since all state changes in the system are triggered through transitions, Helix can control the number of transitions that can happen in parallel. Some of the transitions may be light weight but some might involve moving data around which is quite expensive. +Helix allows applications to set threshold on transitions. The threshold can be set at the multiple scopes. + +* MessageType e.g STATE_TRANSITION +* TransitionType e.g SLAVE-MASTER +* Resource e.g database +* Node i.e per node max transitions in parallel. + +See HelixManager.getHelixAdmin.addMessageConstraint() + +### Health monitoring and alerting + +This in currently in development mode, not yet productionized. + +Helix provides ability for each node in the system to report health metrics on a periodic basis. +Helix supports multiple ways to aggregate these metrics like simple SUM, AVG, EXPONENTIAL DECAY, WINDOW. Helix will only persist the aggregated value. +Applications can define threshold on the aggregate values according to the SLA's and when the SLA is violated Helix will fire an alert. +Currently Helix only fires an alert but eventually we plan to use this metrics to either mark the node dead or load balance the partitions. +This feature will be valuable in for distributed systems that support multi-tenancy and have huge variation in work load patterns. Another place this can be used is to detect skewed partitions and rebalance the cluster. + +This feature is not yet stable and do not recommend to be used in production. + + +### Controller deployment modes + +Read Architecture wiki for more details on the Role of a controller. In simple words, it basically controls the participants in the cluster by issuing transitions. + +Helix provides multiple options to deploy the controller. + +#### STANDALONE + +Controller can be started as a separate process to manage a cluster. This is the recommended approach. How ever since one controller can be a single point of failure, multiple controller processes are required for reliability. +Even if multiple controllers are running only one will be actively managing the cluster at any time and is decided by a leader election process. If the leader fails, another leader will resume managing the cluster. + +Even though we recommend this method of deployment, it has the drawback of having to manage an additional service for each cluster. See Controller As a Service option. + +#### EMBEDDED + +If setting up a separate controller process is not viable, then it is possible to embed the controller as a library in each of the participant. + +#### CONTROLLER AS A SERVICE + +One of the cool feature we added in helix was use a set of controllers to manage a large number of clusters. +For example if you have X clusters to be managed, instead of deploying X*3(3 controllers for fault tolerance) controllers for each cluster, one can deploy only 3 controllers. Each controller can manage X/3 clusters. +If any controller fails the remaining two will manage X/2 clusters. At LinkedIn, we always deploy controllers in this mode. + + + + + + + + http://git-wip-us.apache.org/repos/asf/helix/blob/39e0d3fb/website/0.7.0-incubating/src/site/markdown/Quickstart.md ---------------------------------------------------------------------- diff --git a/website/0.7.0-incubating/src/site/markdown/Quickstart.md b/website/0.7.0-incubating/src/site/markdown/Quickstart.md new file mode 100644 index 0000000..942b386 --- /dev/null +++ b/website/0.7.0-incubating/src/site/markdown/Quickstart.md @@ -0,0 +1,666 @@ +<!--- +Licensed to the Apache Software Foundation (ASF) under one +or more contributor license agreements. See the NOTICE file +distributed with this work for additional information +regarding copyright ownership. The ASF licenses this file +to you under the Apache License, Version 2.0 (the +"License"); you may not use this file except in compliance +with the License. You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + +Unless required by applicable law or agreed to in writing, +software distributed under the License is distributed on an +"AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +KIND, either express or implied. See the License for the +specific language governing permissions and limitations +under the License. +--> + +<head> + <title>Quickstart</title> +</head> + +Quickstart +--------- + +Get Helix +--------- + +First, let\'s get Helix. Either build it, or download it. + +### Build + +``` +git clone https://git-wip-us.apache.org/repos/asf/helix.git +cd helix +git checkout tags/helix-0.7.0-incubating +mvn install package -DskipTests +# This folder contains quickstart.sh and start-helix-participant.sh +cd helix-examples/target/helix-examples-pkg/bin +chmod +x * +# This folder contains helix-admin.sh, start-standalone-zookeeper.sh, and run-helix-controller.sh +cd ../../../../helix-core/target/helix-core-pkg/bin +``` + +### Download + +Download the 0.7.0-incubating release package [here](./download.html) + +Overview +-------- + +In this Quickstart, we\'ll set up a master-slave replicated, partitioned system. Then we\'ll demonstrate how to add a node, rebalance the partitions, and show how Helix manages failover. + + +Let\'s Do It +------------ + +Helix provides command line interfaces to set up the cluster and view the cluster state. The best way to understand how Helix views a cluster is to build a cluster. + +### Get to the Tools Directory + +If you built the code: + +``` +cd helix/helix/helix-examples/target/helix-examples-pkg/bin +``` + +If you downloaded the release package, extract it. + + +Short Version +------------- +You can observe the components working together in this demo, which does the following: + +* Create a cluster +* Add 2 nodes (participants) to the cluster +* Set up a resource with 6 partitions and 2 replicas: 1 Master, and 1 Slave per partition +* Show the cluster state after Helix balances the partitions +* Add a third node +* Show the cluster state. Note that the third node has taken mastership of 2 partitions. +* Kill the third node (Helix takes care of failover) +* Show the cluster state. Note that the two surviving nodes take over mastership of the partitions from the failed node + +### Run the Demo + +``` +cd helix/helix/helix-examples/target/helix-examples-pkg/bin +./quickstart.sh +``` + +#### The Initial Setup + +2 nodes are set up and the partitions are rebalanced. + +The cluster state is as follows: + +``` +CLUSTER STATE: After starting 2 nodes + localhost_12000 localhost_12001 +MyResource_0 M S +MyResource_1 S M +MyResource_2 M S +MyResource_3 M S +MyResource_4 S M +MyResource_5 S M +``` + +Note there is one master and one slave per partition. + +#### Add a Node + +A third node is added and the cluster is rebalanced. + +The cluster state changes to: + +``` +CLUSTER STATE: After adding a third node + localhost_12000 localhost_12001 localhost_12002 +MyResource_0 S M S +MyResource_1 S S M +MyResource_2 M S S +MyResource_3 S S M +MyResource_4 M S S +MyResource_5 S M S +``` + +Note there is one master and _two_ slaves per partition. This is expected because there are three nodes. + +#### Kill a Node + +Finally, a node is killed to simulate a failure + +Helix makes sure each partition has a master. The cluster state changes to: + +``` +CLUSTER STATE: After the 3rd node stops/crashes + localhost_12000 localhost_12001 localhost_12002 +MyResource_0 S M - +MyResource_1 S M - +MyResource_2 M S - +MyResource_3 M S - +MyResource_4 M S - +MyResource_5 S M - +``` + + +Long Version +------------ +Now you can run the same steps by hand. In this detailed version, we\'ll do the following: + +* Define a cluster +* Add two nodes to the cluster +* Add a 6-partition resource with 1 master and 2 slave replicas per partition +* Verify that the cluster is healthy and inspect the Helix view +* Expand the cluster: add a few nodes and rebalance the partitions +* Failover: stop a node and verify the mastership transfer + +### Install and Start ZooKeeper + +Zookeeper can be started in standalone mode or replicated mode. + +More information is available at + +* http://zookeeper.apache.org/doc/r3.3.3/zookeeperStarted.html +* http://zookeeper.apache.org/doc/trunk/zookeeperAdmin.html#sc_zkMulitServerSetup + +In this example, let\'s start zookeeper in local mode. + +#### Start ZooKeeper Locally on Port 2199 + +``` +./start-standalone-zookeeper.sh 2199 & +``` + +### Define the Cluster + +The helix-admin tool is used for cluster administration tasks. In the Quickstart, we\'ll use the command line interface. Helix supports a REST interface as well. + +zookeeper_address is of the format host:port e.g localhost:2199 for standalone or host1:port,host2:port for multi-node. + +Next, we\'ll set up a cluster MYCLUSTER cluster with these attributes: + +* 3 instances running on localhost at ports 12913,12914,12915 +* One database named myDB with 6 partitions +* Each partition will have 3 replicas with 1 master, 2 slaves +* ZooKeeper running locally at localhost:2199 + +#### Create the Cluster MYCLUSTER + +``` +# ./helix-admin.sh --zkSvr <zk_address> --addCluster <clustername> +./helix-admin.sh --zkSvr localhost:2199 --addCluster MYCLUSTER +``` + +### Add Nodes to the Cluster + +In this case we\'ll add three nodes: localhost:12913, localhost:12914, localhost:12915 + +``` +# helix-admin.sh --zkSvr <zk_address> --addNode <clustername> <host:port> +./helix-admin.sh --zkSvr localhost:2199 --addNode MYCLUSTER localhost:12913 +./helix-admin.sh --zkSvr localhost:2199 --addNode MYCLUSTER localhost:12914 +./helix-admin.sh --zkSvr localhost:2199 --addNode MYCLUSTER localhost:12915 +``` + +### Define the Resource and Partitioning + +In this example, the resource is a database, partitioned 6 ways. Note that in a production system, it\'s common to over-partition for better load balancing. Helix has been used in production to manage hundreds of databases each with 10s or 100s of partitions running on 10s of physical nodes. + +#### Create a Database with 6 Partitions using the MasterSlave State Model + +Helix ensures there will be exactly one master for each partition. + +``` +# helix-admin.sh --zkSvr <zk_address> --addResource <clustername> <resourceName> <numPartitions> <StateModelName> +./helix-admin.sh --zkSvr localhost:2199 --addResource MYCLUSTER myDB 6 MasterSlave +``` + +#### Let Helix Assign Partitions to Nodes + +This command will distribute the partitions amongst all the nodes in the cluster. In this example, each partition has 3 replicas. + +``` +# helix-admin.sh --zkSvr <zk_address> --rebalance <clustername> <resourceName> <replication factor> +./helix-admin.sh --zkSvr localhost:2199 --rebalance MYCLUSTER myDB 3 +``` + +Now the cluster is defined in ZooKeeper. The nodes (localhost:12913, localhost:12914, localhost:12915) and resource (myDB, with 6 partitions using the MasterSlave model) are all properly configured. And the _IdealState_ has been calculated, assuming a replication factor of 3. + +### Start the Helix Controller + +Now that the cluster is defined in ZooKeeper, the Helix controller can manage the cluster. + +``` +# Start the cluster manager, which will manage MYCLUSTER +./run-helix-controller.sh --zkSvr localhost:2199 --cluster MYCLUSTER 2>&1 > /tmp/controller.log & +``` + +### Start up the Cluster to be Managed + +We\'ve started up ZooKeeper, defined the cluster, the resources, the partitioning, and started up the Helix controller. Next, we\'ll start up the nodes of the system to be managed. Each node is a Participant, which is an instance of the system component to be managed. Helix assigns work to Participants, keeps track of their roles and health, and takes action when a node fails. + +``` +# start up each instance. These are mock implementations that are actively managed by Helix +./start-helix-participant.sh --zkSvr localhost:2199 --cluster MYCLUSTER --host localhost --port 12913 --stateModelType MasterSlave 2>&1 > /tmp/participant_12913.log +./start-helix-participant.sh --zkSvr localhost:2199 --cluster MYCLUSTER --host localhost --port 12914 --stateModelType MasterSlave 2>&1 > /tmp/participant_12914.log +./start-helix-participant.sh --zkSvr localhost:2199 --cluster MYCLUSTER --host localhost --port 12915 --stateModelType MasterSlave 2>&1 > /tmp/participant_12915.log +``` + +### Inspect the Cluster + +Now, let\'s see the Helix view of our cluster. We\'ll work our way down as follows: + +``` +Clusters -> MYCLUSTER -> instances -> instance detail + -> resources -> resource detail + -> partitions +``` + +A single Helix controller can manage multiple clusters, though so far, we\'ve only defined one cluster. Let\'s see: + +``` +# List existing clusters +./helix-admin.sh --zkSvr localhost:2199 --listClusters + +Existing clusters: +MYCLUSTER +``` + +Now, let\'s see the Helix view of MYCLUSTER: + +``` +# helix-admin.sh --zkSvr <zk_address> --listClusterInfo <clusterName> +./helix-admin.sh --zkSvr localhost:2199 --listClusterInfo MYCLUSTER + +Existing resources in cluster MYCLUSTER: +myDB +Instances in cluster MYCLUSTER: +localhost_12915 +localhost_12914 +localhost_12913 +``` + +Let\'s look at the details of an instance: + +``` +# ./helix-admin.sh --zkSvr <zk_address> --listInstanceInfo <clusterName> <InstanceName> +./helix-admin.sh --zkSvr localhost:2199 --listInstanceInfo MYCLUSTER localhost_12913 + +InstanceConfig: { + "id" : "localhost_12913", + "mapFields" : { + }, + "listFields" : { + }, + "simpleFields" : { + "HELIX_ENABLED" : "true", + "HELIX_HOST" : "localhost", + "HELIX_PORT" : "12913" + } +} +``` + + +#### Query Information about a Resource + +``` +# helix-admin.sh --zkSvr <zk_address> --listResourceInfo <clusterName> <resourceName> +./helix-admin.sh --zkSvr localhost:2199 --listResourceInfo MYCLUSTER myDB + +IdealState for myDB: +{ + "id" : "myDB", + "mapFields" : { + "myDB_0" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "MASTER", + "localhost_12915" : "SLAVE" + }, + "myDB_1" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "SLAVE", + "localhost_12915" : "MASTER" + }, + "myDB_2" : { + "localhost_12913" : "MASTER", + "localhost_12914" : "SLAVE", + "localhost_12915" : "SLAVE" + }, + "myDB_3" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "SLAVE", + "localhost_12915" : "MASTER" + }, + "myDB_4" : { + "localhost_12913" : "MASTER", + "localhost_12914" : "SLAVE", + "localhost_12915" : "SLAVE" + }, + "myDB_5" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "MASTER", + "localhost_12915" : "SLAVE" + } + }, + "listFields" : { + "myDB_0" : [ "localhost_12914", "localhost_12913", "localhost_12915" ], + "myDB_1" : [ "localhost_12915", "localhost_12913", "localhost_12914" ], + "myDB_2" : [ "localhost_12913", "localhost_12915", "localhost_12914" ], + "myDB_3" : [ "localhost_12915", "localhost_12913", "localhost_12914" ], + "myDB_4" : [ "localhost_12913", "localhost_12914", "localhost_12915" ], + "myDB_5" : [ "localhost_12914", "localhost_12915", "localhost_12913" ] + }, + "simpleFields" : { + "IDEAL_STATE_MODE" : "AUTO", + "REBALANCE_MODE" : "SEMI_AUTO", + "NUM_PARTITIONS" : "6", + "REPLICAS" : "3", + "STATE_MODEL_DEF_REF" : "MasterSlave", + "STATE_MODEL_FACTORY_NAME" : "DEFAULT" + } +} + +ExternalView for myDB: +{ + "id" : "myDB", + "mapFields" : { + "myDB_0" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "MASTER", + "localhost_12915" : "SLAVE" + }, + "myDB_1" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "SLAVE", + "localhost_12915" : "MASTER" + }, + "myDB_2" : { + "localhost_12913" : "MASTER", + "localhost_12914" : "SLAVE", + "localhost_12915" : "SLAVE" + }, + "myDB_3" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "SLAVE", + "localhost_12915" : "MASTER" + }, + "myDB_4" : { + "localhost_12913" : "MASTER", + "localhost_12914" : "SLAVE", + "localhost_12915" : "SLAVE" + }, + "myDB_5" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "MASTER", + "localhost_12915" : "SLAVE" + } + }, + "listFields" : { + }, + "simpleFields" : { + "BUCKET_SIZE" : "0" + } +} +``` + +Now, let\'s look at one of the partitions: + +``` +# helix-admin.sh --zkSvr <zk_address> --listResourceInfo <clusterName> <partition> +./helix-admin.sh --zkSvr localhost:2199 --listResourceInfo mycluster myDB_0 +``` + +### Expand the Cluster + +Next, we\'ll show how Helix does the work that you\'d otherwise have to build into your system. When you add capacity to your cluster, you want the work to be evenly distributed. In this example, we started with 3 nodes, with 6 partitions. The partitions were evenly balanced, 2 masters and 4 slaves per node. Let\'s add 3 more nodes: localhost:12916, localhost:12917, localhost:12918 + +``` +./helix-admin.sh --zkSvr localhost:2199 --addNode MYCLUSTER localhost:12916 +./helix-admin.sh --zkSvr localhost:2199 --addNode MYCLUSTER localhost:12917 +./helix-admin.sh --zkSvr localhost:2199 --addNode MYCLUSTER localhost:12918 +``` + +And start up these instances: + +``` +# start up each instance. These are mock implementations that are actively managed by Helix +./start-helix-participant.sh --zkSvr localhost:2199 --cluster MYCLUSTER --host localhost --port 12916 --stateModelType MasterSlave 2>&1 > /tmp/participant_12916.log +./start-helix-participant.sh --zkSvr localhost:2199 --cluster MYCLUSTER --host localhost --port 12917 --stateModelType MasterSlave 2>&1 > /tmp/participant_12917.log +./start-helix-participant.sh --zkSvr localhost:2199 --cluster MYCLUSTER --host localhost --port 12918 --stateModelType MasterSlave 2>&1 > /tmp/participant_12918.log +``` + + +And now, let Helix do the work for you. To shift the work, simply rebalance. After the rebalance, each node will have one master and two slaves. + +``` +./helix-admin.sh --zkSvr localhost:2199 --rebalance MYCLUSTER myDB 3 +``` + +### View the Cluster + +OK, let\'s see how it looks: + + +``` +./helix-admin.sh --zkSvr localhost:2199 --listResourceInfo MYCLUSTER myDB + +IdealState for myDB: +{ + "id" : "myDB", + "mapFields" : { + "myDB_0" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "SLAVE", + "localhost_12917" : "MASTER" + }, + "myDB_1" : { + "localhost_12916" : "SLAVE", + "localhost_12917" : "SLAVE", + "localhost_12918" : "MASTER" + }, + "myDB_2" : { + "localhost_12913" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_3" : { + "localhost_12915" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_4" : { + "localhost_12916" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_5" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "MASTER", + "localhost_12915" : "SLAVE" + } + }, + "listFields" : { + "myDB_0" : [ "localhost_12917", "localhost_12913", "localhost_12914" ], + "myDB_1" : [ "localhost_12918", "localhost_12917", "localhost_12916" ], + "myDB_2" : [ "localhost_12913", "localhost_12917", "localhost_12918" ], + "myDB_3" : [ "localhost_12915", "localhost_12917", "localhost_12918" ], + "myDB_4" : [ "localhost_12916", "localhost_12917", "localhost_12918" ], + "myDB_5" : [ "localhost_12914", "localhost_12915", "localhost_12913" ] + }, + "simpleFields" : { + "IDEAL_STATE_MODE" : "AUTO", + "REBALANCE_MODE" : "SEMI_AUTO", + "NUM_PARTITIONS" : "6", + "REPLICAS" : "3", + "STATE_MODEL_DEF_REF" : "MasterSlave", + "STATE_MODEL_FACTORY_NAME" : "DEFAULT" + } +} + +ExternalView for myDB: +{ + "id" : "myDB", + "mapFields" : { + "myDB_0" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "SLAVE", + "localhost_12917" : "MASTER" + }, + "myDB_1" : { + "localhost_12916" : "SLAVE", + "localhost_12917" : "SLAVE", + "localhost_12918" : "MASTER" + }, + "myDB_2" : { + "localhost_12913" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_3" : { + "localhost_12915" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_4" : { + "localhost_12916" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_5" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "MASTER", + "localhost_12915" : "SLAVE" + } + }, + "listFields" : { + }, + "simpleFields" : { + "BUCKET_SIZE" : "0" + } +} +``` + +Mission accomplished. The partitions are nicely balanced. + +### How about Failover? + +Building a fault tolerant system isn\'t trivial, but with Helix, it\'s easy. Helix detects a failed instance, and triggers mastership transfer automatically. + +First, let's fail an instance. In this example, we\'ll kill localhost:12918 to simulate a failure. + +We lost localhost:12918, so myDB_1 lost its MASTER. Helix can fix that, it will transfer mastership to a healthy node that is currently a SLAVE, say localhost:12197. Helix balances the load as best as it can, given there are 6 partitions on 5 nodes. Let\'s see: + + +``` +./helix-admin.sh --zkSvr localhost:2199 --listResourceInfo MYCLUSTER myDB + +IdealState for myDB: +{ + "id" : "myDB", + "mapFields" : { + "myDB_0" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "SLAVE", + "localhost_12917" : "MASTER" + }, + "myDB_1" : { + "localhost_12916" : "SLAVE", + "localhost_12917" : "SLAVE", + "localhost_12918" : "MASTER" + }, + "myDB_2" : { + "localhost_12913" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_3" : { + "localhost_12915" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_4" : { + "localhost_12916" : "MASTER", + "localhost_12917" : "SLAVE", + "localhost_12918" : "SLAVE" + }, + "myDB_5" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "MASTER", + "localhost_12915" : "SLAVE" + } + }, + "listFields" : { + "myDB_0" : [ "localhost_12917", "localhost_12913", "localhost_12914" ], + "myDB_1" : [ "localhost_12918", "localhost_12917", "localhost_12916" ], + "myDB_2" : [ "localhost_12913", "localhost_12918", "localhost_12917" ], + "myDB_3" : [ "localhost_12915", "localhost_12918", "localhost_12917" ], + "myDB_4" : [ "localhost_12916", "localhost_12917", "localhost_12918" ], + "myDB_5" : [ "localhost_12914", "localhost_12915", "localhost_12913" ] + }, + "simpleFields" : { + "IDEAL_STATE_MODE" : "AUTO", + "REBALANCE_MODE" : "SEMI_AUTO", + "NUM_PARTITIONS" : "6", + "REPLICAS" : "3", + "STATE_MODEL_DEF_REF" : "MasterSlave", + "STATE_MODEL_FACTORY_NAME" : "DEFAULT" + } +} + +ExternalView for myDB: +{ + "id" : "myDB", + "mapFields" : { + "myDB_0" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "SLAVE", + "localhost_12917" : "MASTER" + }, + "myDB_1" : { + "localhost_12916" : "SLAVE", + "localhost_12917" : "MASTER" + }, + "myDB_2" : { + "localhost_12913" : "MASTER", + "localhost_12917" : "SLAVE" + }, + "myDB_3" : { + "localhost_12915" : "MASTER", + "localhost_12917" : "SLAVE" + }, + "myDB_4" : { + "localhost_12916" : "MASTER", + "localhost_12917" : "SLAVE" + }, + "myDB_5" : { + "localhost_12913" : "SLAVE", + "localhost_12914" : "MASTER", + "localhost_12915" : "SLAVE" + } + }, + "listFields" : { + }, + "simpleFields" : { + "BUCKET_SIZE" : "0" + } +} +``` + +As we\'ve seen in this Quickstart, Helix takes care of partitioning, load balancing, elasticity, failure detection and recovery. + +### ZooInspector + +You can view all of the underlying data by going direct to zookeeper. Use ZooInspector that comes with zookeeper to browse the data. This is a java applet (make sure you have X windows) + +To start zooinspector run the following command from <zk_install_directory>/contrib/ZooInspector + +``` +java -cp zookeeper-3.3.3-ZooInspector.jar:lib/jtoaster-1.0.4.jar:../../lib/log4j-1.2.15.jar:../../zookeeper-3.3.3.jar org.apache.zookeeper.inspector.ZooInspector +``` + +### Next + +Now that you understand the idea of Helix, read the [tutorial](./Tutorial.html) to learn how to choose the right state model and constraints for your system, and how to implement it. In many cases, the built-in features meet your requirements. And best of all, Helix is a customizable framework, so you can plug in your own behavior, while retaining the automation provided by Helix. +
