July 2004 - openSUSE AMD64 - openSUSE Mailing Lists

RE: RE: [suse-amd64] Raid 0 + ReiserFS
by Joel Wiramu Pauling 01 Jul '04

01 Jul '04

I was using the latest suse 9.1 amd release kernel. Have an MSI Master2-Far which uses the via vt8xxx sata controller. I had raid0 software stripe, of two identical hitachi 160gb's each drive had 1gb unstripped ext2/swap partition for boot. I won't rule out the possibility that it was somthing to do with the sata_via kernel module. Also in the past I have noted that alot of people have wierd expereinces with the way which suse addresses the sata/raid of this chip Suse uses /dev/hdx to address the sata drives. Wheras all other distro's use sd layout. I had to not setup an array in the controller bios, and do it entirely through linux for it to work. Because if I setup the raid arrya first in the controller bios (stripe both discs full), then suse would still insist on recognising them as two seperate /dev/hdX... Because I have heard other people with the board and setup have sucsesfully got other distros (i.e mandrake 10) to recognise the bios setup raid array as an sdX device. I beleive that suse is to blame partly. I would like comment on this. Like I said I have posted extensively on the setup of this system in suse, and the way which I was using it was the only configuration that worked... Kind regards Joel -----Original Message----- From: "Mike Tierney" <miket(a)marketview.co.nz> Sent: Thu, 1 Jul 2004 16:13:15 +1200 To: <suse-amd64(a)suse.com> Subject: RE: [suse-amd64] Raid 0 + ReiserFS I've been running SLES 8 under Raid 0 on Reiserfs at home for a few months now with no problems but then all I do with my home Linux installation is played Neverwinter Nights or tunnel into work via vmware, SSH and PCAnywhere..... You could always use LVM to stripe your reiserfs, thats what I use (under SLES 8 AMD-64) for our production database server. And yes of course the striped ResierFS is mirrored! BTW, which exact version of SuSE Linux are you having this problem with?? The ResierFS faq ( http://www.namesys.com/faq.html )does mention these issues with _earlier_ kernels. "Q. Can I use ReiserFS with the software RAID?" "A. Yes, for all Raid levels using any Linux >= 2.4.1, but DO NOT use Raid5 with Linux 2.2.x. Our journaling and their Raid code step on each other in the buffering code. Also, mirroring is not safe in the 2.2.x kernels because online mirror rebuilds in 2.2.x break the write ordering requirements for the log. If you crash in the middle of an online rebuild, your meta-data may be corrupted. The only Raid level that is safe with ReiserFS in the 2.2.x kernels is the striping/concatenation level." Anyway, if you store any important data on Raid 0 and don't have super-tight backups you are asking for a beating IMO. I would take this problem to the ReiserFS guys (www.namesys.com) as they take any reported problems VERY seriously. -----Original Message----- From: Joel Wiramu Pauling [mailto:aenertia@aenertia.net] Sent: Thursday, 1 July 2004 3:29 p.m. To: suse-amd64(a)suse.com Subject: [suse-amd64] Raid 0 + ReiserFS Hi There, I have a big Warning, Don't use raid 0 .and reiserfs. bad Bad BAd BAD!. I had been noticing strange errors for the last two months (certain files reporting 0 length and being unable to be removed). And I ran a single user session and rebuilt the reiser tree. This corrected all the issues, but, created more in the process. Basically I have been left with holes all over my filesystem, in a fairly random order. I believe reiserfs 3.6 can't handle the stripe nature of software raid, and screws up where things are because of this. There is absolutely nothing wrong with the discs physically. And i'm pretty certain it's just the combination of reiserfs and raid0. Anyway. The fact is I've lost a lot of data. Suse people... you need to change the default type of filesystem for raid stripes.... This progressive fall to doom will be happening to a number of people with suse installed and a raid 0 stripe. I've posted to the reiserfs guys aswell. Currently I'm trying to rescue what I can but because the problems are random, even taring and sending it across the network to a temporary machine may only be moving the errors in much of the cases. Kind regards Joel -- Check the List-Unsubscribe header to unsubscribe For additional commands, email: suse-amd64-help(a)suse.com -- Check the List-Unsubscribe header to unsubscribe For additional commands, email: suse-amd64-help(a)suse.com

3 2

BUG: linuxrc fails to load x86_64 compiled module
by Kevin.Wilson＠comtrol.com 01 Jul '04

01 Jul '04

Distro: SuSE 9.1 x86_64-SMP No updates applied. Using 2.6.4-52 kernel sources. Installed to regular IDE disk to do controller module build. Hardware: AMD Opteron System 3ware 9500S-8 SATA RAID Controller Steps to reproduce: Compile a fresh raid controller module using 3ware sources. Successfully run insmod to install it to IDE disk. (RAID array detection successful, ready to partition.) Create update cd using new module. Shutdown and disconnect IDE disk. Power up and run a fresh install from the dvd disk. (target = RAID array) (installation with ACPI disabled) Select F6 to do driver update. Load update cd when prompted. select cdrom as update source in Linuxrc. Linuxrc fails with the following messages after loading update: insmod: error inserting '/modules/3w-9xxx.ko': -1 invalid module format Other messages seen during install: 3w_9xxx: no version for "struct_module" found: kernel tainted 3w_9xxx: version magic '2.6.4-52-smp SMP gcc 3.3' should be '2.6.4-52-default gcc-3.3' 1) It isn't invalid because I used module-init-tools 3.0_pre10-32 x86_64 from the distro disk to successfully load it on build disk. 2) I am installing to a SMP system so I shouldn't see this message at all. 3) Have no clue why the "struct_module" message appears or what it means, couldn't find any information other than it happens to many people after doing a search. Now would the change to UTF-8 file encoding cause Linuxrc's insmod to do this? Could it be that Linuxrc cannot load this module because it uses a older version of module-init-utils? Must I build boot disks from the installed system to use in conjunction with the 9.1 installation DVD to allow proper loading of the module? How can I get a x86_64 driver module loaded during first installation?

2 1

New to Suse
by Ray Carlino 01 Jul '04

01 Jul '04

I am new to Suse used to use mandrake but they totally messed up their amd64 version. Anyway where is a good place to get new applications and kernel updates? I would like to get a newer kernel that has support for my Treo 600 phone using the vision driver. Or where can I get the Suse config file for the kernel and I will compile myself? I just like to keep the kernel config as close to the distro as I can. Thanks -- Ray

3 2

Raid 0 + ReiserFS
by Joel Wiramu Pauling 01 Jul '04

01 Jul '04

Hi There, I have a big Warning, Don't use raid 0 .and reiserfs. bad Bad BAd BAD!. I had been noticing strange errors for the last two months (certain files reporting 0 length and being unable to be removed). And I ran a single user session and rebuilt the reiser tree. This corrected all the issues, but, created more in the process. Basically I have been left with holes all over my filesystem, in a fairly random order. I believe reiserfs 3.6 can't handle the stripe nature of software raid, and screws up where things are because of this. There is absolutely nothing wrong with the discs physically. And i'm pretty certain it's just the combination of reiserfs and raid0. Anyway. The fact is I've lost a lot of data. Suse people... you need to change the default type of filesystem for raid stripes.... This progressive fall to doom will be happening to a number of people with suse installed and a raid 0 stripe. I've posted to the reiserfs guys aswell. Currently I'm trying to rescue what I can but because the problems are random, even taring and sending it across the network to a temporary machine may only be moving the errors in much of the cases. Kind regards Joel

2 1

Re: PXE/TFTP'ing 9.1 on AMD64
by Alan Scheinine 01 Jul '04

01 Jul '04

Charles Smith had a problem with PXE/TFTP booting/autoinstalling. For Red Hat 32-bit and 64-bit both work, and aside from the last step of "autoinstalling", the procedure for SUSE should be the same. For Intel x86 and AMD x86_64, /tftpboot/pxelinux.0 works fine. In pxelinux.cfg the files for 32-bit machines has the lines default local #default linux_install prompt 1 timeout 100 display msgs/boot.msg F1 msgs/boot.msg F7 msgs/snake.msg label linux_install kernel vmlinuz append ksdevice=eth0 initrd=initrd.img network ks=http://156.148.64.101/kickst art/scivu6/ks.cfg lang= devfs=nomount ramdisk_size=8192 label local localboot 0 where vmlinuz and initrd.img have been taken from the appropriate 32-bit distribution whereas for 64-bit the names vmlinuz and initrd.img are slightly different to refer to different files, the ones taken from the 64-bit distribution. It works fine for both architectures. As Charles Smith wrote, perhaps the installation procedure for 64-bit is different from 32-bit for SUSE. From the description of the problem it would seem that the last step starting YAST is the problem, that the machine reboots rather than install. Alan

2 1