Support Questions
Find answers, ask questions, and share your expertise
Announcements
Alert: Welcome to the Unified Cloudera Community. Former HCC members be sure to read and learn how to activate your account here.

HBase High Availability - Minimum requirements

HBase High Availability - Minimum requirements

Explorer

I've set up a two node cluster using Ambari with HBase installed. I know that we can't achieve Namenode HA without three nodes. But can we make HBase high available with two nodes, by just adding a Standby HBase Master. Does Hbase high availability depend on Namenode high availability?

3 REPLIES 3
Highlighted

Re: HBase High Availability - Minimum requirements

@Thanuja Kularathna

Yes, we can set up HBase HA with 2 nodes. This doesn't depend on the namenode but you need to take care of memory management. Again this is not the best practice for production.

Re: HBase High Availability - Minimum requirements

Guru

HBase high availability is a bit of a complicated topic. See this doc for details:

https://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.6.1/bk_hadoop-high-availability/content/ch_HA-H...

Fundamentally, you need to make configurations in hbase-site. xml and define your tables as HA using CREATE 't1', 'f1', {REGION_REPLICATION => 2}

Regarding namenode HA, most services including HBase depend on that because if a (non-HA) namenode goes down you will have no access to HDFS. You can set up HBase HA without namenode HA, but HBase would not truly be HA because of the above scenario.

Having said all that ... working with only two nodes is extremely minimal. If you intend to do development or evaluation that may be ok and it is usually not worth worrying about HA. If you want to do critical work, HA should be implemented but cluster size also should be well over 2. Many see 8 as absolutely minimal for production.

Re: HBase High Availability - Minimum requirements

You can have active and passive HBase master on two nodes cluster independently of your HDFS configuration.
But if you don't have HA namenode, then you will have single point of failure which can bring your whole HBase cluster down in case of failure as HBase cluster relies on HDFS for storage.

Don't have an account?
Coming from Hortonworks? Activate your account here