Home | About | Sematext search-lucene.com search-hadoop.com
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB
 Search Hadoop and all its subprojects:

Switch to Plain View
HBase >> mail # user >> Can Hbase run on NFS cluster


+
Ling Kun 2012-12-20, 06:04
Copy link to this message
-
RE: Can Hbase run on NFS cluster

Can you check for the logs at the region server side where the region split happens? Paste the logs at this point if some exception happens.
You can see the split code from CompactSpiltThread.java
Mean while not used HBase with NFS. I have not seen yet here in the mailing list some one using HBase with NFS.

-Anoop-
________________________________________
From: Ling Kun [[EMAIL PROTECTED]]
Sent: Thursday, December 20, 2012 11:34 AM
To: dev; user
Subject: Can Hbase run on NFS cluster

Dear all,
   I am doing some test on Hbase running on a NFS cluster. Does anyone have any experience?

   regionserver: slave1, slave2 and slave3 all mount  the same NFS server /mnt/nfs_disk .

   Theoretically, I can use the following configuration in hbase-site.xml to make it work in distributed mode, no matter what the performance is.

--------------------hbase-site.xml ------------------------
   <property>
      <name>hbase.rootdir</name>
      <value>file:///mnt/nfs_disk/hbase/</value>
   </property>
  <property>
    <name>hbase.cluster.distributed</name>
    <value>true</value>
  </property>
------------------hbase-site.xml-------------------------

    According to my testing, it is ok for small data write. While for huge amount of data write, which causes region split, it seems the daughter region can not be online correctly.

   The following is my questions:
   1. Does anyone have setup a hbase cluster on NFS disk experience?
   2. Since the basic hbse operation works fine except the region split and daughter region online.  could some one please show which part of the code is responsible for this type of works?
   3.  When HDFS + Hbase is used, it is easy to get high performance because the scan, read and write operations can be done locally when the hbase cluster is the same as HDFS cluster. Is this correct to explain the performance descending?
   Thanks.

Ling Kun
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB