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 yuzhihong@gmail.com designates
 209.85.160.181 as permitted sender)
MIME-Version: 1.0
In-Reply-To: 
 <CABNOTEUEefdrTYWEo9A77UOyRe64H02L4BFnO0ireP4fv_V5Ug@mail.gmail.com>
References: 
 <CABNOTEUEefdrTYWEo9A77UOyRe64H02L4BFnO0ireP4fv_V5Ug@mail.gmail.com>
Date: Thu, 19 Feb 2015 06:55:55 -0800
Message-ID: 
 <CALte62wsuqR+S5o_1pqqvbJvxZmCgx7LXgZUnoWNdzO7T1KYmQ@mail.gmail.com>
Subject: Re: HBase mttr
From: Ted Yu <yuzhihong@gmail.com>
To: "user@hbase.apache.org" <user@hbase.apache.org>
Content-Type: multipart/alternative; boundary=001a1133d5661177fe050f722212

--001a1133d5661177fe050f722212
Content-Type: text/plain; charset=UTF-8

Have you read '9.3.4. Better Mean Time to Recover (MTTR)' under
http://hbase.apache.org/book.html#important_configurations ?

For #2, can you take a look at
https://issues.apache.org/jira/browse/HBASE-10070 ?

Cheers

On Thu, Feb 19, 2015 at 12:46 AM, Serega Sheypak <serega.sheypak@gmail.com>
wrote:

> Hi, we are running HBase on super-low-cost HW :)
> Sometimes random node goes down. And HBase needs a time to move regions
> from failed RS.
>
> What are the practices to:
> 1. minimize MTTR?
> 2. is there any possibility to gracefully handle situation when region is
> not accessible for r/w?
> I can just drop the data, but I can't wait seconds or minutes for response.
> I would like to break request after 100ms and return empty result.
>

--001a1133d5661177fe050f722212--