Mailing-List: contact user-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hbase.apache.org
Received-SPF: pass (nike.apache.org: domain of mdnguyen@gmail.com designates
 74.125.82.51 as permitted sender)
MIME-Version: 1.0
In-Reply-To: 
 <CAKnDEhgF4Nvkq1jVm5tV3EsAbLrua_7MCprETxJPsNZ5++=13w@mail.gmail.com>
References: 
 <CAKnDEhgF4Nvkq1jVm5tV3EsAbLrua_7MCprETxJPsNZ5++=13w@mail.gmail.com>
From: Minh Duc Nguyen <mdnguyen@gmail.com>
Date: Sun, 22 Jul 2012 12:58:36 -0400
Message-ID: 
 <CAGM+5B7BjCAC_x4zy+K8Pct0JxZLmVF1KD0qQZrW7Bgb22BmNg@mail.gmail.com>
Subject: Re: HBase Cannot Start Again After Sudden Power-Off
To: user@hbase.apache.org, bing.li@asu.edu
Cc: hbase-user@hadoop.apache.org
Content-Type: multipart/alternative; boundary=bcaec53d588777e6a604c56e0b86

--bcaec53d588777e6a604c56e0b86
Content-Type: text/plain; charset=ISO-8859-1

Bing, as far as why HBase is configured as such, it's by design.  The HBase
documentation is very clear in explaining that pseudo-distributed mode is
for "configuration testing and prototyping."

http://hbase.apache.org/book.html#standalone_dist

In order to make it as easy as possible for people to get up and running
with HBase, the default configuration for HBase in pseudo-distributed mode
is to use /tmp.  Of course, when linux restarts, it will clear out the /tmp
directory.  Since you've been running HBase for 4 months now, it may be
worth your while to setup a small HBase cluster in fully-distributed mode.
 While convenient, HBase in pseudo-distributed mode will not be as reliable
as fully-distributed.

   ~ Minh

On Sun, Jul 22, 2012 at 5:29 AM, Bing Li <lblabs@gmail.com> wrote:

> Dear all,
>
> I have used HBase in the pseudo-distributed mode for four months. Last
> night a sudden power-off shut down my HBase machine abnormally. Thereafter,
> HBase could be started.
>
> I had to reinstall my HBase again. This is a very bad experience. It is
> fortunate that my system is in a development stage. If it works for a
> business, that must be a disaster.
>
> I guess the problem is caused because some configuration data is written in
> /tmp. I really DON'T understand why the designers of Hadoop/HBase did that!
> I have to change a lot of configurations to avoid configuration data loss
> in /tmp.
>
> After reinstallation, HBase can be started normally. Then, I shut down
> HBase/Hadoop and turned the machine off.
>
> But, today I got another problem. When starting the system, HBase could not
> be started again! I checked the files in /tmp and noticed that HBase wrote
> some Jetty-related files in it.
>
>    ...
>    drwxr-xr-x  4 libing libing 4096 2012-07-22 16:27
> Jetty_0_0_0_0_50070_hdfs____w2cu08
>    drwxr-xr-x  4 libing libing 4096 2012-07-22 16:27
> Jetty_0_0_0_0_50090_secondary____y6aanv
>    drwxr-xr-x  4 libing libing 4096 2012-07-22 16:27
> Jetty_0_0_0_0_50075_datanode____hwtdwq
>    ...
>
> I didn't see them before. I guess that's the problem why HBase cannot be
> started? If so, could you please tell me how to force HBase to write the
> Jetty-related files to the directories other than /tmp?
>
> Your help is highly appreciated!
>
> Best regards,
> Bing
>

--bcaec53d588777e6a604c56e0b86--