linux-raid.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: PFC <lists@peufeu.com>
To: Mitchell Laks <mlaks@verizon.net>,
	"linux-raid@vger.kernel.org" <linux-raid@vger.kernel.org>
Subject: Re: Kanotix crashed my raid...
Date: Mon, 09 Jan 2006 19:30:25 +0100	[thread overview]
Message-ID: <op.s24i0zf0cigqcu@apollo13> (raw)
In-Reply-To: <200601090803.03588.mlaks@verizon.net>

>> 	OK, I bit the bullet and removed the "goto abort" in raid5.c
>>
>> 	I was then able to mount everything and recover all of my data without
>> any problem. Hm.
>>
>> 	There should be a way to do this with mdadm without recompiling the
>> kernel, but anyway, opensource saved my ass xDDD
> Could you explain to me what you did? It sounds very important for me to
> understand!

	Well, I ought to update.

	So the kernel refused to start my raid because it was dirty and degraded.  
I understand this, but getting some data out is better than none. I knew  
that the PC had crashed while starting the array. So, it wouldn't have had  
much time to cause a lot of corruption. Most likely everything was  
alright, just marked dirty. So I just removed the test in the kernel which  
refuses to start the array in this condition (that's why I love  
opensource). And it worked.

	raid5.c in the 2.6.14 kernel :

if (mddev->degraded == 1 &&
     mddev->recovery_cp != MaxSector) {
printk(KERN_ERR "raid5: cannot start dirty degraded array for  
%s",mdname(mddev));
goto abort;
}

	I just commented out the goto abort.

	Anyway, that's not the end ! I got 3 harddisk failures in a week. All  
maxtor 250G SATA HDDs. I RMA'd the first one and got a new drive. It  
worked for 2 days and then boom. So I went to all the computer shops  
around here and they only had maxtor so I bought another maxtor. AND  
YESTERDAY IT DIED TOO.

	I spent a lot of time with google, and :	

	Kanotix was not the culprit ; it's nvidia. Turns out the nforce 3 and 4  
chipsets have some SATA problems. This might be a hardware issue, or maybe  
a driver issue, who knows, but the end result is this :

	- My nforce 3 / Athlon 64 PC running Linux has 4 SATA ports (2x 2 ports)
	- 2 of these ports (sda and sdb) are compatible with maxtor harddrives  
(ie. it seems it works)
	- the other two (sdc and sdd) are not compatible with maxtor drives.

	My other computer, which has a nforce4 mobo / Athlon 64 PC and runs  
windows, has the exactly same problem !

	- If I plug a seagate harddrive as sda,b,c,d, it works.
	- If I plug a maxtor harddrive as sda or sdb, it works.
	- If I plug a maxtor harddrive as sdc or sdd (the other sata ports), on  
linux it works for a day or two then the drive "dies" ; on windows it just  
fucks everything up (takes forever to boot, disk management console  
crashes, mouse freezes, disk exists then disappears, etc).

	Re-plugging the "dead" drive as sda or sdb makes it work ! So now all is  
well. I have plugged my 2 maxtor drives on the "maxtorphile" sata sockets,  
and the 2 seagate drives on the "maxtorphobe" sata sockets. Everything  
works. I feel like banging head against wall.

	Oh yeah, on linux I also deactivated USB, firewire, and unplugged teh  
CDROM drive. Who knows. It seems to work now. At least it has worked for  
24 hours now.

	Actually it's quite hallucinating. And I found people in forums with the  
same experience ! Some guy had to unplug his IDE CDR to get his SATA hdd  
to work. On windows. Cool. Actually sata + nforce = broken.

	(the "maxtorphile" sata sockets are not driven by the nforce chipset but  
by a SIS chip, or so it seems).

	DON'T BUY NFORCE FOR MAKING SATA RAID !!!












  parent reply	other threads:[~2006-01-09 18:30 UTC|newest]

Thread overview: 29+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <fd8d0180601050104x15079396h@mail.gmail.com>
2006-01-05  9:06 ` Fwd: Linux MD raid5 and reiser4... Any experience ? Francois Barre
2006-01-05 10:14   ` Daniel Pittman
2006-01-05 11:21     ` Francois Barre
2006-01-05 11:31       ` Gordon Henderson
2006-01-06  6:33       ` Daniel Pittman
2006-01-06  9:47       ` Simon Valiquette
2006-01-06 10:50         ` Francois Barre
2006-01-06 19:28           ` Forrest Taylor
2006-01-06 11:03         ` Kanotix crashed my raid PFC
2006-01-06 12:02           ` PFC
2006-01-06 12:08             ` PFC
2006-01-06 22:01               ` PFC
     [not found]                 ` <200601090803.03588.mlaks@verizon.net>
2006-01-09 18:30                   ` PFC [this message]
2006-01-06 19:05         ` Fwd: Linux MD raid5 and reiser4... Any experience ? Mike Hardy
2006-01-08  2:53         ` Daniel Pittman
2006-01-05 11:26     ` berk walker
2006-01-05 11:35       ` Francois Barre
2006-01-05 11:43         ` Gordon Henderson
2006-01-05 11:59         ` berk walker
2006-01-05 13:13         ` Bill Rugolsky Jr.
2006-01-05 13:38   ` John Stoffel
2006-01-05 14:03     ` Francois Barre
2006-01-05 18:55       ` John Stoffel
2006-01-06  9:08         ` Francois Barre
2006-01-06 10:49           ` Andre Majorel
2006-01-09  8:00             ` Molle Bestefich
2006-01-09  8:16               ` Gordon Henderson
2006-01-09  9:00                 ` Francois Barre
2006-01-09  9:24                   ` Molle Bestefich

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=op.s24i0zf0cigqcu@apollo13 \
    --to=lists@peufeu.com \
    --cc=linux-raid@vger.kernel.org \
    --cc=mlaks@verizon.net \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).