From: Marcus Bointon Date: October 1 2012 1:55pm Subject: Re: Replication breaks - slave differs from master List-Archive: http://lists.mysql.com/replication/2386 Message-Id: MIME-Version: 1.0 (Mac OS X Mail 6.1 \(1498\)) Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: quoted-printable On 1 Oct 2012, at 14:33, "Stallmann, Andreas" = wrote: > One single table out of many differs after a while between mysql = master and slave. The statement altering the table is: >=20 > UPDATE SEQ_GEN SET SEQUENCE_VALUE =3D SEQUENCE_VALUE + 10 WHERE = SEQUENCE_NAME =3D 'Client_Object_SEQ' >=20 > We already checked, that no other process but the replication is = altering the slave. >=20 > The logs show, that some transactions are completed in a different = order on the master and the slave. Still, it=92s in no way the obvious = reason for the difference between the two hosts. >=20 > Any ideas? Any (kind) RTFMs? :-) I assume you're using InnoDB? InnoDB should treat that update like a = transaction, so it should be atomic and will lock the row, so there = shouldn't be an opportunity to get out of sync, and as you say, the = order of transactions shouldn't make any difference to this operation = anyway. You could try making the transaction more literal: BEGIN; SELECT * FROM SEQ_GEN WHERE SEQUENCE_NAME =3D 'Client_Object_SEQ' FOR = UPDATE; UPDATE SEQ_GEN SET SEQUENCE_VALUE =3D SEQUENCE_VALUE + 10 WHERE = SEQUENCE_NAME =3D 'Client_Object_SEQ'; COMMIT; Marcus --=20 Marcus Bointon Synchromedia Limited: Creators of http://www.smartmessages.net/ UK info@hand CRM solutions marcus@stripped | http://www.synchromedia.co.uk/