HPE Ezmeral Software platform

Is there any Split Brain Syndrome Situation in CLDB HA configuration.

 
SOLVED
Go to solution
Ranav
Occasional Visitor

Is there any Split Brain Syndrome Situation in CLDB HA configuration.

HI Team,

I am new to Mapr .. Had knowledge in cloudera distribution. Would like to know how CLDB HA works. How to overcome Split Brain Syndrome Situation if it happens.

 

Regards,

Ranav

 

1 REPLY 1
tdunning
HPE Pro
Solution

Re: Is there any Split Brain Syndrome Situation in CLDB HA configuration.

The Container Location Database (CLDB) does not have a split brain problem.

The CLDB nodes use Zookeeper to detect live instances and to designate a leader. Each instance has a local container (container #1) that is used to persist CLDB state. All updates to CLDB state are sent to the primary CLDB and are persisted to all live replicas. If an instance goes away and then comes back, the local copy of container #1 is resynchronized before that CLDB rejoins as an active instance.

As long as a single valid instance survives, CLDB data is safe. Normally, the system is configured to prevent updates to CLDB state if fewer than 2 instances are active, but this can be set to allow operations to continue with only a single active instance.

 

I work for HPE