This post describes how to do an Express Upgrade of Hortonworks Data Platform (HDP) with Ambari.

Ugrading HDP begins with upgrading Ambari, Ambari Metrics and, not mandatory but recommended, adding Grafana.

When this is in place and all services are up and running, Upgrading HDP to 2.4 can begin.

Backup

File backup

Creating a backup of all the important files and databases is the first step. The following steps are done on the NameNode.

Create backup directory

mkdir /home/ubuntu/HDP-2.3.4-backup

Run HDFS filesystem check and save the ouptut to a file in the backup directory

sudo -u hdfs hdfs fsck / -files -blocks -locations > /home/ubuntu/HDP-2.3.4-backup/dfs-old-fsck-1.log

Gather basic filesystem information and statistics in a report

sudo -u hdfs hdfs dfsadmin -report > /home/ubuntu/HDP-2.3.4-backup/dfs-old-report-1.log

List the whole HDFS directory and save the ouptut to a file

sudo -u hdfs hdfs dfs -ls -R > /home/ubuntu/HDP-2.3.4-backup/dfs-old-lsr-1.log

Enter Safemode, mandatory for next steps

sudo -u hdfs hdfs dfsadmin -safemode enter

Save current namespace and reset edits log

sudo -u hdfs hdfs dfsadmin -saveNamespace

Make a copy of the VERSION file (here is HDP’s default directoy, file VERSION should reside in ${dfs.namenode.name.dir}/current)

sudo cp /hadoop/hdfs/namenode/current/VERSION /home/ubuntu/HDP-2.3.4-backup/

Leave Safemode

sudo -u hdfs hdfs dfsadmin -safemode leave

Finalize upgrade of HDFS
According to the Apache Hadoop documentation:

“Datanodes delete their previous version working directories, followed by Namenode doing the same. This completes the upgrade process.”

sudo -u hdfs hdfs dfsadmin -finalizeUpgrade

Database backup

My cluster has MySql database that is used by Hive and Ranger. That means I have 3 databases to back up: hive, ranger and ranger_audit (since I am storing audit data in a database).

hive

DAT=`date +%Y%m%d_%H%M%S`
mysqldump -u root -proot hive > /home/ubuntu/HDP-2.3.4-backup/hive_$DAT.sql

This is done beforehand so that you can check the checkbox and move on in the process of upgrade

Ranger

This is done beforehand so that you can check the checkbox and move on in the process of upgrade

ranger_admin

DAT=`date +%Y%m%d_%H%M%S`
mysqldump -u root -proot ranger > /home/ubuntu/HDP-2.3.4-backup/ranger_$DAT.sql

ranger_audit

DAT=`date +%Y%m%d_%H%M%S`
mysqldump -u root -proot ranger_audit > /home/ubuntu/HDP-2.3.4-backup/ranger_audit_$DAT.sql

 

Content of backup folder

/home/ubuntu/HDP-2.3.4-backup/
├── dfs-old-fsck-1.log
├── dfs-old-lsr-1.log
├── dfs-old-report-1.log
├── hive_20160804_074811.sql
├── ranger_20160804_074907.sql
├── ranger_audit_20160804_074914.sql
└── VERSION

Click below on Page 2 to continue with the process.