Remove unused site src module
Project: http://git-wip-us.apache.org/repos/asf/helix/repo Commit: http://git-wip-us.apache.org/repos/asf/helix/commit/fc59b125 Tree: http://git-wip-us.apache.org/repos/asf/helix/tree/fc59b125 Diff: http://git-wip-us.apache.org/repos/asf/helix/diff/fc59b125 Branch: refs/heads/master Commit: fc59b12537a264a8ca728a085ae4e7ac6afa8ae4 Parents: 39e0d3f Author: Junkai Xue <[email protected]> Authored: Thu Jun 22 18:05:49 2017 -0700 Committer: Junkai Xue <[email protected]> Committed: Thu Jun 22 18:05:49 2017 -0700 ---------------------------------------------------------------------- src/site/apt/privacy-policy.apt | 52 -- .../releasenotes/release-0.6.0-incubating.apt | 77 --- .../releasenotes/release-0.6.1-incubating.apt | 110 ---- src/site/apt/releasing.apt | 107 ---- src/site/markdown/Architecture.md | 248 -------- src/site/markdown/Concepts.md | 268 -------- src/site/markdown/Features.md | 306 --------- src/site/markdown/Quickstart.md | 621 ------------------- src/site/markdown/Tutorial.md | 199 ------ src/site/markdown/UseCases.md | 109 ---- src/site/markdown/index.md | 127 ---- src/site/markdown/involved/building.md | 30 - src/site/markdown/recipes/lock_manager.md | 253 -------- .../markdown/recipes/rabbitmq_consumer_group.md | 227 ------- .../recipes/rsync_replicated_file_store.md | 165 ----- src/site/markdown/recipes/service_discovery.md | 191 ------ src/site/markdown/recipes/task_dag_execution.md | 204 ------ src/site/markdown/tutorial_admin.md | 167 ----- src/site/markdown/tutorial_controller.md | 90 --- src/site/markdown/tutorial_health.md | 42 -- src/site/markdown/tutorial_messaging.md | 67 -- src/site/markdown/tutorial_participant.md | 100 --- src/site/markdown/tutorial_propstore.md | 30 - src/site/markdown/tutorial_rebalance.md | 168 ----- src/site/markdown/tutorial_spectator.md | 72 --- src/site/markdown/tutorial_state.md | 127 ---- src/site/markdown/tutorial_throttling.md | 34 - src/site/resources/.htaccess | 20 - src/site/resources/download.cgi | 22 - src/site/resources/images/HELIX-components.png | Bin 82112 -> 0 bytes src/site/resources/images/PFS-Generic.png | Bin 72435 -> 0 bytes src/site/resources/images/RSYNC_BASED_PFS.png | Bin 78007 -> 0 bytes .../resources/images/bootstrap_statemodel.gif | Bin 24919 -> 0 bytes .../resources/images/helix-architecture.png | Bin 282390 -> 0 bytes src/site/resources/images/helix-logo.jpg | Bin 13659 -> 0 bytes .../resources/images/helix-znode-layout.png | Bin 53074 -> 0 bytes src/site/resources/images/statemachine.png | Bin 41641 -> 0 bytes src/site/resources/images/system.png | Bin 79791 -> 0 bytes src/site/site.xml | 122 ---- src/site/xdoc/download.xml.vm | 191 ------ 40 files changed, 4546 deletions(-) ---------------------------------------------------------------------- http://git-wip-us.apache.org/repos/asf/helix/blob/fc59b125/src/site/apt/privacy-policy.apt ---------------------------------------------------------------------- diff --git a/src/site/apt/privacy-policy.apt b/src/site/apt/privacy-policy.apt deleted file mode 100644 index ada9363..0000000 --- a/src/site/apt/privacy-policy.apt +++ /dev/null @@ -1,52 +0,0 @@ - ---- - 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/fc59b125/src/site/apt/releasenotes/release-0.6.0-incubating.apt ---------------------------------------------------------------------- diff --git a/src/site/apt/releasenotes/release-0.6.0-incubating.apt b/src/site/apt/releasenotes/release-0.6.0-incubating.apt deleted file mode 100644 index 16e2fbf..0000000 --- a/src/site/apt/releasenotes/release-0.6.0-incubating.apt +++ /dev/null @@ -1,77 +0,0 @@ - ----- - Release Notes for 0.6.0-incubating Apache Helix - ----- - -~~ 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 0.6.0-incubating Apache Helix - - The Apache Helix would like to announce the release of Apache Helix 0.6.0-incubating - - This is the first release in 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 - - [] - -* Changes - -** Bug - - * [HELIX-1] - Use org.apache.helix package for java sources. - - * [HELIX-2] - Remove jsqlparser dependency from Helix - - * [HELIX-3] - Fix license headers in sources. - - * [HELIX-12] - Issue with starting multiple controllers with same name - - * [HELIX-14] - error in helix-core ivy file - - [] - -** Task - - * [HELIX-4] - Remove deprecated file based implementation - - * [HELIX-5] - Remove deprecated Accessors - - * [HELIX-13] - New usecase to replicate files between replicas using simple rsync - - * [HELIX-15] - Distributed lock manager recipe - - [] - - Have Fun - -- - The Apache Helix Team http://git-wip-us.apache.org/repos/asf/helix/blob/fc59b125/src/site/apt/releasenotes/release-0.6.1-incubating.apt ---------------------------------------------------------------------- diff --git a/src/site/apt/releasenotes/release-0.6.1-incubating.apt b/src/site/apt/releasenotes/release-0.6.1-incubating.apt deleted file mode 100644 index 9305214..0000000 --- a/src/site/apt/releasenotes/release-0.6.1-incubating.apt +++ /dev/null @@ -1,110 +0,0 @@ - ----- - Release Notes for 0.6.1-incubating Apache Helix - ----- - -~~ 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 0.6.1-incubating Apache Helix - - The Apache Helix would like to announce the release of Apache Helix 0.6.1-incubating - - This is the second release in 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 - - [] - -* Changes - -** Bug - - * [HELIX-25] - setConfig should check if instance exist or not when setting PARTICIPANT config - - * [HELIX-29] - Not receiving transitions after participant reconnection - - * [HELIX-30] - ZkHelixManager.carryOverPreviousCurrentState() should use a special merge logic - - * [HELIX-34] - Remove watches after the node /resource is deleted - - * [HELIX-35] - Support custom instance id format in CLI - - * [HELIX-41] - fix intermittent test failures - - * [HELIX-44] - ZkHelix property store is not checking the validity of path - - * [HELIX-55] - Session timeout setting not honoured - - * [HELIX-75] - ZKHelixManager declares throws Exception - - * [HELIX-76] - ChangeLogGenerator split bug - - -** Improvements - - * [HELIX-31] - Detect flapping and disable the participant/controller/spectator - - * [HELIX-32] - Flapping detection: if a helix manager starts connect/disconnect frequently it should be disconnected - - * [HELIX-64] - Allow application to provide additional metadata while connecting to cluster - - * [HELIX-73] - Remove assumption that Instance.id is always host_port - - * [HELIX-81] - org.apache.helix.manager.zk.ZKUtil#isClusterSetup() should not log error - - -** New Features - - * [HELIX-19] - Allow process to join the cluster dynamically - - * [HELIX-43] - Add support for error->dropped transition - - * [HELIX-45] - Standalone helix agent - - * [HELIX-63] - Make the idealstate computation code pluggable - - * [HELIX-72] - Allow pluggable rebalancer in controller - - -** Tasks - - * [HELIX-16] - Distributed task execution framework sample app - - - [] - - Cheers, - -- - The Apache Helix Team - - - http://git-wip-us.apache.org/repos/asf/helix/blob/fc59b125/src/site/apt/releasing.apt ---------------------------------------------------------------------- diff --git a/src/site/apt/releasing.apt b/src/site/apt/releasing.apt deleted file mode 100644 index 11d0cd9..0000000 --- a/src/site/apt/releasing.apt +++ /dev/null @@ -1,107 +0,0 @@ - ----- - Helix release process - ----- - ----- - 2012-12-15 - ----- - -~~ 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 - -Helix release process - - [[1]] Post to the dev list a few days before you plan to do an Helix release - - [[2]] Your maven setting must contains the entry to be able to deploy. - - ~/.m2/settings.xml - -+------------- - <server> - <id>apache.releases.https</id> - <username></username> - <password></password> - </server> -+------------- - - [[3]] Apache DAV passwords - -+------------- - Add the following info into your ~/.netrc - machine git-wip-us.apache.org login <apache username> <password> - -+------------- - [[4]] Release Helix - You should have a GPG agent running in the session you will run the maven release commands(preferred), and confirm it works by running "gpg -ab" (type some text and press Ctrl-D). - If you do not have a GPG agent running, make sure that you have the "apache-release" profile set in your settings.xml as shown below. - - Run the release - -+------------- -mvn release:prepare release:perform -B -+------------- - - GPG configuration in maven settings xml: - -+------------- -<profile> - <id>apache-release</id> - <properties> - <gpg.passphrase>[GPG_PASSWORD]</gpg.passphrase> - </properties> -</profile> -+------------- - - [[4]] go to https://repository.apache.org and close your staged repository. Note the repository url (format https://repository.apache.org/content/repositories/orgapachehelix-019/org/apache/helix/helix/0.6-incubating/) - -+------------- -svn co https://dist.apache.org/repos/dist/dev/incubator/helix helix-dev-release -cd helix-dev-release -sh ./release-script-svn.sh version stagingRepoUrl -then svn add <new directory created with new version as name> -then svn ci -+------------- - - [[5]] Validating the release - -+------------- - * Download sources, extract, build and run tests - mvn clean package - * Verify license headers - mvn -Prat -DskipTests - * Download binaries and .asc files - * Download release manager's public key - From the KEYS file, get the release manager's public key finger print and run gpg --keyserver pgpkeys.mit.edu --recv-key <key> - * Validate authenticity of key - run gpg --fingerprint <key> - * Check signatures of all the binaries using gpg <binary> -+------------- - - [[6]] Call for a vote in the dev list and wait for 72 hrs. for the vote results. 3 binding votes are necessary for the release to be finalized. example - After the vote has passed, move the files from dist dev to dist release: svn mv https://dist.apache.org/repos/dist/dev/incubator/helix/version to https://dist.apache.org/repos/dist/release/incubator/helix/ - - [[7]] Prepare release note. Add a page in src/site/apt/releasenotes/ and change value of \<currentRelease> in parent pom. - - - [[8]] Send out an announcement of the release to: - - * [email protected] - - * [email protected] - - [[9]] Celebrate ! - - http://git-wip-us.apache.org/repos/asf/helix/blob/fc59b125/src/site/markdown/Architecture.md ---------------------------------------------------------------------- diff --git a/src/site/markdown/Architecture.md b/src/site/markdown/Architecture.md deleted file mode 100644 index 7acf590..0000000 --- a/src/site/markdown/Architecture.md +++ /dev/null @@ -1,248 +0,0 @@ -<!--- -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. ---> - - -Helix aims to provide the following abilities to a distributed system: - -* Automatic management of a cluster hosting partitioned, replicated resources. -* Soft and hard failure detection and handling. -* Automatic load balancing via smart placement of resources on servers(nodes) based on server capacity and resource profile (size of partition, access patterns, etc). -* Centralized config management and self discovery. Eliminates the need to modify config on each node. -* Fault tolerance and optimized rebalancing during cluster expansion. -* Manages entire operational lifecycle of a node. Addition, start, stop, enable/disable without downtime. -* Monitor cluster health and provide alerts on SLA violation. -* Service discovery mechanism to route requests. - -To build such a system, we need a mechanism to co-ordinate between different nodes/components in the system. This mechanism can be achieved with a software that reacts to any change in the cluster and comes up with a set of tasks needed to bring the cluster to a stable state. The set of tasks will be assigned to one or more nodes in the cluster. Helix serves this purpose of managing the various components in the cluster. - - - -Distributed System Components - -In general any distributed system cluster will have the following - -* Set of nodes also referred to as an instance. -* Set of resources which can be a database, lucene index or a task. -* Each resource is also partitioned into one or more Partitions. -* Each partition may have one or more copies called replicas. -* Each replica can have a state associated with it. For example Master, Slave, Leader, Standby, Online, Offline etc - -Roles ------ - - - -Not all nodes in a distributed system will perform similar functionality. For e.g, a few nodes might be serving requests, few nodes might be sending the request and some nodes might be controlling the nodes in the cluster. Based on functionality we have grouped them into - -We have divided Helix in 3 logical components based on their responsibility - -1. PARTICIPANT: The nodes that actually host the distributed resources. -2. SPECTATOR: The nodes that simply observe the PARTICIPANT State and route the request accordingly. Routers, for example, need to know the Instance on which a partition is hosted and its state in order to route the request to the appropriate end point. -3. CONTROLLER: The controller observes and controls the PARTICIPANT nodes. It is responsible for coordinating all transitions in the cluster and ensuring that state constraints are satisfied and cluster stability is maintained. - - -These are simply logical components and can be deployed as per the system requirements. For example: - -1. Controller can be deployed as a separate service -2. Controller can be deployed along with a Participant but only one Controller will be active at any given time. - -Both have pros and cons, which will be discussed later and one can chose the mode of deployment as per system needs. - - -## Cluster state/metadata store - -We need a distributed store to maintain the state of the cluster and a notification system to notify if there is any change in the cluster state. Helix uses Zookeeper to achieve this functionality. - -Zookeeper provides: - -* A way to represent PERSISTENT state which basically remains until its deleted. -* A way to represent TRANSIENT/EPHEMERAL state which vanishes when the process that created the STATE dies. -* Notification mechanism when there is a change in PERSISTENT/EPHEMERAL STATE - -The namespace provided by ZooKeeper is much like that of a standard file system. A name is a sequence of path elements separated by a slash (/). Every node[ZNODE] in ZooKeeper\'s namespace is identified by a path. - -More info on Zookeeper can be found here http://zookeeper.apache.org - -## Statemachine and constraints - -Even though the concept of Resource, Partition, Replicas is common to most distributed systems, one thing that differentiates one distributed system from another is the way each partition is assigned a state and the constraints on each state. - -For example: - -1. If a system is serving READ ONLY data then all partition\'s replicas are equal and they can either be ONLINE or OFFLINE. -2. If a system takes BOTH READ and WRITES but ensure that WRITES go through only one partition then the states will be MASTER, SLAVE and OFFLINE. Writes go through the MASTER and is replicated to the SLAVES. Optionally, READS can go through SLAVES. - -Apart from defining STATE for each partition, the transition path to each STATE can be application specific. For example, in order to become MASTER it might be a requirement to first become a SLAVE. This ensures that if the SLAVE does not have the data as part of OFFLINE-SLAVE transition it can bootstrap data from other nodes in the system. - -Helix provides a way to configure an application specific state machine along with constraints on each state. Along with constraints on STATE, Helix also provides a way to specify constraints on transitions. (More on this later.) - -``` - OFFLINE | SLAVE | MASTER - _____________________________ - | | | | -OFFLINE | N/A | SLAVE | SLAVE | - |__________|________|_________| - | | | | -SLAVE | OFFLINE | N/A | MASTER | - |__________|________|_________| - | | | | -MASTER | SLAVE | SLAVE | N/A | - |__________|________|_________| - -``` - - - -## Concepts - -The following terminologies are used in Helix to model a state machine. - -* IDEALSTATE: The state in which we need the cluster to be in if all nodes are up and running. In other words, all state constraints are satisfied. -* CURRENTSTATE: Represents the current state of each node in the cluster -* EXTERNALVIEW: Represents the combined view of CURRENTSTATE of all nodes. - -The goal of Helix is always to make the CURRENTSTATE of the system same as the IDEALSTATE. Some scenarios where this may not be true are: - -* When all nodes are down -* When one or more nodes fail -* New nodes are added and the partitions need to be reassigned - -### IDEALSTATE - -Helix lets the application define the IdealState on a resource basis which basically consists of: - -* List of partitions. Example: 64 -* Number of replicas for each partition. Example: 3 -* Node and State for each replica. - -Example: - -* Partition-1, replica-1, Master, Node-1 -* Partition-1, replica-2, Slave, Node-2 -* Partition-1, replica-3, Slave, Node-3 -* ..... -* ..... -* Partition-p, replica-3, Slave, Node-n - -Helix comes with various algorithms to automatically assign the partitions to nodes. The default algorithm minimizes the number of shuffles that happen when new nodes are added to the system - -### CURRENTSTATE - -Every instance in the cluster hosts one or more partitions of a resource. Each of the partitions has a State associated with it. - -Example Node-1 - -* Partition-1, Master -* Partition-2, Slave -* .... -* .... -* Partition-p, Slave - -### EXTERNALVIEW - -External clients needs to know the state of each partition in the cluster and the Node hosting that partition. Helix provides one view of the system to SPECTATORS as EXTERNAL VIEW. EXTERNAL VIEW is simply an aggregate of all CURRENTSTATE - -* Partition-1, replica-1, Master, Node-1 -* Partition-1, replica-2, Slave, Node-2 -* Partition-1, replica-3, Slave, Node-3 -* ..... -* ..... -* Partition-p, replica-3, Slave, Node-n - -## Process Workflow - -Mode of operation in a cluster - -A node process can be one of the following: - -* PARTICIPANT: The process registers itself in the cluster and acts on the messages received in its queue and updates the current state. Example: Storage Node -* SPECTATOR: The process is simply interested in the changes in the Externalview. The Router is a spectator of the Storage cluster. -* CONTROLLER: This process actively controls the cluster by reacting to changes in Cluster State and sending messages to PARTICIPANTS. - - -### Participant Node Process - -* When Node starts up, it registers itself under LIVEINSTANCES -* After registering, it waits for new Messages in the message queue -* When it receives a message, it will perform the required task as indicated in the message -* After the task is completed, depending on the task outcome it updates the CURRENTSTATE - -### Controller Process - -* Watches IDEALSTATE -* Node goes down/comes up or Node is added/removed. Watches LIVEINSTANCES and CURRENTSTATE of each Node in the cluster -* Triggers appropriate state transition by sending message to PARTICIPANT - -### Spectator Process - -* When the process starts, it asks cluster manager agent to be notified of changes in ExternalView -* Whenever it receives a notification, it reads the Externalview and performs required duties. For the Router, it updates its routing table. - -#### Interaction between controller, participant and spectator - -The following picture shows how controllers, participants and spectators interact with each other. - - - -## Core algorithm - -* Controller gets the IdealState and the CurrentState of active storage nodes from Zookeeper -* Compute the delta between IdealState and CurrentState for each partition across all participant nodes -* For each partition compute tasks based on the State Machine Table. It\'s possible to configure priority on the state Transition. For example, in case of Master-Slave: - * Attempt mastership transfer if possible without violating constraint. - * Partition Addition - * Drop Partition -* Add the tasks in parallel if possible to the respective queue for each storage node (if the tasks added are mutually independent) -* If a task is dependent on another task being completed, do not add that task -* After any task is completed by a Participant, Controllers gets notified of the change and the State Transition algorithm is re-run until the CurrentState is same as IdealState. - -## Helix znode layout - -Helix organizes znodes under clusterName in multiple levels. - -The top level (under clusterName) znodes are all Helix defined and in upper case: - -* PROPERTYSTORE: application property store -* STATEMODELDEFES: state model definitions -* INSTANCES: instance runtime information including current state and messages -* CONFIGS: configurations -* IDEALSTATES: ideal states -* EXTERNALVIEW: external views -* LIVEINSTANCES: live instances -* CONTROLLER: cluster controller runtime information - -Under INSTANCES, there are runtime znodes for each instance. An instance organizes znodes as follows: - -* CURRENTSTATES - * sessionId - * resourceName -* ERRORS -* STATUSUPDATES -* MESSAGES -* HEALTHREPORT - -Under CONFIGS, there are different scopes of configurations: - -* RESOURCE: contains resource scope configurations -* CLUSTER: contains cluster scope configurations -* PARTICIPANT: contains participant scope configurations - -The following image shows an example of Helix znodes layout for a cluster named "test-cluster": - - http://git-wip-us.apache.org/repos/asf/helix/blob/fc59b125/src/site/markdown/Concepts.md ---------------------------------------------------------------------- diff --git a/src/site/markdown/Concepts.md b/src/site/markdown/Concepts.md deleted file mode 100644 index 02d7406..0000000 --- a/src/site/markdown/Concepts.md +++ /dev/null @@ -1,268 +0,0 @@ -<!--- -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. ---> - -Helix is based on the idea that a given task has the following attributes associated with it: - -* _Location of the task_. For example it runs on Node N1 -* _State_. For example, it is running, stopped etc. - -In Helix terminology, a task is referred to as a _resource_. - -### IdealState - -IdealState simply allows one to map tasks to location and state. A standard way of expressing this in Helix: - -``` - "TASK_NAME" : { - "LOCATION" : "STATE" - } - -``` -Consider a simple case where you want to launch a task \'myTask\' on node \'N1\'. The IdealState for this can be expressed as follows: - -``` -{ - "id" : "MyTask", - "mapFields" : { - "myTask" : { - "N1" : "ONLINE", - } - } -} -``` -### Partition - -If this task get too big to fit on one box, you might want to divide it into subTasks. Each subTask is referred to as a _partition_ in Helix. Let\'s say you want to divide the task into 3 subTasks/partitions, the IdealState can be changed as shown below. - -\'myTask_0\', \'myTask_1\', \'myTask_2\' are logical names representing the partitions of myTask. Each tasks runs on N1, N2 and N3 respectively. - -``` -{ - "id" : "myTask", - "simpleFields" : { - "NUM_PARTITIONS" : "3", - } - "mapFields" : { - "myTask_0" : { - "N1" : "ONLINE", - }, - "myTask_1" : { - "N2" : "ONLINE", - }, - "myTask_2" : { - "N3" : "ONLINE", - } - } -} -``` - -### Replica - -Partitioning allows one to split the data/task into multiple subparts. But let\'s say the request rate each partition increases. The common solution is to have multiple copies for each partition. Helix refers to the copy of a partition as a _replica_. Adding a replica also increases the availability of the system during failures. One can see this methodology employed often in Search systems. The index is divided into shards, and each shard has multiple copies. - -Let\'s say you want to add one additional replica for each task. The IdealState can simply be changed as shown below. - -For increasing the availability of the system, it\'s better to place the replica of a given partition on different nodes. - -``` -{ - "id" : "myIndex", - "simpleFields" : { - "NUM_PARTITIONS" : "3", - "REPLICAS" : "2", - }, - "mapFields" : { - "myIndex_0" : { - "N1" : "ONLINE", - "N2" : "ONLINE" - }, - "myIndex_1" : { - "N2" : "ONLINE", - "N3" : "ONLINE" - }, - "myIndex_2" : { - "N3" : "ONLINE", - "N1" : "ONLINE" - } - } -} -``` - -### State - -Now let\'s take a slightly complicated scenario where a task represents a database. Unlike an index which is in general read-only, a database supports both reads and writes. Keeping the data consistent among the replicas is crucial in distributed data stores. One commonly applied technique is to assign one replica as MASTER and remaining replicas as SLAVE. All writes go to the MASTER and are then replicated to the SLAVE replicas. - -Helix allows one to assign different states to each replica. Let\'s say you have two MySQL instances N1 and N2, where one will serve as MASTER and another as SLAVE. The IdealState can be changed to: - -``` -{ - "id" : "myDB", - "simpleFields" : { - "NUM_PARTITIONS" : "1", - "REPLICAS" : "2", - }, - "mapFields" : { - "myDB" : { - "N1" : "MASTER", - "N2" : "SLAVE", - } - } -} - -``` - - -### State Machine and Transitions - -IdealState allows one to exactly specify the desired state of the cluster. Given an IdealState, Helix takes up the responsibility of ensuring that the cluster reaches the IdealState. The Helix _controller_ reads the IdealState and then commands the Participant to take appropriate actions to move from one state to another until it matches the IdealState. These actions are referred to as _transitions_ in Helix. - -The next logical question is: how does the _controller_ compute the transitions required to get to IdealState? This is where the finite state machine concept comes in. Helix allows applications to plug in a finite state machine. A state machine consists of the following: - -* State: Describes the role of a replica -* Transition: An action that allows a replica to move from one State to another, thus changing its role. - -Here is an example of MASTERSLAVE state machine, - -``` - OFFLINE | SLAVE | MASTER - _____________________________ - | | | | -OFFLINE | N/A | SLAVE | SLAVE | - |__________|________|_________| - | | | | -SLAVE | OFFLINE | N/A | MASTER | - |__________|________|_________| - | | | | -MASTER | SLAVE | SLAVE | N/A | - |__________|________|_________| - -``` - -Helix allows each resource to be associated with one state machine. This means you can have one resource as an index and another as a database in the same cluster. One can associate each resource with a state machine as follows: - -``` -{ - "id" : "myDB", - "simpleFields" : { - "NUM_PARTITIONS" : "1", - "REPLICAS" : "2", - "STATE_MODEL_DEF_REF" : "MasterSlave", - }, - "mapFields" : { - "myDB" : { - "N1" : "MASTER", - "N2" : "SLAVE", - } - } -} - -``` - -### Current State - -CurrentState of a resource simply represents its actual state at a PARTICIPANT. In the below example: - -* INSTANCE_NAME: Unique name representing the process -* SESSION_ID: ID that is automatically assigned every time a process joins the cluster - -``` -{ - "id":"MyResource" - ,"simpleFields":{ - ,"SESSION_ID":"13d0e34675e0002" - ,"INSTANCE_NAME":"node1" - ,"STATE_MODEL_DEF":"MasterSlave" - } - ,"mapFields":{ - "MyResource_0":{ - "CURRENT_STATE":"SLAVE" - } - ,"MyResource_1":{ - "CURRENT_STATE":"MASTER" - } - ,"MyResource_2":{ - "CURRENT_STATE":"MASTER" - } - } -} -``` -Each node in the cluster has its own CurrentState. - -### External View - -In order to communicate with the PARTICIPANTs, external clients need to know the current state of each of the PARTICIPANTs. The external clients are referred to as SPECTATORS. In order to make the life of SPECTATOR simple, Helix provides an EXTERNALVIEW that is an aggregated view of the current state across all nodes. The EXTERNALVIEW has a similar format as IDEALSTATE. - -``` -{ - "id":"MyResource", - "mapFields":{ - "MyResource_0":{ - "N1":"SLAVE", - "N2":"MASTER", - "N3":"OFFLINE" - }, - "MyResource_1":{ - "N1":"MASTER", - "N2":"SLAVE", - "N3":"ERROR" - }, - "MyResource_2":{ - "N1":"MASTER", - "N2":"SLAVE", - "N3":"SLAVE" - } - } -} -``` - -### Rebalancer - -The core component of Helix is the CONTROLLER which runs the REBALANCER algorithm on every cluster event. Cluster events can be one of the following: - -* Nodes start/stop and soft/hard failures -* New nodes are added/removed -* Ideal state changes - -There are few more such as config changes, etc. The key takeaway: there are many ways to trigger the rebalancer. - -When a rebalancer is run it simply does the following: - -* Compares the IdealState and current state -* Computes the transitions required to reach the IdealState -* Issues the transitions to each PARTICIPANT - -The above steps happen for every change in the system. Once the current state matches the IdealState, the system is considered stable which implies \'IDEALSTATE = CURRENTSTATE = EXTERNALVIEW\' - -### Dynamic IdealState - -One of the things that makes Helix powerful is that IdealState can be changed dynamically. This means one can listen to cluster events like node failures and dynamically change the ideal state. Helix will then take care of triggering the respective transitions in the system. - -Helix comes with a few algorithms to automatically compute the IdealState based on the constraints. For example, if you have a resource of 3 partitions and 2 replicas, Helix can automatically compute the IdealState based on the nodes that are currently active. See the [tutorial](./tutorial_rebalance.html) to find out more about various execution modes of Helix like AUTO_REBALANCE, AUTO and CUSTOM. - - - - - - - - - - - - http://git-wip-us.apache.org/repos/asf/helix/blob/fc59b125/src/site/markdown/Features.md ---------------------------------------------------------------------- diff --git a/src/site/markdown/Features.md b/src/site/markdown/Features.md deleted file mode 100644 index 4f78812..0000000 --- a/src/site/markdown/Features.md +++ /dev/null @@ -1,306 +0,0 @@ -<!--- -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. ---> - - -### 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/fc59b125/src/site/markdown/Quickstart.md ---------------------------------------------------------------------- diff --git a/src/site/markdown/Quickstart.md b/src/site/markdown/Quickstart.md deleted file mode 100644 index 4e09d24..0000000 --- a/src/site/markdown/Quickstart.md +++ /dev/null @@ -1,621 +0,0 @@ -<!--- -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. ---> - -Get Helix ---------- - -First, let\'s get Helix, either build it, or download. - -### Build - - git clone https://git-wip-us.apache.org/repos/asf/incubator-helix.git - cd incubator-helix - mvn install package -DskipTests - cd helix-core/target/helix-core-pkg/bin //This folder contains all the scripts used in following sections - chmod +x * - -### Download - -Download the 0.6.1-incubating release package [here](https://dist.apache.org/repos/dist/dev/incubator/helix/0.6.1-incubating/binaries/helix-core-0.6.1-incubating-pkg.tar) - -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. - -#### First, get to the tools directory - -If you built the code - -``` -cd helix/incubator-helix/helix-core/target/helix-core-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/incubator-helix/helix-core/target/helix-core-pkg/bin -./quickstart.sh -``` - -##### 2 nodes are set up and the partitions 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. - -##### A third node is added and the cluster 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. - -##### 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 the 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/Start zookeeper - -Zookeeper can be started in standalone mode or replicated mode. - -More info 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. (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 - -##### Now we can 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). And the _ideal state_ 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 info of 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", - "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", - "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", - "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. - http://git-wip-us.apache.org/repos/asf/helix/blob/fc59b125/src/site/markdown/Tutorial.md ---------------------------------------------------------------------- diff --git a/src/site/markdown/Tutorial.md b/src/site/markdown/Tutorial.md deleted file mode 100644 index 27f9fd9..0000000 --- a/src/site/markdown/Tutorial.md +++ /dev/null @@ -1,199 +0,0 @@ -<!--- -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. ---> - -# Helix Tutorial - -In this tutorial, we will cover the roles of a Helix-managed cluster, and show the code you need to write to integrate with it. In many cases, there is a simple default behavior that is often appropriate, but you can also customize the behavior. - -Convention: we first cover the _basic_ approach, which is the easiest to implement. Then, we'll describe _advanced_ options, which give you more control over the system behavior, but require you to write more code. - - -### Prerequisites - -1. Read [Concepts/Terminology](./Concepts.html) and [Architecture](./Architecture.html) -2. Read the [Quickstart guide](./Quickstart.html) to learn how Helix models and manages a cluster -3. Install Helix source. See: [Quickstart](./Quickstart.html) for the steps. - -### Tutorial Outline - -1. [Participant](./tutorial_participant.html) -2. [Spectator](./tutorial_spectator.html) -3. [Controller](./tutorial_controller.html) -4. [Rebalancing Algorithms](./tutorial_rebalance.html) -5. [State Machines](./tutorial_state.html) -6. [Messaging](./tutorial_messaging.html) -7. [Customized health check](./tutorial_health.html) -8. [Throttling](./tutorial_throttling.html) -9. [Application Property Store](./tutorial_propstore.html) -10. [Admin Interface](./tutorial_admin.html) - -### Preliminaries - -First, we need to set up the system. Let\'s walk through the steps in building a distributed system using Helix. - -### Start Zookeeper - -This starts a zookeeper in standalone mode. For production deployment, see [Apache Zookeeper](http://zookeeper.apache.org) for instructions. - -``` - ./start-standalone-zookeeper.sh 2199 & -``` - -### Create a cluster - -Creating a cluster will define the cluster in appropriate znodes on zookeeper. - -Using the java API: - -``` - // Create setup tool instance - // Note: ZK_ADDRESS is the host:port of Zookeeper - String ZK_ADDRESS = "localhost:2199"; - admin = new ZKHelixAdmin(ZK_ADDRESS); - - String CLUSTER_NAME = "helix-demo"; - //Create cluster namespace in zookeeper - admin.addCluster(CLUSTER_NAME); -``` - -OR - -Using the command-line interface: - -``` - ./helix-admin.sh --zkSvr localhost:2199 --addCluster helix-demo -``` - - -### Configure the nodes of the cluster - -First we\'ll add new nodes to the cluster, then configure the nodes in the cluster. Each node in the cluster must be uniquely identifiable. -The most commonly used convention is hostname:port. - -``` - String CLUSTER_NAME = "helix-demo"; - int NUM_NODES = 2; - String hosts[] = new String[]{"localhost","localhost"}; - String ports[] = new String[]{7000,7001}; - for (int i = 0; i < NUM_NODES; i++) - { - - InstanceConfig instanceConfig = new InstanceConfig(hosts[i]+ "_" + ports[i]); - instanceConfig.setHostName(hosts[i]); - instanceConfig.setPort(ports[i]); - instanceConfig.setInstanceEnabled(true); - - //Add additional system specific configuration if needed. These can be accessed during the node start up. - instanceConfig.getRecord().setSimpleField("key", "value"); - admin.addInstance(CLUSTER_NAME, instanceConfig); - - } -``` - -### Configure the resource - -A _resource_ represents the actual task performed by the nodes. It can be a database, index, topic, queue or any other processing entity. -A _resource_ can be divided into many sub-parts known as _partitions_. - - -#### Define the _state model_ and _constraints_ - -For scalability and fault tolerance, each partition can have one or more replicas. -The _state model_ allows one to declare the system behavior by first enumerating the various STATES, and the TRANSITIONS between them. -A simple model is ONLINE-OFFLINE where ONLINE means the task is active and OFFLINE means it\'s not active. -You can also specify how many replicas must be in each state, these are known as _constraints_. -For example, in a search system, one might need more than one node serving the same index to handle the load. - -The allowed states: - -* MASTER -* SLAVE -* OFFLINE - -The allowed transitions: - -* OFFLINE to SLAVE -* SLAVE to OFFLINE -* SLAVE to MASTER -* MASTER to SLAVE - -The constraints: - -* no more than 1 MASTER per partition -* the rest of the replicas should be slaves - -The following snippet shows how to declare the _state model_ and _constraints_ for the MASTER-SLAVE model. - -``` - - StateModelDefinition.Builder builder = new StateModelDefinition.Builder(STATE_MODEL_NAME); - - // Add states and their rank to indicate priority. A lower rank corresponds to a higher priority - builder.addState(MASTER, 1); - builder.addState(SLAVE, 2); - builder.addState(OFFLINE); - - // Set the initial state when the node starts - builder.initialState(OFFLINE); - - // Add transitions between the states. - builder.addTransition(OFFLINE, SLAVE); - builder.addTransition(SLAVE, OFFLINE); - builder.addTransition(SLAVE, MASTER); - builder.addTransition(MASTER, SLAVE); - - // set constraints on states. - - // static constraint: upper bound of 1 MASTER - builder.upperBound(MASTER, 1); - - // dynamic constraint: R means it should be derived based on the replication factor for the cluster - // this allows a different replication factor for each resource without - // having to define a new state model - // - builder.dynamicUpperBound(SLAVE, "R"); - - StateModelDefinition statemodelDefinition = builder.build(); - admin.addStateModelDef(CLUSTER_NAME, STATE_MODEL_NAME, myStateModel); -``` - -#### Assigning partitions to nodes - -The final goal of Helix is to ensure that the constraints on the state model are satisfied. -Helix does this by assigning a STATE to a partition (such as MASTER, SLAVE), and placing it on a particular node. - -There are 3 assignment modes Helix can operate on - -* AUTO_REBALANCE: Helix decides the placement and state of a partition. -* AUTO: Application decides the placement but Helix decides the state of a partition. -* CUSTOM: Application controls the placement and state of a partition. - -For more info on the assignment modes, see [Rebalancing Algorithms](./tutorial_rebalance.html) of the tutorial. - -``` - String RESOURCE_NAME = "MyDB"; - int NUM_PARTITIONS = 6; - STATE_MODEL_NAME = "MasterSlave"; - String MODE = "AUTO"; - int NUM_REPLICAS = 2; - - admin.addResource(CLUSTER_NAME, RESOURCE_NAME, NUM_PARTITIONS, STATE_MODEL_NAME, MODE); - admin.rebalance(CLUSTER_NAME, RESOURCE_NAME, NUM_REPLICAS); -``` -
