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

Switch to Threaded View
HBase, mail # user - Simple but Reliable configuration for pseudo distributed mode on ubuntu


Copy link to this message
-
Re: Simple but Reliable configuration for pseudo distributed mode on ubuntu
Stack 2013-05-22, 22:59
I can work on the doc. part.  If you have a moment, would suggest filing an
issue w/ snippets of the logs where hbase is lost.

Thanks Stephen,
St.Ack
On Wed, May 22, 2013 at 2:45 PM, Stephen Boesch <[EMAIL PROTECTED]> wrote:

> Hi Stack,
>    This section of the docs is clear. I would suggest it should be included
> in *all *docs not just the standalone section.  I had seen that section of
> the docs before, and it was on my mind..  I had made the change from
> 127.0.1.1 to 127.0.0.1 in the past and it had not worked -but I had
> apparently made the change incorrectly - given that  this time after
> restarting hdfs, mapred, and then hbase everything came up nicely.
>
> In terms of "who could not find whom" the master could not find the region
> servers because  the master had not properly registered with zookeeper. The
> master had  a kind of halfbaked registration in which the /hbase/masters
> node was created but no actual entry inserted. Also the master was unable
> to obtain the region serer info from ZK even though regionserver entries
> were correct: and this was a symptom of the underlying master/zk handshake
> problem.
>
> With a more experienced hbase engineer we worked this out to some extent
> through the hbase provided logs as well as using the zookeeper cli (thus
> the transition from hbase managed zk to standalone zk - with the benefit of
> the zkCli.sh tool)
>
>
> 2013/5/22 Stack <[EMAIL PROTECTED]>
>
> > What did the logs show regards who could not find who? (I would like to
> > answer Jay Vyas but thought I'd ask here first to see if could see what
> > tight-coupling to /etc/hosts we are guilty of).
> >
> > Thanks,
> > St.Ack
> >
> >
> > On Wed, May 22, 2013 at 12:46 PM, Stephen Boesch <[EMAIL PROTECTED]>
> > wrote:
> >
> > > OK found the issue, it was the old ubuntu localhost anomaly of
>  127.0.1.1
> > > vs 127.0.0.1   Changing /etc/hosts to use the latter fixed the issue
> > >
> > > Here is the quick start reference to it
> > >
> > > http://hbase.apache.org/book/quickstart.html
> > > Loopback IP
> > >
> > > HBase expects the loopback IP address to be 127.0.0.1. Ubuntu and some
> > > other distributions, for example, will default to 127.0.1.1 and this
> will
> > > cause problems for you.
> > >
> > > /etc/hosts should look something like this:
> > >
> > >             127.0.0.1 localhost
> > >             127.0.0.1 ubuntu.ubuntu-domain ub
> > >
> > >
> > >
> > > 2013/5/22 Stephen Boesch <[EMAIL PROTECTED]>
> > >
> > > >
> > > > I have not been able to bring up a simple standalone hbase pseuo
> > > > distributed mode cluster on a single ubuntu 12.x machine.  A couple
> of
> > > > co-workers that have been using hbase for years also looked at it and
> > we
> > > > have not been able to resolve.
> > > >
> > > > Before I jump in to the details, i'll just say : "hey is there a
> > pointer
> > > > to a well known configuration that just works?"  I don't have any
> > special
> > > > requirements beyond the following:
> > > >        ubuntu 12.x
> > > >        single machine pseudo distributed
> > > >        preference for hbase 0.92.2 but not a showstopper for newer
> > > versions
> > > >        apache (not cdh, etc) hadoop 1.X (version not important to me
> > > > beyond that)
> > > >
> > > >
> > > > Some more details on the setup:
> > > >
> > > >    using hadoop 1.0.3 .  HDFS and map/reduce been working fine for
> > months
> > > >    hbase 0.92.2
> > > >    zookeeper: have tried both hbase manages zk=true as well as false
> > (and
> > > > with zk 3.3.6)
> > > >    hdfs: using localhost:8020
> > > >    Launching hbase with start-hbase.sh
> > > >          master comes up
> > > >          regiion server comes up
> > > >          region server shows as "region in transition" and never
> > > > completes/never starts serving regions
> > > >          Appears that master does not register properly with ZK -
> > though
> > > > there *is * a new  /hbase/master node in zookeeper , it oes not have
> > any
> > > > entries.
> >