Re: file systems

To: debian-user@lists.debian.org
Subject: Re: file systems
From: Stan Hoeppner <stan@hardwarefreak.com>
Date: Fri, 22 Apr 2011 12:43:23 -0500
Message-id: <[🔎] 4DB1BE3B.2030203@hardwarefreak.com>
In-reply-to: <[🔎] 20110422151809.34bc5ddf@celsius.local>
References: <[🔎] 87wriqjd0t.fsf@towardsfreedom.com> <[🔎] 4DAE0FF0.5070805@hardwarefreak.com> <BANLkTinVLv+z1uJPDU+eJNT76PZ-gY3aiA@mail.gmail.com> <[🔎] 4DAEADF2.30600@hardwarefreak.com> <BANLkTimSKRkJYMHcAV9AkxPM0NwwsYC2qw@mail.gmail.com> <[🔎] BANLkTimexBrLdZQFSpCQ3V299--w74c5WQ@mail.gmail.com> <[🔎] 4DB04070.1030500@affinityvision.com.au> <[🔎] 20110422151809.34bc5ddf@celsius.local>

Arnt Karlsen put forth on 4/22/2011 8:18 AM:

> ..I've had about half a dozen crashes on ext3 where the journal 
> was written to disk but not the data, so the journal trashed my 
> old data "that were gone" according to the journal, on the next 
> fsck.

This is a result of using data=writeback journal mode and has been well
known for a long time.  If you use data=ordered mode (the default on
most distros) the journal isn't written until after the data, which
eliminates the problem you've experienced.

https://ext4.wiki.kernel.org/index.php/Ext3_Data%3DOrdered_vs_Data%3DWriteback_mode

-- 
Stan

Reply to:

Follow-Ups:
- Re: file systems
  - From: Jörg-Volker Peetz <jvpeetz@web.de>

References:
- file systems
  - From: prad <prad@towardsfreedom.com>
- Re: file systems
  - From: Stan Hoeppner <stan@hardwarefreak.com>
- Re: file systems
  - From: Stan Hoeppner <stan@hardwarefreak.com>
- Fwd: file systems
  - From: Heddle Weaver <weaver2world@gmail.com>
- Re: Fwd: file systems
  - From: Andrew McGlashan <andrew.mcglashan@affinityvision.com.au>
- Re: file systems
  - From: Arnt Karlsen <arnt@c2i.net>

Prev by Date: Re: lenny to squeeze dbus-daemon
Next by Date: Re: Using Files Without Mounting A Share From Another System
Previous by thread: Re: file systems
Next by thread: Re: file systems
Index(es):
- Date
- Thread