hadoop-hdfs-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Hamed Ghavamnia <ghavamni...@gmail.com>
Subject Re: hbase Multi Node Setup
Date Tue, 03 Jan 2012 09:17:54 GMT
Is it normal that stopping the hbase cluster is really slow?
Is it different to start/stop the hbase service through
/etc/init.d/hadoop-hbase-master or hbase/bin/start-hdfs.sh?

On Tue, Jan 3, 2012 at 8:55 AM, Hamed Ghavamnia <ghavamnia.h@gmail.com>wrote:

> To TS.chia,
> Thanks, but that's for running a hadoop cluster on multiple nodes, which
> isn't the problem, I've already got that running, I want to know how to run
> HBase on that multi-node cluster.
> To Alexander,
> Thanks, I checked them out. I'm really having a hard time understanding
> the relation between zookeeper and hbase, and how to configure the region
> servers and the master. My actual question is, which configs should be done
> on the master and which on the slaves. For example for hdfs, we would set
> the namenode for all the datanodes in the hdfs-site.xml file, but hbase
> doesn't seem to have something like that. Where is the master run, does the
> master automatically run on the system which I run bin/hbase-start.sh?
> On my simple test I'm running, I have a regionserver on one of my machines
> (slave_two), and the zookeeper on another (slave_one), and I run
> bin/start-hbase.sh on my hdfs datanode (master).
> When I start hbase on the master, it starts the zookeeper successfully, it
> can start the regionserver as well, but for the master it says the master
> is already running, but I've stopped hbase on all of the machines (with
> /etc/init.d/hadoop-hbase-master stop).
> On Tue, Jan 3, 2012 at 7:19 AM, Alexander Lorenz <wget.null@googlemail.com
> > wrote:
>> Hi,
>> here two links for:
>> http://wiki.apache.org/hadoop/Hbase/10Minutes
>> https://ccp.cloudera.com/display/CDHDOC/HBase+Installation
>> Install hbase and zookeeper, edit the configs (for single or multinode),
>> pro host a unique ID (myid) for zookeeper, start zookeeper on all nodes as
>> hbase too. Don't forget to identify your servers as master and slaves in
>> hbase-site.xml and setup a path in hdfs (/hbase) as example. Use this path
>> in your config.
>> If the server does not start check the logs for errors.
>> For testing I setup a image with 4 servers, you can download it here:
>> http://mapredit.blogspot.com/p/all-in-one-hadoop-multi-node-appliance.html
>>  - Alex
>> Alexander Lorenz
>> http://mapredit.blogspot.com
>> On Jan 2, 2012, at 5:07 AM, Hamed Ghavamnia <ghavamnia.h@gmail.com>
>> wrote:
>> I've checked that out, but I don't understand what changes should be made
>> to the config files on the hbase-master and the region servers. I've added
>> my region servers(which are the same as my datanodes in hdfs) to the
>> regionservers file, and I've changed the hbase-site.xml configuration file,
>> but the hbase master web UI doesn't show any region servers.
>> On Mon, Jan 2, 2012 at 4:02 PM, Sudharsan Sampath <sudhan65@gmail.com>wrote:
>>> If you are just looking for just steps The HBase Definitive Guide does a
>>> decent brief. I am not sure of any automated tools handy though creating
>>> one based on the steps is straight forward
>>> -Sudhan S
>>> On Mon, Jan 2, 2012 at 5:52 PM, Hamed Ghavamnia <ghavamnia.h@gmail.com>wrote:
>>>> Hi,
>>>> Does anyone have a link for a good tutorial regarding how to setup a
>>>> fully distributed hbase on top of a multi-node hadoop cluster?
>>>> Thanks,
>>>> Hamed

View raw message