Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
Kafka, mail # user - Leader election in the failure mode


Copy link to this message
-
Re: Leader election in the failure mode
Vadim Keylis 2013-08-18, 22:19
How often should we run preferred-replica-election tool?
On Sun, Aug 18, 2013 at 3:18 PM, Vadim Keylis <[EMAIL PROTECTED]> wrote:

> Thanks so much Jay for clarifying that.
>
>
> On Thu, Aug 15, 2013 at 8:26 PM, Jay Kreps <[EMAIL PROTECTED]> wrote:
>
>> Leader election is automatic. The preferred-replica-election tool isn't
>> about electing a leader it is about migrating leaderships to ensure each
>> broker is a leader for a fair share of partitions.
>>
>> -Jay
>>
>> On Thu, Aug 15, 2013 at 6:56 PM, Vadim Keylis <[EMAIL PROTECTED]>
>> wrote:
>>
>> > Hello guys. I am not clear about leader election rules in case of server
>> > or service failure. If one of the servers went down that was a leader
>> for
>> > some partitions. Will new leader be elected automatically on server that
>> > has replica or manual intervention is required like running leader
>> election
>> > tool?
>> >  How did you automate the process of running leader election tool in
>> case
>> > leaders are not re-elected automatically in case of failure?
>> >
>> > Thanks,
>> > Vadim
>> >
>> > Sent from my iPhone
>>
>
>