Apache
/
druid
mirror of https://github.com/apache/druid.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

remove bad docs on setting up clusters (#3188)

This commit is contained in:
Fangjin Yang 2016-07-01 13:41:40 -07:00 committed by Himanshu
parent 99844dfeb5
commit 8eeae2e844
4 changed files with 1 additions and 775 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

360
docs/content/configuration/hadoop.md
View File

@ -1,360 +0,0 @@
---
layout: doc_page
---
Example Production Hadoop Configuration
=======================================
The following configuration should work relatively well for Druid indexing and Hadoop. In the example, we are using Hadoop 2.4 with EC2 m1.xlarge nodes for NameNodes and cc2.8xlarge nodes for DataNodes.
### Core-site.xml
```
<configuration>
<!-- Temporary directory on HDFS (but also sometimes local!) -->
<property>
<name>hadoop.tmp.dir</name>
<value>/mnt/persistent/hadoop</value>
</property>
<!-- S3 -->
<property>
<name>fs.defaultFS</name>
<value>hdfs://#{IP}:9000</value>
</property>
<property>
<name>fs.s3.impl</name>
<value>org.apache.hadoop.fs.s3native.NativeS3FileSystem</value>
</property>
<property>
<name>fs.s3.awsAccessKeyId</name>
<value>#{S3_ACCESS_KEY}</value>
</property>
<property>
<name>fs.s3.awsSecretAccessKey</name>
<value>#{S3_SECRET_KEY}</value>
</property>
<property>
<name>fs.s3.buffer.dir</name>
<value>/mnt/persistent/hadoop-s3n</value>
</property>
<property>
<name>fs.s3n.awsAccessKeyId</name>
<value>#{S3N_ACCESS_KEY}</value>
</property>
<property>
<name>fs.s3n.awsSecretAccessKey</name>
<value>#{S3N_SECRET_KEY}</value>
</property>
<!-- Compression -->
<property>
<name>io.compression.codecs</name>
<value>org.apache.hadoop.io.compress.GzipCodec,org.apache.hadoop.io.compress.DefaultCodec,org.apache.hadoop.io.compress.Lz4Codec,org.apache.hadoop.io.compress.BZip2Codec,org.apache.hadoop.io.compress.SnappyCodec</value>
</property>
<!-- JBOD -->
<property>
<name>io.seqfile.local.dir</name>
<value>/mnt/persistent/hadoop/io/local</value>
</property>
</configuration>
```
### Mapred-site.xml
```
<configuration>
<property>
<name>mapreduce.framework.name</name>
<value>yarn</value>
</property>
<property>
<name>mapreduce.jobtracker.address</name>
<value>#{JT_ADDR}:9001</value>
</property>
<property>
<name>mapreduce.jobtracker.http.address</name>
<value>#{JT_HTTP_ADDR}:9100</value>
</property>
<property>
<name>mapreduce.jobhistory.address</name>
<value>#{JH_ADDR}:10020</value>
</property>
<property>
<name>mapreduce.jobhistory.webapp.address</name>
<value>#{JH_WEBAPP_ADDR}:19888</value>
</property>
<property>
<name>mapreduce.tasktracker.http.address</name>
<value>#{TT_ADDR}:9103</value>
</property>
<!-- Memory and concurrency tuning -->
<property>
<name>mapreduce.job.reduces</name>
<value>21</value>
</property>
<property>
<property>
<name>mapreduce.job.jvm.numtasks</name>
<value>20</value>
</property>
<property>
<name>mapreduce.map.memory.mb</name>
<value>2048</value>
</property>
<property>
<name>mapreduce.map.java.opts</name>
<value>-server -Xmx1536m -Duser.timezone=UTC -Dfile.encoding=UTF-8 -XX:+PrintGCDetails -XX:+PrintGCTimeStamps</value>
</property>
<property>
<name>mapreduce.reduce.memory.mb</name>
<value>6144</value>
</property>
<property>
<name>mapreduce.reduce.java.opts</name>
<value>-server -Xmx2560m -Duser.timezone=UTC -Dfile.encoding=UTF-8 -XX:+PrintGCDetails -XX:+PrintGCTimeStamps</value>
</property>
<property>
<name>mapreduce.reduce.shuffle.parallelcopies</name>
<value>50</value>
</property>
<property>
<name>mapreduce.reduce.shuffle.input.buffer.percent</name>
<value>0.5</value>
</property>
<property>
<name>mapreduce.task.io.sort.mb</name>
<value>256</value>
</property>
<property>
<name>mapreduce.task.io.sort.factor</name>
<value>100</value>
</property>
<property>
<name>mapreduce.jobtracker.handler.count</name>
<value>64</value>
</property>
<property>
<name>mapreduce.tasktracker.http.threads</name>
<value>20</value>
</property>
<!-- JBOD -->
<property>
<name>mapreduce.cluster.local.dir</name>
<value>/mnt/persistent/hadoop/mapred/local</value>
</property>
<!-- Job history server persistent state -->
<property>
<name>mapreduce.jobhistory.recovery.enable</name>
<value>true</value>
</property>
<property>
<name>mapreduce.jobhistory.recovery.store.class</name>
<value>org.apache.hadoop.mapreduce.v2.hs.HistoryServerFileSystemStateStoreService</value>
</property>
<property>
<name>mapreduce.jobhistory.recovery.store.fs.uri</name>
<value>file://${hadoop.tmp.dir}/mapred-jobhistory-state</value>
</property>
<!-- Compression -->
<property>
<!-- Off by default, because it breaks Druid indexing (at least, it does it druid-0.6.10+). Jobs should turn
it on if they need it. -->
<name>mapreduce.output.fileoutputformat.compress</name>
<value>false</value>
</property>
<property>
<name>mapreduce.map.output.compress</name>
<value>true</value>
</property>
<property>
<name>mapreduce.output.fileoutputformat.compress.type</name>
<value>BLOCK</value>
</property>
<property>
<name>mapreduce.map.output.compress.codec</name>
<value>org.apache.hadoop.io.compress.Lz4Codec</value>
</property>
<property>
<name>mapreduce.output.fileoutputformat.compress.codec</name>
<value>org.apache.hadoop.io.compress.GzipCodec</value>
</property>
<!-- Speculative execution would violate various assumptions we've made in our system design -->
<property>
<name>mapreduce.map.speculative</name>
<value>false</value>
</property>
<property>
<name>mapreduce.reduce.speculative</name>
<value>false</value>
</property>
<!-- Sometimes jobs take a long time to run, but really, they're okay. Examples: Long index persists,
hadoop reading lots of empty files into a single mapper. Let's increase the timeout to 30 minutes. -->
<property>
<name>mapreduce.task.timeout</name>
<value>1800000</value>
</property>
</configuration>
```
### Yarn-site.xml
```
<configuration>
<property>
<name>yarn.resourcemanager.hostname</name>
<value>#{RM_HOSTNAME}</value>
</property>
<property>
<name>yarn.resourcemanager.scheduler.class</name>
<value>org.apache.hadoop.yarn.server.resourcemanager.scheduler.capacity.CapacityScheduler</value>
</property>
<property>
<name>yarn.nodemanager.aux-services</name>
<value>mapreduce_shuffle</value>
</property>
<property>
<name>yarn.log-aggregation-enable</name>
<value>true</value>
</property>
<property>
<name>yarn.log.server.url</name>
<value>http://#{IP_LOG_SERVER}:19888/jobhistory/logs/</value>
</property>
<property>
<name>yarn.nodemanager.hostname</name>
<value>#{IP_ADDR}</value>
</property>
<property>
<name>yarn.scheduler.minimum-allocation-mb</name>
<value>512</value>
</property>
<property>
<name>yarn.nodemanager.resource.memory-mb</name>
<value>1024</value>
</property>
<property>
<name>yarn.nodemanager.resource.cpu-vcores</name>
<value>1</value>
</property>
<property>
<name>yarn.nodemanager.vmem-check-enabled</name>
<value>false</value>
</property>
<!-- JBOD -->
<property>
<name>yarn.nodemanager.local-dirs</name>
<value>/mnt/persistent/hadoop/nm-local-dir</value>
</property>
<!-- ResourceManager persistent state doesn't work well in tests yet, so disable it -->
<property>
<name>yarn.resourcemanager.recovery.enabled</name>
<value>false</value>
</property>
<property>
<name>yarn.resourcemanager.store.class</name>
<value>org.apache.hadoop.yarn.server.resourcemanager.recovery.FileSystemRMStateStore</value>
</property>
<property>
<name>yarn.resourcemanager.fs.state-store.uri</name>
<value>file://${hadoop.tmp.dir}/yarn-resourcemanager-state</value>
</property>
<!-- Ability to exclude hosts -->
<property>
<name>yarn.resourcemanager.nodes.exclude-path</name>
<value>/mnt/persistent/hadoop/yarn-exclude.txt</value>
</property>
</configuration>
```
### HDFS-site.xml
```
<configuration>
<property>
<name>dfs.replication</name>
<value>3</value>
</property>
<property>
<name>dfs.namenode.datanode.registration.ip-hostname-check</name>
<value>false</value>
</property>
<property>
<name>dfs.hosts.exclude</name>
<value>/mnt/persistent/hadoop/hdfs-exclude.txt</value>
</property>
<!-- JBOD -->
<property>
<name>dfs.datanode.data.dir</name>
<value>file:///mnt/persistent/hadoop/dfs/data</value>
</property>
</configuration>
```
### Capacity-scheduler.xml
```
<configuration>
<property>
<name>yarn.scheduler.capacity.maximum-am-resource-percent</name>
<value>0.1</value>
</property>
<property>
<name>yarn.scheduler.capacity.root.queues</name>
<value>default</value>
</property>
<property>
<name>yarn.scheduler.capacity.root.default.capacity</name>
<value>100</value>
</property>
<property>
<name>yarn.scheduler.capacity.root.default.user-limit-factor</name>
<value>1</value>
</property>
<property>
<name>yarn.scheduler.capacity.root.queues</name>
<value>default</value>
</property>
<property>
<name>yarn.scheduler.capacity.root.default.maximum-capacity</name>
<value>100</value>
</property>
<property>
<name>yarn.scheduler.capacity.root.default.state</name>
<value>RUNNING</value>
</property>
<property>
<name>yarn.scheduler.capacity.root.default.acl_submit_applications</name>
<value>*</value>
</property>
<property>
<name>yarn.scheduler.capacity.root.default.acl_administer_queue</name>
<value>*</value>
</property>
<property>
<name>yarn.scheduler.capacity.node-locality-delay</name>
<value>-1</value>
</property>
</configuration>
```

404
docs/content/configuration/production-cluster.md
View File

@ -1,404 +0,0 @@
---
layout: doc_page
---
Production Cluster Configuration
================================
<div class="note info">
This configuration is an example of what a production cluster could look like. Many other hardware combinations are
possible! Cheaper hardware is absolutely possible.
</div>
This production Druid cluster assumes that metadata storage and Zookeeper are already set up. The deep storage that is
used for examples is [S3](https://aws.amazon.com/s3/) and [memcached](http://memcached.org/) is used for a distributed cache.
<div class="note info">
The nodes in this example do not need to be on their own individual servers. Overlord and Coordinator nodes should be
co-located on the same hardware.
</div>
The nodes that respond to queries (Historical, Broker, and MiddleManager nodes) will use as many cores as are available,
depending on usage, so it is best to keep these on dedicated machines. The upper limit of effectively utilized cores is
not well characterized yet and would depend on types of queries, query load, and the schema. Historical daemons should
have a heap size of at least 1GB per core for normal usage, but could be squeezed into a smaller heap for testing.
Since in-memory caching is essential for good performance, even more RAM is better.
Broker nodes will use RAM for caching, so they do more than just route queries.
SSDs are highly recommended for Historical nodes when all they have more segments loaded than available memory.
The nodes that are responsible for coordination (Coordinator and Overlord nodes) require much less processing.
The effective utilization of cores by Zookeeper, metadata storage, and Coordinator nodes is likely to be between 1 and 2
for each process/daemon, so these could potentially share a machine with lots of cores. These daemons work with heap
size between 500MB and 1GB.
We'll use [EC2](https://aws.amazon.com/ec2/) r3.8xlarge nodes for query facing nodes and m1.xlarge nodes for coordination nodes.
The following examples work relatively well in production, however, a more optimized tuning for the nodes we selected and
more optimal hardware for a Druid cluster are both definitely possible.
<div class="note caution">
For high availability, there should be at least a redundant copy of every process running on separate hardware.
</div>
### Common Configuration (common.runtime.properties)
```
# Extensions
druid.extensions.loadList=["druid-s3-extensions", "druid-histogram", "mysql-metadata-storage"]
# Zookeeper
druid.zk.service.host=#{ZK_IPs}
druid.zk.paths.base=/druid/prod
druid.discovery.curator.path=/prod/discovery
# Request logging, monitoring, and metrics
druid.request.logging.type=emitter
druid.request.logging.feed=druid_requests
druid.monitoring.monitors=["com.metamx.metrics.JvmMonitor"]
druid.emitter=http
druid.emitter.http.recipientBaseUrl=#{EMITTER_URL}
# Metadata storage
druid.metadata.storage.type=mysql
druid.metadata.storage.connector.connectURI=jdbc:mysql://#{MYSQL_URL}:3306/druid?characterEncoding=UTF-8
druid.metadata.storage.connector.user=#{MYSQL_USER}
druid.metadata.storage.connector.password=#{MYSQL_PW}
# Deep storage
druid.storage.type=s3
druid.s3.accessKey=#{S3_ACCESS_KEY}
druid.s3.secretKey=#{S3_SECRET_KEY}
# Caching
druid.cache.type=memcached
druid.cache.hosts=#{MEMCACHED_IPS}
druid.cache.expiration=2147483647
druid.cache.memcachedPrefix=d1
druid.cache.maxOperationQueueSize=1073741824
druid.cache.readBufferSize=10485760
# Indexing Service Service Discovery
druid.selectors.indexing.serviceName=druid/overlord
# Coordinator Service Discovery
druid.selectors.coordinator.serviceName=druid/prod/coordinator
```
### Overlord Node
Run:
```
io.druid.cli.Main server overlord
```
Hardware:
```
m1.xlarge (Cores: 4, Memory: 15.0 GB)
```
JVM Configuration:
```
-server
-Xmx4g
-Xms4g
-XX:NewSize=256m
-XX:MaxNewSize=256m
-XX:+UseConcMarkSweepGC
-XX:+PrintGCDetails
-XX:+PrintGCTimeStamps
-Duser.timezone=UTC
-Dfile.encoding=UTF-8
-Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager
-Djava.io.tmpdir=/mnt/tmp
```
Runtime.properties:
```
druid.host=#{IP_ADDR}
druid.port=8080
druid.service=druid/overlord
# Only required if you are autoscaling middle managers
druid.indexer.autoscale.doAutoscale=true
druid.indexer.autoscale.strategy=ec2
druid.indexer.autoscale.workerIdleTimeout=PT90m
druid.indexer.autoscale.terminatePeriod=PT5M
druid.indexer.autoscale.workerVersion=#{WORKER_VERSION}
# Upload all task logs to deep storage
druid.indexer.logs.type=s3
druid.indexer.logs.s3Bucket=druid
druid.indexer.logs.s3Prefix=prod/logs/v1
# Run in remote mode
druid.indexer.runner.type=remote
druid.indexer.runner.minWorkerVersion=#{WORKER_VERSION}
# Store all task state in the metadata storage
druid.indexer.storage.type=metadata
```
### MiddleManager Node
Run:
```
io.druid.cli.Main server middleManager
```
Hardware:
```
r3.8xlarge (Cores: 32, Memory: 244 GB, SSD)
```
JVM Configuration:
```
-server
-Xmx64m
-Xms64m
-XX:+UseConcMarkSweepGC
-XX:+PrintGCDetails
-XX:+PrintGCTimeStamps
-Duser.timezone=UTC
-Dfile.encoding=UTF-8
-Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager
-Djava.io.tmpdir=/mnt/tmp
```
Runtime.properties:
```
druid.host=#{IP_ADDR}
druid.port=8080
druid.service=druid/middlemanager
# Store task logs in deep storage
druid.indexer.logs.type=s3
druid.indexer.logs.s3Bucket=#{LOGS_BUCKET}
druid.indexer.logs.s3Prefix=prod/logs/v1
# Resources for peons
druid.indexer.runner.javaOpts=-server -Xmx3g -XX:+UseG1GC -XX:MaxGCPauseMillis=100 -XX:+PrintGCDetails -XX:+PrintGCTimeStamps
druid.indexer.task.baseTaskDir=/mnt/persistent/task/
# Peon properties
druid.indexer.fork.property.druid.monitoring.monitors=["com.metamx.metrics.JvmMonitor"]
druid.indexer.fork.property.druid.processing.buffer.sizeBytes=536870912
druid.indexer.fork.property.druid.processing.numThreads=2
druid.indexer.fork.property.druid.segmentCache.locations=[{"path": "/mnt/persistent/zk_druid", "maxSize": 0}]
druid.indexer.fork.property.druid.server.http.numThreads=50
druid.indexer.fork.property.druid.storage.archiveBaseKey=prod
druid.indexer.fork.property.druid.storage.archiveBucket=aws-prod-druid-archive
druid.indexer.fork.property.druid.storage.baseKey=prod/v1
druid.indexer.fork.property.druid.storage.bucket=druid
druid.indexer.fork.property.druid.storage.type=s3
druid.worker.capacity=9
druid.worker.ip=#{IP_ADDR}
druid.worker.version=#{WORKER_VERSION}
```
### Coordinator Node
Run:
```
io.druid.cli.Main server coordinator
```
Hardware:
```
m1.xlarge (Cores: 4, Memory: 15.0 GB)
```
JVM Configuration:
```
-server
-Xmx10g
-Xms10g
-XX:NewSize=512m
-XX:MaxNewSize=512m
-XX:+UseG1GC
-XX:+PrintGCDetails
-XX:+PrintGCTimeStamps
-Duser.timezone=UTC
-Dfile.encoding=UTF-8
-Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager
-Djava.io.tmpdir=/mnt/tmp
```
Runtime.properties:
```
druid.host=#{IP_ADDR}
druid.port=8080
druid.service=druid/coordinator
```
### Historical Node
Run:
```
io.druid.cli.Main server historical
```
Hardware:
```
r3.8xlarge (Cores: 32, Memory: 244 GB, SSD)
```
JVM Configuration:
```
-server
-Xmx12g
-Xms12g
-XX:NewSize=6g
-XX:MaxNewSize=6g
-XX:MaxDirectMemorySize=32g
-XX:+UseConcMarkSweepGC
-XX:+PrintGCDetails
-XX:+PrintGCTimeStamps
-Duser.timezone=UTC
-Dfile.encoding=UTF-8
-Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager
-Djava.io.tmpdir=/mnt/tmp
```
Runtime.properties:
```
druid.host=#{IP_ADDR}
druid.port=8080
druid.service=druid/historical
druid.historical.cache.useCache=true
druid.historical.cache.populateCache=true
druid.processing.buffer.sizeBytes=1073741824
druid.processing.numThreads=31
druid.server.http.numThreads=50
druid.server.maxSize=300000000000
druid.segmentCache.locations=[{"path": "/mnt/persistent/zk_druid", "maxSize": 300000000000}]
druid.monitoring.monitors=["io.druid.server.metrics.HistoricalMetricsMonitor", "com.metamx.metrics.JvmMonitor"]
```
### Broker Node
Run:
```
io.druid.cli.Main server broker
```
Hardware:
```
r3.8xlarge (Cores: 32, Memory: 244 GB, SSD - this hardware is a bit overkill for the broker but we choose it for simplicity)
```
JVM Configuration:
```
-server
-Xmx25g
-Xms25g
-XX:NewSize=6g
-XX:MaxNewSize=6g
-XX:MaxDirectMemorySize=64g
-XX:+UseConcMarkSweepGC
-XX:+PrintGCDetails
-XX:+PrintGCTimeStamps
-Duser.timezone=UTC
-Dfile.encoding=UTF-8
-Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager
-Djava.io.tmpdir=/mnt/tmp
-Dcom.sun.management.jmxremote.port=17071
-Dcom.sun.management.jmxremote.authenticate=false
-Dcom.sun.management.jmxremote.ssl=false
```
Runtime.properties:
```
druid.host=#{IP_ADDR}
druid.port=8080
druid.service=druid/broker
druid.broker.http.numConnections=20
druid.broker.http.readTimeout=PT5M
druid.processing.buffer.sizeBytes=2147483647
druid.processing.numThreads=31
druid.server.http.numThreads=50
```
### Real-time Node
Run:
```
io.druid.cli.Main server realtime
```
Hardware (this is a little overkill):
```
r3.8xlarge (Cores: 32, Memory: 244 GB, SSD - this hardware is way overkill for the real-time node but we choose it for simplicity)
```
JVM Configuration:
```
-server
-Xmx13g
-Xms13g
-XX:NewSize=2g
-XX:MaxNewSize=2g
-XX:MaxDirectMemorySize=9g
-XX:+UseConcMarkSweepGC
-XX:+PrintGCDetails
-XX:+PrintGCTimeStamps
-XX:+HeapDumpOnOutOfMemoryError
-Duser.timezone=UTC
-Dfile.encoding=UTF-8
-Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager
-Djava.io.tmpdir=/mnt/tmp
-Dcom.sun.management.jmxremote.port=17071
-Dcom.sun.management.jmxremote.authenticate=false
-Dcom.sun.management.jmxremote.ssl=false
```
Runtime.properties:
```
druid.host=#{IP_ADDR}
druid.port=8080
druid.service=druid/realtime
druid.processing.buffer.sizeBytes=1073741824
druid.processing.numThreads=7
druid.server.http.numThreads=50
druid.monitoring.monitors=["io.druid.segment.realtime.RealtimeMetricsMonitor", "com.metamx.metrics.JvmMonitor"]
```

7
docs/content/configuration/zookeeper.md
View File

@ -1,7 +0,0 @@
---
layout: doc_page
---
Production Zookeeper Configuration
==================================
You can find great information about configuring and running Zookeeper [here](https://cwiki.apache.org/confluence/display/KAFKA/Operations#Operations-Zookeeper).

5
docs/content/toc.md
View File

@ -81,10 +81,7 @@ layout: toc
* [Broker](/docs/VERSION/configuration/broker.html)
* [Realtime](/docs/VERSION/configuration/realtime.html)
* [Configuring Logging](/docs/VERSION/configuration/logging.html)
* [Production Cluster Configuration](/docs/VERSION/configuration/production-cluster.html)
* [Production Hadoop Configuration](/docs/VERSION/configuration/hadoop.html)
* [Production Zookeeper Configuration](/docs/VERSION/configuration/zookeeper.html)
## Development
* [Overview](/docs/VERSION/development/overview.html)
* [Libraries](/docs/VERSION/development/libraries.html)