zookeeper-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jordan Zimmerman <jor...@jordanzimmerman.com>
Subject Re: Leader elections between 2 nodes.
Date Thu, 10 May 2018 01:43:47 GMT
Yes. In ZooKeeper reads are local. In Cassandra, for example, reads are quorum usually. 

====================
Jordan Zimmerman

> On May 9, 2018, at 8:40 PM, Prasanth Mathialagan <prasanthmathialagan@gmail.com>
wrote:
> 
> For reads, is Quorum necessary? Reads are served from local the local
> replica, right?
> 
> On Wed, May 9, 2018 at 8:29 AM, Jordan Zimmerman <jordan@jordanzimmerman.com
>> wrote:
> 
>> A quorum means (n/2)+1 - In a 3 node cluster that is 2. So, if 2 servers
>> are up you can achieve quorum.
>> 
>>> we need odd
>>> numbers in order to form a quorum
>> 
>> The "odd" in this formula refers to the number of configured servers not
>> how many are running at a given time. That's the reasoning for quorum in
>> the first place. Quorum I/O ensures that a majority of nodes is referenced
>> for reads or writes (in ZooKeeper writes are what matter for quorum).
>> 
>> -Jordan
>> 
>>> On May 9, 2018, at 10:26 AM, Kaushal Shriyan <kaushalshriyan@gmail.com>
>> wrote:
>>> 
>>> Hi,
>>> 
>>> In a ensemble of 3 nodes with 1 leader and 2 followers, if the leader
>> goes
>>> down then two servers can elect a leader among themselves. But we need
>> odd
>>> numbers in order to form a quorum. I am confused. Please correct me if i
>> am
>>> understanding it wrong.
>>> 
>>> Any help will be highly appreciable
>>> 
>>> Best Regards,
>>> 
>>> Kaushal
>> 
>> 

Mime
View raw message