Mailing-List: contact user-help@hbase.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hbase.apache.org
Received-SPF: pass (athena.apache.org: local policy)
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
  s=s1024; d=yahoo.com;
  h=X-YMail-OSG:Received:X-Mailer:References:Message-ID:Date:From:Reply-To:Subject:To:In-Reply-To:MIME-Version:Content-Type;
  b=ZNxfnl0MvW3GK1XVDJ3ezOvyIvFqJzEnorrAl4gJH15+0lrf7qbkstJqdbhsAcgjRH9/mCjxlBLBzVImGF0qRs7d/HxDdSTnm6TkDPpGr02QORoe+1/Dh3Pe1G4Hd1PYm8FtssLjcYGAKkGJrM6nAYf00ahBJ71+1phLzdKj8WI=;
References: 
 <CAH6qSq+iP0AKrcLsNHgL10rYFEoScZEae-XuTLN_XmpdV6ZJgA@mail.gmail.com>
 <F2DB34CE-2FFB-495B-A3D6-D51B464B1E7E@salesforce.com>
 <CAH6qSqKvwiF5VgYzMgH-MOBHip99k7tX6LmHgZ5pM=u-hpN53Q@mail.gmail.com>
 <1329767911.88706.YahooMailNeo@web121705.mail.ne1.yahoo.com>
 <CADcMMgEebdmjyMXkxAMaUCYvqE31WsO-q4iUC+RhfAnDhtbM2g@mail.gmail.com>
 <CAH6qSqJM_PA9j5=MDsd1TvJ0-qUZC=uqEKbZqG7cazGzkhHxsg@mail.gmail.com>
 <1329871128.53618.YahooMailNeo@web121706.mail.ne1.yahoo.com>
Message-ID: <1329874155.37832.YahooMailNeo@web121701.mail.ne1.yahoo.com>
Date: Tue, 21 Feb 2012 17:29:15 -0800 (PST)
From: lars hofhansl <lhofhansl@yahoo.com>
Reply-To: lars hofhansl <lhofhansl@yahoo.com>
Subject: Re: Flushing to HDFS sooner
To: "user@hbase.apache.org" <user@hbase.apache.org>,
  Manuel de Ferran <manuel.deferran@gmail.com>,
  lars hofhansl <lhofhansl@yahoo.com>
In-Reply-To: <1329871128.53618.YahooMailNeo@web121706.mail.ne1.yahoo.com>
MIME-Version: 1.0
Content-Type: multipart/alternative;
 boundary="-887950874-2023108928-1329874155=:37832"

---887950874-2023108928-1329874155=:37832
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

I tried on trunk, and this scenario seems to works fine.=0A=0AIn fact I fir=
st forgot to enable appends (I just switched my local end to hadoop-1.0.0),=
 and without that I did in fact loose the edits.=0AWith appends enables thi=
s works as designed.=0A=0A=0AMight still be a timing issue or only occur in=
 0.90.x.=0A=0AMaybe somebody else could mine the attached logs for clues?=
=0A=0A=0A-- Lars=0A=0A=0A=0A________________________________=0A From: lars =
hofhansl <lhofhansl@yahoo.com>=0ATo: Manuel de Ferran <manuel.deferran@gmai=
l.com>; "user@hbase.apache.org" <user@hbase.apache.org> =0ASent: Tuesday, F=
ebruary 21, 2012 4:38 PM=0ASubject: Re: Flushing to HDFS sooner=0A =0AYou s=
till should not loose data this way.=0ALooks like something that is easily =
reproducible. I'll try with the latest trunk.=0A=0A-- Lars=0A=0A=0A=0A_____=
___________________________=0AFrom: Manuel de Ferran <manuel.deferran@gmail=
.com>=0ATo: user@hbase.apache.org =0ACc: lars hofhansl <lhofhansl@yahoo.com=
> =0ASent: Tuesday, February 21, 2012 3:51 AM=0ASubject: Re: Flushing to HD=
FS sooner=0A=0A=0AOn Mon, Feb 20, 2012 at 9:43 PM, Stack <stack@duboce.net>=
 wrote:=0A=0AOn Mon, Feb 20, 2012 at 11:58 AM, lars hofhansl <lhofhansl@yah=
oo.com> wrote:=0A>> Are there any messages about log replay when you restar=
t the region server?=0A>>=0A>>=0A>=0A>What Lars says', whats it say in the =
logs on master on restart?=0A>St.Ack=0A>=0A=0AI just noticed that if I rest=
art the master before starting the regionserver, I still have my rows and I=
 have "recovered.edits" logs in the master log :=0A=0A=0A2012-02-21 10:09:5=
5,541 INFO org.apache.hadoop.hbase.regionserver.wal.HLogSplitter: Closed pa=
th hdfs://lxc167.nightly-dev.com:9100/hbase/.META./1028785192/recovered.edi=
ts/0000000000000029079=A0(wrote 4 edits in 22ms)=0A=0ADuring my previous te=
sts, I was doing the following :=0A- kill datanode=0A- start datanode=0A- s=
tart regionserver=0A- restart master (because .META. and -ROOT- were the on=
ly regions online): i did not realize that I could loose any data doing tha=
t.=0A=0AHere are the master logs to both cases :=0Ahttp://pastebin.com/KD03=
P0pD : master restart before regionserver=0Ahttp://pastebin.com/FvYaBMdm : =
master restart after regionserver
---887950874-2023108928-1329874155=:37832--