From mboxrd@z Thu Jan 1 00:00:00 1970 From: Chris Mason Subject: RE: [PATCH CFT] tons of logging patches Date: 04 Jun 2002 12:55:28 -0400 Message-ID: <1023209728.31683.170.camel@tiny> References: <93F527C91A6ED411AFE10050040665D0049BFAF9@corpusmx1.us.dg.com> Mime-Version: 1.0 Content-Transfer-Encoding: 7bit Return-path: list-help: list-unsubscribe: list-post: In-Reply-To: <93F527C91A6ED411AFE10050040665D0049BFAF9@corpusmx1.us.dg.com> List-Id: Content-Type: text/plain; charset="us-ascii" To: berthiaume_wayne@emc.com Cc: reiserfs-list@namesys.com, manuel.krause@mb.tu-ilmenau.de On Tue, 2002-06-04 at 11:48, berthiaume_wayne@emc.com wrote: > Chris, isn't there a danger, remotely, that with data=ordered you > could lose your data? For Ext3FS, if the machine should crash after the > filesystem data is updated but before the metadata is updated, fsck(8) will > reclaim the newly allocated blocks; therefore, the updated file is no longer > correct. Yes, reiserfs does the same. data=ordered only promises that before the updated metadata is on disk, the data blocks they reference will be on disk too. This way you know that if you add new bytes to a file, after a crash you will either have valid data in the file, or no new bytes added at all. The only way you can be sure the actual metadata is there is by waiting for an fsync(file) to finish. -chris