Mailing-List: contact user-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hbase.apache.org
MIME-Version: 1.0
In-Reply-To: 
 <CALte62xFar5xA=Ay+Vq70Q6u+sbOr8eTYJ-RQ8FmeQGn6XkFBQ@mail.gmail.com>
References: 
 <CACCK_aRmKs-9bJQkkRCLGwtDXFsF++Z3g7ib_M2sr3SnHLfSdg@mail.gmail.com>
	<CALte62xFar5xA=Ay+Vq70Q6u+sbOr8eTYJ-RQ8FmeQGn6XkFBQ@mail.gmail.com>
Date: Sun, 2 Aug 2015 21:54:17 +0800
Message-ID: 
 <CACCK_aQRbEBX3kcTGSc8ECe_FJCuX0uFPX_nsBydBrMExC6A8g@mail.gmail.com>
Subject: Re: How does HBase manage Memstore
From: Chang Chen <baibaichen@gmail.com>
To: user@hbase.apache.org
Content-Type: multipart/alternative; boundary=20cf307ac7ab98d3c4051c5463b5

--20cf307ac7ab98d3c4051c5463b5
Content-Type: text/plain; charset=UTF-8

Hi Ted

we are using 0.96, so no hbase.hregion.percolumnfamilyflush.size.lower.bound
 exists.

Besides these options, hbase.regionserver.optionalcacheflushinterval is
also related to flush memstore.
Thanks
Chang


On Sat, Aug 1, 2015 at 5:11 PM, Ted Yu <yuzhihong@gmail.com> wrote:

> Here're some other memstore related config parameters:
>
> hbase.regionserver.global.memstore.size
> hbase.regionserver.global.memstore.size.lower.limit
> hbase.hregion.preclose.flush.size
> hbase.hregion.memstore.block.multiplier
>
> hbase.hregion.percolumnfamilyflush.size.lower.bound
>
> The last one is for per column family flush.
> You can find their meaning in
> either hbase-common/src/main/resources/hbase-default.xml or, in
> http://hbase.apache.org/book.html
>
> FYI
>
> On Sat, Aug 1, 2015 at 1:14 AM, Chang Chen <baibaichen@gmail.com> wrote:
>
> > Hi All
> >
> > How does HBase flush memstore  besides  the configuration in
> > [hbase.hregion.memstore.flush.size]? I believe we set flush size to 64M,
> > but I find many *small* memstore flush logs as below
> >
> > 2015-06-25 10:19:09,243 INFO
> org.apache.hadoop.hbase.regionserver.HRegion:
> > Finished memstore flush of ~69.8 K/71520, currentsize=0/0 for region
> >
> user_intent_exp,1696512044,1426251005088.448ddc0071414b6af5729e4476ea35e8.
> > in 26ms, sequenceid=18149727777, compaction requested=true
> >
> > 2015-06-25 10:20:54,428 INFO
> org.apache.hadoop.hbase.regionserver.HRegion:
> > Finished memstore flush of ~6.9 K/7104, currentsize=0/0 for region
> > user_intent_exp,472446392,1426251005088.8d6041bd68dd431f5840357a2fd8bce2.
> > in 30ms, sequenceid=18149728414, compaction requested=false
> >
> >
> > Is there any other configuration which can affect flush? Will
> HTable.close
> > cause it happen?
> >
> > Thanks
> > Chang
> >
>

--20cf307ac7ab98d3c4051c5463b5--