source: gs3-extensions/maori-lang-detection/MoreReading/Vagrant-Spark-Hadoop.txt@ 33443

To run firefox/anything graphical inside the VM run by vagrant, have to ssh -Y onto both analytics and then to the vagrant VM from analytics:
1. ssh analytics -Y
2. [anupama@analytics vagrant-hadoop-hive-spark]$ vagrant ssh -- -Y
or
vagrant ssh -- -Y node1
(the -- flag tells the vagrant command that the subsequent -Y flag should be passed to the ssh cmd that vagrant runs)

Only once ssh-ed with vagrant into the VM whose hostname is "node1", do you have access to node1's assigned IP: 10.211.55.101
- Connecting machines, like analytics, must access node1 or use port forwarding to view the VM's servers on localhost. For example, on analytics, can view Yarn pages at http://localhost:8088/
- If firefox is launched inside the VM (so inside node1), then can access pages off their respective ports at any of localhost|10.211.55.101|node1.

-------------------------


WET example from https://github.com/commoncrawl/cc-warc-examples

vagrant@node1:~/cc-warc-examples$ hdfs dfs -mkdir /user/vagrant/data
vagrant@node1:~/cc-warc-examples$ hdfs dfs -put data/CC-MAIN-20190715175205-20190715200159-00000.warc.wet.gz hdfs:///user/vagrant/data/.
vagrant@node1:~/cc-warc-examples$ hdfs dfs -ls data
Found 1 items
-rw-r--r--   1 vagrant supergroup  154823265 2019-08-19 08:23 data/CC-MAIN-20190715175205-20190715200159-00000.warc.wet.gz
vagrant@node1:~/cc-warc-examples$ hadoop jar target/cc-warc-examples-0.3-SNAPSHOT-jar-with-dependencies.jar org.commoncrawl.examples.mapreduce.WETWordCount

<ONCE FINISHED:>

vagrant@node1:~/cc-warc-examples$ hdfs dfs -cat /tmp/cc/part*



INFO ON HADOOP/HDFS:
https://www.bluedata.com/blog/2016/08/hadoop-hdfs-upgrades-painful/

SPARK:
configure option example: https://stackoverflow.com/questions/46152526/how-should-i-configure-spark-to-correctly-prune-hive-metastore-partitions



LIKE '%isl%'

cd cc-index-table
APPJAR=target/cc-spark-0.2-SNAPSHOT-jar-with-dependencies.jar
> $SPARK_HOME/bin/spark-submit \
#    $SPARK_ON_YARN \
   --conf spark.hadoop.parquet.enable.dictionary=true \
   --conf spark.hadoop.parquet.enable.summary-metadata=false \
   --conf spark.sql.hive.metastorePartitionPruning=true \
   --conf spark.sql.parquet.filterPushdown=true \
   --conf spark.sql.parquet.mergeSchema=true \
   --class org.commoncrawl.spark.examples.CCIndexWarcExport $APPJAR \
   --query "SELECT url, warc_filename, warc_record_offset, warc_record_length
            FROM ccindex
            WHERE crawl = 'CC-MAIN-2019-30' AND subset = 'warc' AND content_languages LIKE '%mri%'" \
   --numOutputPartitions 12 \
   --numRecordsPerWarcFile 20000 \
   --warcPrefix ICELANDIC-CC-2018-43 \
   s3://commoncrawl/cc-index/table/cc-main/warc/ \
   .../my_output_path/


----
TIME
----
1. https://dzone.com/articles/need-billions-of-web-pages-dont-bother-crawling
http://digitalpebble.blogspot.com/2017/03/need-billions-of-web-pages-dont-bother_29.html

"So, not only have CommonCrawl given you loads of web data for free, they’ve also made your life easier by preprocessing the data for you. For many tasks, the content of the WAT or WET files will be sufficient and you won’t have to process the WARC files.

This should not only help you simplify your code but also make the whole processing faster. We recently ran an experiment on CommonCrawl where we needed to extract anchor text from HTML pages. We initially wrote some MapReduce code to extract the binary content of the pages from their WARC representation, processed the HTML with JSoup and reduced on the anchor text. Processing a single WARC segment took roughly 100 minutes on a 10-node EMR cluster. We then simplified the extraction logic, took the WAT files as input and the processing time dropped to 17 minutes on the same cluster. This gain was partly due to not having to parse the web pages, but also to the fact that WAT files are a lot smaller than their WARC counterparts."

2. https://spark-in.me/post/parsing-common-crawl-in-two-simple-commands
"Both the parsing part and the processing part take just a couple of minutes per index file / WET file - the bulk of the “compute” lies within actually downloading these files.

Essentially if you have some time to spare and an unlimited Internet connection, all of this processing can be done on one powerful machine. You can be fancy and go ahead and rent some Amazon server(s) to minimize the download time, but that can be costly.

In my experience - parsing the whole index for Russian websites (just filtering by language) takes approximately 140 hours - but the majority of this time is just downloading (my speed averaged ~300-500 kb/s)."

=========================================================
Configuring spark to work on Amazon AWS s3a dataset:
=========================================================
https://stackoverflow.com/questions/30385981/how-to-access-s3a-files-from-apache-spark
http://deploymentzone.com/2015/12/20/s3a-on-spark-on-aws-ec2/
https://answers.dataiku.com/1734/common-crawl-s3
https://stackoverflow.com/questions/2354525/what-should-be-hadoop-tmp-dir
https://stackoverflow.com/questions/40169610/where-exactly-should-hadoop-tmp-dir-be-set-core-site-xml-or-hdfs-site-xml

https://stackoverflow.com/questions/43759896/spark-truncated-the-string-representation-of-a-plan-since-it-was-too-large-w


https://sparkour.urizone.net/recipes/using-s3/
Configuring Spark to Use Amazon S3
"Some Spark tutorials show AWS access keys hardcoded into the file paths. This is a horribly insecure approach and should never be done. Use exported environment variables or IAM Roles instead, as described in Configuring Amazon S3 as a Spark Data Source."

"No FileSystem for scheme: s3n

java.io.IOException: No FileSystem for scheme: s3n

This message appears when dependencies are missing from your Apache Spark distribution. If you see this error message, you can use the --packages parameter and Spark will use Maven to locate the missing dependencies and distribute them to the cluster. Alternately, you can use --jars if you manually downloaded the dependencies already. These parameters also works on the spark-submit script."

===========================================
IAM Role (or user) and commoncrawl profile
===========================================

"iam" role or user for commoncrawl(er) profile


aws management console:
[email protected]
lab pwd, capital R and ! (maybe g)

commoncrawl profile created while creating the user/role, by following the instructions at: https://answers.dataiku.com/1734/common-crawl-s3

<!--
    <property>
      <name>fs.s3a.awsAccessKeyId</name>
      <value>XXX</value>
    </property>
    <property>
      <name>fs.s3a.awsSecretAccessKey</name>
      <value>XXX</value>
    </property>
-->


But instead of putting the access and secret keys in hadoop's core-site.xml as above (with sudo emacs /usr/local/hadoop-2.7.6/etc/hadoop/core-site.xml)

you'll want to put the Amazon AWS access key and secret key in the spark properties file:

       sudo emacs /usr/local/spark-2.3.0-bin-hadoop2.7/conf/spark-defaults.conf


The spark properties should contain:

spark.hadoop.fs.s3a.impl=org.apache.hadoop.fs.s3a.S3AFileSystem  
spark.hadoop.fs.s3a.access.key=ACCESSKEY  
spark.hadoop.fs.s3a.secret.key=SECRETKEY  



When the job is running, can visit the Spark Context at http://node1:4040/jobs/ (http://node1:4041/jobs/ for me, since I forwarded the vagrant VM's ports at +1)

-------------

APPJAR=target/cc-spark-0.2-SNAPSHOT-jar-with-dependencies.jar
$SPARK_HOME/bin/spark-submit \
    --conf spark.hadoop.parquet.enable.dictionary=true \
    --conf spark.hadoop.parquet.enable.summary-metadata=false \
    --conf spark.sql.hive.metastorePartitionPruning=true \
    --conf spark.sql.parquet.filterPushdown=true \
    --conf spark.sql.parquet.mergeSchema=true \
    --class org.commoncrawl.spark.examples.CCIndexExport $APPJAR \
        --query "SELECT url, warc_filename, warc_record_offset, warc_record_length   
            FROM ccindex                                                        
            WHERE crawl = 'CC-MAIN-2019-30' AND subset = 'warc' AND content_languages LIKE '%mri%'" \
    --outputFormat csv \
    --numOutputPartitions 10 \
    --outputCompression gzip \
    s3://commoncrawl/cc-index/table/cc-main/warc/ \
    hdfs:///user/vagrant/cc-mri-csv

----------------
Exception in thread "main" java.lang.RuntimeException: java.lang.ClassNotFoundException: Class org.apache.hadoop.fs.s3a.S3AFileSystem not found


https://stackoverflow.com/questions/39355354/spark-no-filesystem-for-scheme-https-cannot-load-files-from-amazon-s3
https://stackoverflow.com/questions/33356041/technically-what-is-the-difference-between-s3n-s3a-and-s3
"2018-01-10 Update Hadoop 3.0 has cut its s3: and s3n implementations: s3a is all you get. It is now significantly better than its predecessor and performs as least as good as the Amazon implementation. Amazon's "s3:" is still offered by EMR, which is their closed source client. Consult the EMR docs for more info."

1. https://stackoverflow.com/questions/30385981/how-to-access-s3a-files-from-apache-spark

"Having experienced first hand the difference between s3a and s3n - 7.9GB of data transferred on s3a was around ~7 minutes while 7.9GB of data on s3n took 73 minutes [us-east-1 to us-west-1 unfortunately in both cases; Redshift and Lambda being us-east-1 at this time] this is a very important piece of the stack to get correct and it's worth the frustration.

Here are the key parts, as of December 2015:

    Your Spark cluster will need a Hadoop version 2.x or greater. If you use the Spark EC2 setup scripts and maybe missed it, the switch for using something other than 1.0 is to specify --hadoop-major-version 2 (which uses CDH 4.2 as of this writing).

    You'll need to include what may at first seem to be an out of date AWS SDK library (built in 2014 as version 1.7.4) for versions of Hadoop as late as 2.7.1 (stable): aws-java-sdk 1.7.4. As far as I can tell using this along with the specific AWS SDK JARs for 1.10.8 hasn't broken anything.

    You'll also need the hadoop-aws 2.7.1 JAR on the classpath. This JAR contains the class org.apache.hadoop.fs.s3a.S3AFileSystem.

    In spark.properties you probably want some settings that look like this:

    spark.hadoop.fs.s3a.impl=org.apache.hadoop.fs.s3a.S3AFileSystem  
    spark.hadoop.fs.s3a.access.key=ACCESSKEY  
    spark.hadoop.fs.s3a.secret.key=SECRETKEY

I've detailed this list in more detail on a post I wrote as I worked my way through this process. In addition I've covered all the exception cases I hit along the way and what I believe to be the cause of each and how to fix them."


2. The classpath used by hadoop can be found by running the command (https://stackoverflow.com/questions/26748811/setting-external-jars-to-hadoop-classpath):
    hadoop classpath


3. Got hadoop-aws 2.7.6 jar
from https://mvnrepository.com/artifact/org.apache.hadoop/hadoop-aws/2.7.6
and put it into /home/vagrant


4. https://stackoverflow.com/questions/26748811/setting-external-jars-to-hadoop-classpath
https://stackoverflow.com/questions/28520821/how-to-add-external-jar-to-hadoop-job/54459211#54459211
vagrant@node1:~$ export LIBJARS=/home/vagrant/hadoop-aws-2.7.6.jar
vagrant@node1:~$ export HADOOP_CLASSPATH=`echo ${LIBJARS} | sed s/,/:/g`
vagrant@node1:~$ hadoop classpath

5. https://community.cloudera.com/t5/Community-Articles/HDP-2-4-0-and-Spark-1-6-0-connecting-to-AWS-S3-buckets/ta-p/245760
"Download the aws sdk for java https://aws.amazon.com/sdk-for-java/ Uploaded it to the hadoop directory. You should see the aws-java-sdk-1.10.65.jar in /usr/hdp/2.4.0.0-169/hadoop/"

I got version 1.11

[Can't find a spark.properties file, but this seems to contain spark specific properties:
$SPARK_HOME/conf/spark-defaults.conf

https://jaceklaskowski.gitbooks.io/mastering-apache-spark/spark-properties.html
"The default Spark properties file is $SPARK_HOME/conf/spark-defaults.conf that could be overriden using spark-submit with the --properties-file command-line option."]

Can SUDO COPY the 2 jar files hadoop-aws-2.7.6.jar and aws-java-sdk-1.11.616.jar to:
/usr/local/hadoop/share/hadoop/common/
(else /usr/local/hadoop/share/hadoop/hdfs/hadoop-aws-2.7.6.jar)

--------
schema
https://commoncrawl.s3.amazonaws.com/cc-index/table/cc-main/index.html

---------------
More examples to try:
https://github.com/commoncrawl/cc-warc-examples


A bit outdated?
https://www.journaldev.com/20342/apache-spark-example-word-count-program-java
https://www.journaldev.com/20261/apache-spark

--------

sudo apt-get install maven
(or sudo apt update
sudo apt install maven)
git clone https://github.com/commoncrawl/cc-index-table.git
cd cc-index-table
mvn package
vagrant@node1:~/cc-index-table$ ./src/script/convert_url_index.sh https://commoncrawl.s3.amazonaws.com/cc-index/collections/CC-MAIN-2019-30/indexes/cdx-00000.gz hdfs:///user/vagrant/cc-index-table




spark:
https://jaceklaskowski.gitbooks.io/mastering-apache-spark/spark-shell.html

============
Dr Bainbridge found the following vagrant file that will set up hadoop and spark presumably for cluster computing:

https://github.com/martinprobson/vagrant-hadoop-hive-spark

Vagrant:
    * Guide: https://www.vagrantup.com/intro/getting-started/index.html
    * Common cmds: https://blog.ipswitch.com/5-vagrant-commands-you-need-to-know
    * vagrant reload = vagrant halt + vagrant up https://www.vagrantup.com/docs/cli/reload.html
    * https://stackoverflow.com/questions/46903623/how-to-use-firefox-ui-in-vagrant-box
    * https://stackoverflow.com/questions/22651399/how-to-install-firefox-in-precise64-vagrant-box
      sudo apt-get -y install firefox
    * vagrant install emacs: https://medium.com/@AnnaJS15/getting-started-with-virtualbox-and-vagrant-8d98aa271d2a

    * hadoop conf: sudo vi /usr/local/hadoop-2.7.6/etc/hadoop/mapred-site.xml
    * https://data-flair.training/forums/topic/mkdir-cannot-create-directory-data-name-node-is-in-safe-mode/
---
==> node1: Forwarding ports...
    node1: 8080 (guest) => 8081 (host) (adapter 1)
    node1: 8088 (guest) => 8089 (host) (adapter 1)
    node1: 9083 (guest) => 9084 (host) (adapter 1)
    node1: 4040 (guest) => 4041 (host) (adapter 1)
    node1: 18888 (guest) => 18889 (host) (adapter 1)
    node1: 16010 (guest) => 16011 (host) (adapter 1)
    node1: 22 (guest) => 2200 (host) (adapter 1)
==> node1: Running 'pre-boot' VM customizations...


==> node1: Checking for guest additions in VM...
    node1: The guest additions on this VM do not match the installed version of
    node1: VirtualBox! In most cases this is fine, but in rare cases it can
    node1: prevent things such as shared folders from working properly. If you see
    node1: shared folder errors, please make sure the guest additions within the
    node1: virtual machine match the version of VirtualBox you have installed on
    node1: your host and reload your VM.
    node1: 
    node1: Guest Additions Version: 5.1.38
    node1: VirtualBox Version: 5.2

------------