Re: [suse-amd64] SuSE 9.1 on MSI K8T Master-FAR SATA disk i/o problem
On Thu, Jun 17, 2004 at 07:41:08AM +1200, Joel Wiramu Pauling wrote:
1) Your power supply is far too weak... especially with 2 proc in you need at LEAST 26Amps on the 12+ volt power rail. (I doubt you 360 watt is anywhere near this). Use a Toppower 450Watt or Enermax PSU. 360W is definitely too thin.
The 12 volt rail shows 17A... I was thinking that with no GPU that would be enough. But you've given me something to think about now.
2) The OCZ ram you have will not work with this board. OCZ is designed for Single proc performance level enthusiast boards, and has very tweaked and high CAS timmings. This board is a server board, and does not work well (at all...) with the OCZ pc3200 ecc reged ram. I made the same mistake when first purchacing and got... many...many problems. I have a number of threads open with OCZ/MSI/2cpu forums on this issue... Scrap the OCZ ram. Get Samsung pc400 ecc ram. I did and I get MUCH MUCH more stabler operation (samsung adheres to jdec standards and therefore is cas 3 at pc3200 not cas 2 like corsair/mushkin/ocz)
Hmm, I know that could be a problem, but I don't think it's what I have to deal with right at the moment. After reading through others' problems, it really seems like I have to do some more before my hard drives to work correctly, and that's the bulk of the problem at the moment, although other instabilities might be tied to the RAM. Note that I had FreeBSD installed on one of these machines for a while, and didn't have any trouble with it.... I just couldn't get MySQL to run on more than one processor.
3) The Raid on this MOBO is software only.... Disable all arrays through the vt8xxx bios util (hit tab at startup). Then use Linux software raid through yast to setup the array. I get Better performance using linux software raid than I do from the Via windows driver designed by the manufacturer... (I use 2 hitachi 160gbs drives, with raptors should should get even better).
I set up the OS on Linux software RAID. I didn't mess with the VIA BIOS after I realized that it was a software raid and required drivers. My problem now is getting the SATA recognized as SATA, I think. My drives are still showing up as hda and hdc rather than sd# like they should, right? Thanks for your help... I've read more than a couple of your messages :) I hope this can be worked out soon. BW
Nope not gong to happen. The drives SHOULD show up as hda/hdc(b)(Note that the second drive appears as hdc not b in my rig, probably somthing to do with the crazzy via bios...). They are just ide drives... not scsi. SATA is just a jacked up interface to ide devices. Some raid sata solutions implement scsi ontop for various reasons, this one doesn't.. Now go run mprimes on your setup... I bet you 100$ that it crashes out within 20mins... you are going to have to get rid of that ram eventually TRUST me. The system will lock up for no reason. You will get memory leaks and errors left right and center and things just won't run right.... Note to get my rig working I setup two 500MB Partitions on each drive in for /boot and swap, which I left out from the raid lvm array. Right now go run mprimes =-) Kind regards Joel On Thu, 2004-06-17 at 09:12, Brock Witherspoon wrote:
On Thu, Jun 17, 2004 at 07:41:08AM +1200, Joel Wiramu Pauling wrote:
1) Your power supply is far too weak... especially with 2 proc in you need at LEAST 26Amps on the 12+ volt power rail. (I doubt you 360 watt is anywhere near this). Use a Toppower 450Watt or Enermax PSU. 360W is definitely too thin.
The 12 volt rail shows 17A... I was thinking that with no GPU that would be enough. But you've given me something to think about now.
2) The OCZ ram you have will not work with this board. OCZ is designed for Single proc performance level enthusiast boards, and has very tweaked and high CAS timmings. This board is a server board, and does not work well (at all...) with the OCZ pc3200 ecc reged ram. I made the same mistake when first purchacing and got... many...many problems. I have a number of threads open with OCZ/MSI/2cpu forums on this issue... Scrap the OCZ ram. Get Samsung pc400 ecc ram. I did and I get MUCH MUCH more stabler operation (samsung adheres to jdec standards and therefore is cas 3 at pc3200 not cas 2 like corsair/mushkin/ocz)
Hmm, I know that could be a problem, but I don't think it's what I have to deal with right at the moment. After reading through others' problems, it really seems like I have to do some more before my hard drives to work correctly, and that's the bulk of the problem at the moment, although other instabilities might be tied to the RAM.
Note that I had FreeBSD installed on one of these machines for a while, and didn't have any trouble with it.... I just couldn't get MySQL to run on more than one processor.
3) The Raid on this MOBO is software only.... Disable all arrays through the vt8xxx bios util (hit tab at startup). Then use Linux software raid through yast to setup the array. I get Better performance using linux software raid than I do from the Via windows driver designed by the manufacturer... (I use 2 hitachi 160gbs drives, with raptors should should get even better).
I set up the OS on Linux software RAID. I didn't mess with the VIA BIOS after I realized that it was a software raid and required drivers. My problem now is getting the SATA recognized as SATA, I think. My drives are still showing up as hda and hdc rather than sd# like they should, right?
Thanks for your help... I've read more than a couple of your messages :) I hope this can be worked out soon.
BW
Joel Wiramu Pauling wrote:
Nope not gong to happen. The drives SHOULD show up as hda/hdc(b)(Note that the second drive appears as hdc not b in my rig, probably somthing to do with the crazzy via bios...). They are just ide drives... not scsi. SATA is just a jacked up interface to ide devices. Some raid sata solutions implement scsi ontop for various reasons, this one doesn't..
All my SATA drives show up as scsi. I run a RAID0 nicely across a pair of raptors. As an FYI: hammer:~ # lsmod | grep -i ata sata_promise 11204 6 libata 43400 1 sata_promise scsi_mod 137600 8 st,sr_mod,sg,ide_scsi,sata_promise,libata,aic7xxx,sd_mod hammer:~ # dmesg | grep "[hs]d[ab]" ide0: BM-DMA at 0xffa0-0xffa7, BIOS settings: hda:DMA, hdb:pio hda: ST340014A, ATA DISK drive hda: max request size: 1024KiB hda: 78165360 sectors (40020 MB) w/2048KiB Cache, CHS=16383/255/63, UDMA(100) hda: hda1 hda2 hda3 hda4 < hda5 hda6 hda7 > SCSI device sda: 72303840 512-byte hdwr sectors (37020 MB) SCSI device sda: drive cache: write back sda: sda1 sda2 sda3 Attached scsi disk sda at scsi1, channel 0, id 0, lun 0 SCSI device sdb: 72303840 512-byte hdwr sectors (37020 MB) SCSI device sdb: drive cache: write back sdb: sdb1 sdb2 sdb3 Attached scsi disk sdb at scsi2, channel 0, id 0, lun 0 and from /var/log/boot.msg <7>libata version 1.02 loaded. <7>sata_promise version 1.00 <6>ata1: SATA max UDMA/133 cmd 0xFFFFFF0000018200 ctl 0xFFFFFF0000018238 bmdma 0 x0 irq 17 <6>ata2: SATA max UDMA/133 cmd 0xFFFFFF0000018280 ctl 0xFFFFFF00000182B8 bmdma 0 x0 irq 17 <7>ata1: dev 0 cfg 49:2f00 82:346b 83:7f21 84:4003 85:3469 86:3c01 87:4003 88:20 7f <6>ata1: dev 0 ATA, max UDMA/133, 72303840 sectors: lba48 <6>ata1: dev 0 configured for UDMA/133 <6>scsi1 : sata_promise <7>ata2: dev 0 cfg 49:2f00 82:346b 83:7f21 84:4003 85:3469 86:3c01 87:4003 88:20 7f <6>ata2: dev 0 ATA, max UDMA/133, 72303840 sectors: lba48 <6>ata2: dev 0 configured for UDMA/133 <6>scsi2 : sata_promise <5> Vendor: ATA Model: WDC WD360GD-00FN Rev: 35.0 <5> Type: Direct-Access ANSI SCSI revision: 05 <5>SCSI device sda: 72303840 512-byte hdwr sectors (37020 MB) <5>SCSI device sda: drive cache: write back <6> sda: sda1 sda2 sda3 <5>Attached scsi disk sda at scsi1, channel 0, id 0, lun 0 <5> Vendor: ATA Model: WDC WD360GD-00FN Rev: 35.0 <5> Type: Direct-Access ANSI SCSI revision: 05 <5>SCSI device sdb: 72303840 512-byte hdwr sectors (37020 MB) <5>SCSI device sdb: drive cache: write back -- Joseph Landman, Ph.D Scalable Informatics LLC, email: landman@scalableinformatics.com web : http://scalableinformatics.com phone: +1 734 612 4615
Your using Promise not vt8xxx. Which show up the sata's as normal ide devices i.e /dev/hdx). Like I mentioned some sata controllers do show the devices as scsi, (for whatever reasons). The Promise is true hardware raid so this is probably why. Whereas the vt8xxx is just an sata controller and thus is more like your standard atapi contoller chip. The SATA drives regardless are pretty much more like atapi ide drives than true scsi array discs. On Thu, 2004-06-17 at 11:47, Joe Landman wrote:
Joel Wiramu Pauling wrote:
Nope not gong to happen. The drives SHOULD show up as hda/hdc(b)(Note that the second drive appears as hdc not b in my rig, probably somthing to do with the crazzy via bios...). They are just ide drives... not scsi. SATA is just a jacked up interface to ide devices. Some raid sata solutions implement scsi ontop for various reasons, this one doesn't..
All my SATA drives show up as scsi. I run a RAID0 nicely across a pair of raptors. As an FYI:
hammer:~ # lsmod | grep -i ata sata_promise 11204 6 libata 43400 1 sata_promise scsi_mod 137600 8 st,sr_mod,sg,ide_scsi,sata_promise,libata,aic7xxx,sd_mod
hammer:~ # dmesg | grep "[hs]d[ab]" ide0: BM-DMA at 0xffa0-0xffa7, BIOS settings: hda:DMA, hdb:pio hda: ST340014A, ATA DISK drive hda: max request size: 1024KiB hda: 78165360 sectors (40020 MB) w/2048KiB Cache, CHS=16383/255/63, UDMA(100) hda: hda1 hda2 hda3 hda4 < hda5 hda6 hda7 > SCSI device sda: 72303840 512-byte hdwr sectors (37020 MB) SCSI device sda: drive cache: write back sda: sda1 sda2 sda3 Attached scsi disk sda at scsi1, channel 0, id 0, lun 0 SCSI device sdb: 72303840 512-byte hdwr sectors (37020 MB) SCSI device sdb: drive cache: write back sdb: sdb1 sdb2 sdb3 Attached scsi disk sdb at scsi2, channel 0, id 0, lun 0
and from /var/log/boot.msg
<7>libata version 1.02 loaded. <7>sata_promise version 1.00 <6>ata1: SATA max UDMA/133 cmd 0xFFFFFF0000018200 ctl 0xFFFFFF0000018238 bmdma 0 x0 irq 17 <6>ata2: SATA max UDMA/133 cmd 0xFFFFFF0000018280 ctl 0xFFFFFF00000182B8 bmdma 0 x0 irq 17 <7>ata1: dev 0 cfg 49:2f00 82:346b 83:7f21 84:4003 85:3469 86:3c01 87:4003 88:20 7f <6>ata1: dev 0 ATA, max UDMA/133, 72303840 sectors: lba48 <6>ata1: dev 0 configured for UDMA/133 <6>scsi1 : sata_promise <7>ata2: dev 0 cfg 49:2f00 82:346b 83:7f21 84:4003 85:3469 86:3c01 87:4003 88:20 7f <6>ata2: dev 0 ATA, max UDMA/133, 72303840 sectors: lba48 <6>ata2: dev 0 configured for UDMA/133 <6>scsi2 : sata_promise <5> Vendor: ATA Model: WDC WD360GD-00FN Rev: 35.0 <5> Type: Direct-Access ANSI SCSI revision: 05 <5>SCSI device sda: 72303840 512-byte hdwr sectors (37020 MB) <5>SCSI device sda: drive cache: write back <6> sda: sda1 sda2 sda3 <5>Attached scsi disk sda at scsi1, channel 0, id 0, lun 0 <5> Vendor: ATA Model: WDC WD360GD-00FN Rev: 35.0 <5> Type: Direct-Access ANSI SCSI revision: 05 <5>SCSI device sdb: 72303840 512-byte hdwr sectors (37020 MB) <5>SCSI device sdb: drive cache: write back
-- Joseph Landman, Ph.D Scalable Informatics LLC, email: landman@scalableinformatics.com web : http://scalableinformatics.com phone: +1 734 612 4615
A snippet from http://www.linuxmafia.com/faq/Hardware/sata.html "Support is mostly in 2.6 kernels, and in circa-2.4.23 and later kernels. Note that device names will change, if you upgrade kernels from 2.4.x to 2.6.0, e.g., move from /dev/hdg to /dev/sda. So, e.g., append "root=/dev/sda3" to kernel's boot options. (According to a 2004-04-15 post to the Linux kernel mailing list, this addressing change occurs only with the ICH5 and SiI3112 chipset.)" Joe Landman wrote:
Joel Wiramu Pauling wrote:
Nope not gong to happen. The drives SHOULD show up as hda/hdc(b)(Note that the second drive appears as hdc not b in my rig, probably somthing to do with the crazzy via bios...). They are just ide drives... not scsi. SATA is just a jacked up interface to ide devices. Some raid sata solutions implement scsi ontop for various reasons, this one doesn't..
All my SATA drives show up as scsi. I run a RAID0 nicely across a pair of raptors. As an FYI:
hammer:~ # lsmod | grep -i ata sata_promise 11204 6 libata 43400 1 sata_promise scsi_mod 137600 8 st,sr_mod,sg,ide_scsi,sata_promise,libata,aic7xxx,sd_mod
hammer:~ # dmesg | grep "[hs]d[ab]" ide0: BM-DMA at 0xffa0-0xffa7, BIOS settings: hda:DMA, hdb:pio hda: ST340014A, ATA DISK drive hda: max request size: 1024KiB hda: 78165360 sectors (40020 MB) w/2048KiB Cache, CHS=16383/255/63, UDMA(100) hda: hda1 hda2 hda3 hda4 < hda5 hda6 hda7 > SCSI device sda: 72303840 512-byte hdwr sectors (37020 MB) SCSI device sda: drive cache: write back sda: sda1 sda2 sda3 Attached scsi disk sda at scsi1, channel 0, id 0, lun 0 SCSI device sdb: 72303840 512-byte hdwr sectors (37020 MB) SCSI device sdb: drive cache: write back sdb: sdb1 sdb2 sdb3 Attached scsi disk sdb at scsi2, channel 0, id 0, lun 0
and from /var/log/boot.msg
<7>libata version 1.02 loaded. <7>sata_promise version 1.00 <6>ata1: SATA max UDMA/133 cmd 0xFFFFFF0000018200 ctl 0xFFFFFF0000018238 bmdma 0 x0 irq 17 <6>ata2: SATA max UDMA/133 cmd 0xFFFFFF0000018280 ctl 0xFFFFFF00000182B8 bmdma 0 x0 irq 17 <7>ata1: dev 0 cfg 49:2f00 82:346b 83:7f21 84:4003 85:3469 86:3c01 87:4003 88:20 7f <6>ata1: dev 0 ATA, max UDMA/133, 72303840 sectors: lba48 <6>ata1: dev 0 configured for UDMA/133 <6>scsi1 : sata_promise <7>ata2: dev 0 cfg 49:2f00 82:346b 83:7f21 84:4003 85:3469 86:3c01 87:4003 88:20 7f <6>ata2: dev 0 ATA, max UDMA/133, 72303840 sectors: lba48 <6>ata2: dev 0 configured for UDMA/133 <6>scsi2 : sata_promise <5> Vendor: ATA Model: WDC WD360GD-00FN Rev: 35.0 <5> Type: Direct-Access ANSI SCSI revision: 05 <5>SCSI device sda: 72303840 512-byte hdwr sectors (37020 MB) <5>SCSI device sda: drive cache: write back <6> sda: sda1 sda2 sda3 <5>Attached scsi disk sda at scsi1, channel 0, id 0, lun 0 <5> Vendor: ATA Model: WDC WD360GD-00FN Rev: 35.0 <5> Type: Direct-Access ANSI SCSI revision: 05 <5>SCSI device sdb: 72303840 512-byte hdwr sectors (37020 MB) <5>SCSI device sdb: drive cache: write back
-- Tony Kubinski tkubinski@dls.net 630.292.5293 (cell) 847.947.3224 (fax)
participants (4)
-
Brock Witherspoon
-
Joe Landman
-
Joel Wiramu Pauling
-
tonykubinski