You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hbase.apache.org by Demai Ni <ni...@gmail.com> on 2013/09/21 01:14:36 UTC

hbase regionserver fail over -- how long it will take?

hi, folks,

My understanding of hbase within cluster highavailability is to keep 3 copy
of data(dfs.replication, default is 3). Whenever a region server
fails(offlline), another regionserver, which contain another copy of the
data,  will pick up.

Is my understanding correct? In that case, usually how long will take the
failover to complete, so that the data will be available?

Can I get a couple pointers? or on-line doc? many thanks

Demai

Re: hbase regionserver fail over -- how long it will take?

Posted by Demai Ni <ni...@gmail.com>.
Ted, 

Thanks a lot. I will look into it

Demai on the run

On Sep 20, 2013, at 8:44 PM, Ted Yu <yu...@gmail.com> wrote:

> bq. another regionserver, which contain another copy of the data,  will
> pick up.
> 
> Replica of data (HFile and WAL) is stored on data nodes.
> 
> In http://hbase.apache.org/book.html, take a look at 9.7.2.2. Failover
> and 9.7.3. Region-RegionServer Locality
> 
> Cheers
> 
> 
> On Fri, Sep 20, 2013 at 4:14 PM, Demai Ni <ni...@gmail.com> wrote:
> 
>> hi, folks,
>> 
>> My understanding of hbase within cluster highavailability is to keep 3 copy
>> of data(dfs.replication, default is 3). Whenever a region server
>> fails(offlline), another regionserver, which contain another copy of the
>> data,  will pick up.
>> 
>> Is my understanding correct? In that case, usually how long will take the
>> failover to complete, so that the data will be available?
>> 
>> Can I get a couple pointers? or on-line doc? many thanks
>> 
>> Demai
>>

Re: hbase regionserver fail over -- how long it will take?

Posted by Ted Yu <yu...@gmail.com>.
bq. another regionserver, which contain another copy of the data,  will
pick up.

Replica of data (HFile and WAL) is stored on data nodes.

In http://hbase.apache.org/book.html, take a look at 9.7.2.2. Failover
and 9.7.3. Region-RegionServer Locality

Cheers


On Fri, Sep 20, 2013 at 4:14 PM, Demai Ni <ni...@gmail.com> wrote:

> hi, folks,
>
> My understanding of hbase within cluster highavailability is to keep 3 copy
> of data(dfs.replication, default is 3). Whenever a region server
> fails(offlline), another regionserver, which contain another copy of the
> data,  will pick up.
>
> Is my understanding correct? In that case, usually how long will take the
> failover to complete, so that the data will be available?
>
> Can I get a couple pointers? or on-line doc? many thanks
>
> Demai
>