hbase-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Michael Segel <michael_se...@hotmail.com>
Subject RE: Problem starting HBase Master for the first time, EOFException
Date Tue, 06 Jul 2010 14:32:49 GMT


Jamie... to answer your question...

If you change your configuration on the name node, you need to make sure that your data nodes
also know about this. (Assuming you're putting a region server on each data node.)

If you don't add a configuration parameter in the hbase-site.xml, it goes to the default value
in defaults xml file.

You have to remember that the web service is one that reports on the status and is not the
TCP/IP port used for communication between nodes.

HTH

-Mike

> From: jamie.cockrill@gmail.com
> Date: Tue, 6 Jul 2010 13:39:31 +0100
> Subject: Re: Problem starting HBase Master for the first time, EOFException
> To: user@hbase.apache.org
> 
> Dear hbase-users,
> 
> Further to this, I've been doing a lot of kicking around with the
> config and learned the following:
> 
> - You get the same error if a process is listening on that port, that
> is not a HDFS namenode;
> - Although my web front-end for my namenode is on port 50071, it seems
> the namenode itself (i.e. anything that wants hdfs://) is listening on
> 50070.
> 
> Having changed my hbase-site.xml to point to 50070 it now seems to
> have started working. Apologies if anyone has wasted any time looking
> at it. This begs the question, why is my namenode web server  on a
> different port, but I'll get to that another time.
> 
> Many thanks,
> 
> James
> 
> 
> On 6 July 2010 12:19, Jamie Cockrill <jamie.cockrill@gmail.com> wrote:
> > Dear hbase-users,
> >
> > I was wondering if you might be able to help. I am trying to setup
> > HBase (and as such, Zookeeper) on Ubuntu 10.04 using the Cloudera
> > Karmic CDH3 distribution. Zookeeper has installed fine, however when
> > it comes to starting an hbase master, it falls over with the following
> > exception:
> >
> > (stack trace summarised to last bit)
> > Caused by: java.io.EOFException
> >    at java.io.DataInputStream.readInt(DataInputStream.java:375)
> >    at org.apache.hadoop.ipc.Client$Connection.receiveResponse(Client.java:508)
> >    at org.apache.hadoop.ipc.Client$Connection.run(Client.java:446)
> >
> > Its clearly connecting to the HDFS, but it's giving it a strange
> > response. The section above of the trace identifies the hostname that
> > I've specified in the hbase.rootdir property in the hbase-site.xml and
> > also the IP address, which it must have looked up in DNS. For
> > information, that is set to:
> > hdfs://master:50071/hbase.
> >
> > Also, as i'm just evaluating it at this stage, I'm installing
> > Zookeeper and hbase-master on the same machine as the namenode
> > (master). The regionservers will go somewhere else when I get to that
> > stage.
> >
> > My hbase-site.xml was blank (between the configuration tags) and the
> > only things I've added so far are:
> >
> > hbase.cluster.distributed="true"
> > hbase.rootdir="hdfs://master:50071/hbase"
> > hbase.zookeeper.quorum="master"
> >
> > obviously in the <name><value> format of the XML file, without the quotes.
> >
> > I'm at a bit of a loss as to what is going on. I've tried a wget to
> > the namenode dfshealth.asp page and that works fine (obviously thats
> > http:// rather than hdfs://). Any pointers on where to look?
> >
> > Many thanks,
> >
> > Jamie
> >
> > I ran a wget to
> >
 		 	   		  
_________________________________________________________________
The New Busy think 9 to 5 is a cute idea. Combine multiple calendars with Hotmail. 
http://www.windowslive.com/campaign/thenewbusy?tile=multicalendar&ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_5
Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message