What to backup? and how? (only metadata, not data)

[#1]

What components and folders to backup (only metadata, not data)? What would be the commands?

- Namenode - Ambari database - Hive Metastore database - Oozie database - Hue database - Ranger database

[#2 · Accepted solution]

@Guilherme Braccialli

Backup metadata Link (dfs.namenode.name.dir - Backup this location)

Backup data too (Falcon for DR, Snapshots)

HCatalog

Ambari DB

Oozie DB

Ranger DB

Hue DB

/etc/*

/usr/hdp/* ( Its good to backup binaries on weekly or month basis)

backup logs for compliance purpose /var/log/*

[#3]

@Guilherme Braccialli / Neeraj Sabharwal

What is the best way to backup the Ranger database 0.5.0(for the purpose of disaster recovery). I see that the REST API is available. Can you please share your experience if you have implemented this in your organization? Thank you.