From mboxrd@z Thu Jan 1 00:00:00 1970 From: Oscar Megia Subject: Re: BUG EXT3-fs: fragsize 1024 != blocksize 4096 (unsupported) Date: Tue, 6 Apr 2010 22:16:28 +0200 Message-ID: References: <20100317162439.GC5799@quack.suse.cz> <20100406200552.GF4420@quack.suse.cz> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: linux-fsdevel@vger.kernel.org To: Jan Kara Return-path: Received: from mail-bw0-f209.google.com ([209.85.218.209]:41203 "EHLO mail-bw0-f209.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753224Ab0DFURA convert rfc822-to-8bit (ORCPT ); Tue, 6 Apr 2010 16:17:00 -0400 Received: by bwz1 with SMTP id 1so254366bwz.21 for ; Tue, 06 Apr 2010 13:16:58 -0700 (PDT) In-Reply-To: <20100406200552.GF4420@quack.suse.cz> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: I agree. My problem was flaky power supply. Is good to know it. Thanks for your time, Oscar 2010/4/6 Jan Kara : > On Tue 06-04-10 21:35:09, Oscar Megia wrote: >> Yes, it was a problem with the ATX power supply. I replaced it and n= ow >> is working fine. >> >> I did what you told and I recover my important data but not all fs. >> >> The matter is if you have a hardware fault the ext3 journal fs don't >> preserve your data. >> >> I'm not an expert about the journal ext3 fs but I though is designed >> to preserve data so if the HD has a fault, ext3 must preserve your >> data. If the fault is during updating the main fs it must go to the >> previous main fs. This could happen when there is a power cut. > =C2=A0Ext3 journal prevents filesystem corruption only in case the po= wer > fails suddently. Also cases like kernel crashes are usually solved by= the > journal quite fine. But in situations like flaky power supply where m= emory > can provide wrong data to disk or disk writes only some data because = of > weak voltage, journaling does not help at all. Generally, only good b= ackups > are a solution to bugs in HW... > > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0= =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0Honza >> 2010/3/17 Jan Kara : >> > =C2=A0Hi, >> > >> > On Tue 16-03-10 13:05:09, Oscar Megia wrote: >> >> Since three days ago mi PC with Fedora 9 started to do random res= ets. >> >> >> >> I started to change the BIOS configuration and one of this change= s leaved >> >> the screen black. Unfortunately, I pressed the reset button think= ing that >> >> didn't boot but the next time (after leave the BIOS configuration= like >> >> before) the system didn't boot. >> > =C2=A0I'd check your HW - memory, power supply, ... Obviously some= thing got >> > wrong. >> > >> >> It showed me this message: "EXT3-fs: fragsize 1024 !=3D blocksize= 4096 >> >> (unsupported)" (you can see two pictures booting with this error = at >> >> http://forums.fedoraforum.org/attachment.php?attachmentid=3D19240= &d=3D1268739264 >> >> and http://forums.fedoraforum.org/attachment.php?attachmentid=3D1= 9239&d=3D1268739254). >> >> >> >> Researching I could get info from the root fs with dumpe2fs: >> > ... >> >> >> >> hda1 is for boot and hda2/lvm/ext3 with the OS and=C2=A0 data. If= the ext3 >> >> has journal, is this a bug in the ext3 journal filesystem? >> > =C2=A0I'd say it is a faulty HW, not a software bug. >> > >> >> My question is how can repair this lvm/ext3 volume without loose = data? >> > =C2=A0I would first try to identify faulty HW (run memtest from a = rescue CD >> > for example). When HW gets fixed, I'd copy the filesystem with 'dd= ' to a >> > different disk as a backup. Then run e2fsck to fix /dev/hda2. >> > >> > =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2= =A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0= =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 =C2=A0 = =C2=A0 =C2=A0Honza >> > -- >> > Jan Kara >> > SUSE Labs, CR >> > > -- > Jan Kara > SUSE Labs, CR > -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel= " in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html