Mailing-List: contact zookeeper-user-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: zookeeper-user@hadoop.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=Message-ID:X-YMail-OSG:Received:X-Mailer:Date:From:Subject:To:MIME-Version:Content-Type:Content-Transfer-Encoding;
  b=Sb6MRllE3BHotCAm89b02RLQtUIjCnEC7+Q1dArPpnFCgRkCuVcoC+QCCsIIfnnVSr80YRuGcvBJ30PsoSn/smRGFS8OKOZezqlogG2hVftCxPot/eH9HQuCRAVdKsVDDA6ZHvLgzof8EErSuEGmZXrQ5VPitdRw4hxHB/Cxyqw=;
Message-ID: <24288.10584.qm@web32004.mail.mud.yahoo.com>
Date: Fri, 27 Mar 2009 18:49:27 -0700 (PDT)
From: "raghul@yahoo.com" <raghul@yahoo.com>
Subject: Divergence in ZK transaction logs in some corner cases?
To: zookeeper-user@hadoop.apache.org
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable


ZK gurus,=0A=0AI think the ZK transaction logs can diverge from one another=
 in some corner cases. I have one such corner case listed below, could you =
please confirm if my understanding is correct?=0A=0AImagine a 5 srever ense=
mble (A,B,C,D,E). All the servers are @ zxid 9. A is the leader and it star=
ts a new PROPOSAL (@zxid 10). A writes the proposal to the log, so A moves =
to zxid 10. Others haven't received the PROPOSAL yet and A crashes. Now the=
 following happens:=0A=0A1. B is elected as the newleader. B bumps up its i=
n-mem zxid to 10. Since other nodes are at the same zxid, it sends a SNAP s=
o that the others can rebuild their data tree. In-memory zxid of all other =
nodes moves to 10.  =0A=0A2.  A comes back now, it accepts B as the leader =
as soon as the leader (B) and N/2 other nodes vouch for B as the leader. So=
 A joins the ensemble. Every zookeeper node is at zxid 10.=0A=0A3. A new re=
quest is submitted to B. B runs PROPOSAL and COMMIT phases and the cluster =
moves up to zxid 11. But the transaction log of A is different from that of=
 everyone else now. So the transaction logs have diverged.=0A=0ACould you c=
onfirm if this can happen? Or am I reading the code wrong?=0A=0AThanks=0ARa=
ghu=0A=0A=0A