Re: bootable raid 1
Hi David, Backup your data first ifyou are using a production box. Read completely this HOWTO as it outlines the process of going from non-raid to bootable raid: http://www.linuxdoc.org/HOWTO/Boot+Root+Raid+LILO.html Please note that the above HOWTO references an older lilo. I recommend using 21.5 or later for the bootable raid 1. Here is my lilo.conf for your reference. boot=/dev/md1 map=/boot/map install=/boot/boot.b prompt timeout=50 default=vmlinux vga=normal keytable=/boot/us.klt message=/boot/message menu-scheme=wb:bw:wb:bw image=/boot/vmlinux label=vmlinux root=/dev/md0 append="ide6=0x168,0x36e,10 ide0=autotune ide1=autotune ide6=autotune" read-only image=/boot/vmlinuz label=vmlinuz root=/dev/md0 read-only image=/boot/bzImage label=z root=/dev/md0 read-only Also reference this: http://www.linuxdoc.org/FAQ/Linux-RAID-FAQ Then go here as this will incorporate the latest patches needed for raid: http://www.linuxraid.org/ If you are using ide then get the latest patches from here: http://www.kernel.org/pub/linux/kernel/people/hedrick/ Another resource for well packaged patches to the 2.2.18 kernel is: http://www.hattaway-associates.com/raidpatches/ and finally, consult this mailing list: [EMAIL PROTECTED] archived at: http://marc.theaimsgroup.com/?l=linux-raid&r=1&w=2 Good luck! Charles p.s. If anyone has anything helpful or more up-to-date to add to this list of resources, please feel free. - Original Message - From: "David Moylan" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Tuesday, December 26, 2000 5:11 PM Subject: bootable raid 1 > greetings, > > i'm keen to get a bootable raid-1 config up and going. are you able to > point me towards a "howto" or a web page with information? alternatively > are you able to give me a run down on what you did to get it working? > > so you know, i'm an advanced linux user. > > cheers, Wiz!! > > --- > > I am pleased that I have > overcome this issue and here are my conclusions. > > My previous raid configuration was set up using CHS method for > harddisk geometry. There was no reason for doing this other than it > was a part of the diagnostic procedure for a previous issues regarding > a western digital harddrive that identifies itself to the tyan bios as > being ata66 but really only operates reliably at ata33. The CHS really > had nothing to do with the timing issue, but I suspect that the > partition table arrangement did. The symptom of this timing issue was > not being able to reliably modify the harddrive's partition tables, > strangely enough. Mandrakeuser.org has the write up about certain > western digital harddrives that have this issue. > - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [EMAIL PROTECTED]
Re: bootable raid 1
On Wed, 20 Dec 2000, Michael wrote: > This is an issue with the firmware on the drive itself. Intrestingly > enough, I have 2 identical (??) 20 gig drives purchased at the same > time that report differently to the kernel at boot. When probed, one > reports CHS and the other reports LBA values. I haven't checked in Well, I've noticed a similar situation too. If I have two identical disks (even the same firmware revision), one connected to primary, the other to secondary IDE controller, the first will report LBA and the later 'normal' CHS. If I connect both to primary as master/slave pair, they both report LBA values. This seems to be 'normal', as I've seen it on virtually all motherboard/disk combinations. > Bottom line, it is important to note the probe values reported by the > kernel and to use those values to set the bios on the motherboard > since fdisk uses the bios values when creating the partitions and it > seems like the kernel uses the probe values read from the ide > interface. Maybe I haven't got it quite right, but this is as close > as I've managed without digging into the code. You can always use the 'hdx=cyl,head,sectors' to make Linux (and fdisk) use the geometry you prefer. I usually adjust the secondary drive params to those detected on the primary. D. - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [EMAIL PROTECTED]
Re: bootable raid 1
Same here. I believe it has to do with the seconday controller typically not being the boot device. The bios on the mainboard supports LBA on the secondary controller, but will not report in by default. However, where linux is concerned, and like Danilo mentioned, you can used fdisk in expert mode and manaully key in the LBA values for consistency with the primary master drive and it will work just fine. In fact, I recommend it. Just be sure that you are entering in the correct values. - Original Message - From: "Danilo Godec" <[EMAIL PROTECTED]> To: "Michael" <[EMAIL PROTECTED]> Cc: <[EMAIL PROTECTED]> Sent: Thursday, December 21, 2000 3:05 AM Subject: Re: bootable raid 1 > On Wed, 20 Dec 2000, Michael wrote: > > > This is an issue with the firmware on the drive itself. Intrestingly > > enough, I have 2 identical (??) 20 gig drives purchased at the same > > time that report differently to the kernel at boot. When probed, one > > reports CHS and the other reports LBA values. I haven't checked in > > Well, I've noticed a similar situation too. If I have two identical disks > (even the same firmware revision), one connected to primary, the other to > secondary IDE controller, the first will report LBA and the later 'normal' > CHS. If I connect both to primary as master/slave pair, they both report > LBA values. > > This seems to be 'normal', as I've seen it on virtually all > motherboard/disk combinations. > > > Bottom line, it is important to note the probe values reported by the > > kernel and to use those values to set the bios on the motherboard > > since fdisk uses the bios values when creating the partitions and it > > seems like the kernel uses the probe values read from the ide > > interface. Maybe I haven't got it quite right, but this is as close > > as I've managed without digging into the code. > > You can always use the 'hdx=cyl,head,sectors' to make Linux (and > fdisk) use the geometry you prefer. I usually adjust the secondary drive > params to those detected on the primary. > > > D. > > > - > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > the body of a message to [EMAIL PROTECTED] - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [EMAIL PROTECTED]
Re: bootable raid 1
> I am not 100% certain, but this is the only change > that I implemented that made the difference. It might explain why > when both drives were installed that the system booted, but when > either was removed, it would not boot. > > Using LBA translation method in both the BIOS and with fdisk using > matching LBA geometries, I can now boot from either drive- the way > raid1 is meant to be . . . > This is an issue with the firmware on the drive itself. Intrestingly enough, I have 2 identical (??) 20 gig drives purchased at the same time that report differently to the kernel at boot. When probed, one reports CHS and the other reports LBA values. I haven't checked in detail, but I suspect that the firmware revision levels on the disks are different. The disks will operate with either setting in the motherboard bios, and will report those settings to fdisk. However, the kernel wants to use the probe values or will not consistently write the correct stuff to the disk. So one is set up as CHS and the other as LBA -- both are in the same raid 1 set and all works fine with those settings. It does not work fine when otherwise. Bottom line, it is important to note the probe values reported by the kernel and to use those values to set the bios on the motherboard since fdisk uses the bios values when creating the partitions and it seems like the kernel uses the probe values read from the ide interface. Maybe I haven't got it quite right, but this is as close as I've managed without digging into the code. - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [EMAIL PROTECTED]
Re: bootable raid 1
Whoops shouls be lilo 21.3 Godfrey Livingstone wrote: > Try the following with lilo 21.2 it works for me Neither 21.6 nor 21.5 do. > > boot=/dev/md1 > map=/boot/map > install=/boot/boot.b > prompt > timeout=50 > default=linux > > image=/boot/vmlinux > label=linux > read-only > root=/dev/md0 > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > > Charles Wilkins wrote: > > > Let me try this again. > > > > I have now tried lilo 21.5 and 21.6 > > I have tried this wth 1 lilo.conf and with 2. > > > > My /boot filesystem exists on device /dev/md1. > > My root filesystem exists on device /dev/md0. > > > > I do have a motherboard that will attempt to boot the secondary master if > > the primary is absent. > > > > When I try to use 1 lilo.conf with /dev/md1 as the boot and /dev/md0 as the > > root, when the system tries to boot, it comes up with hieroglyphics and > > needless to say, does not boot. > > > > When I try to use 2 lilo.confs (listed below), the system boots, but only > > with both drives physically connected. This somewhat defeats the purpose of > > raid1 as inevitably a drive will eventually fail. > > > > Is anyone successfully booting raid1 with ide that has tested booting the > > system with either drive removed? If so, would you mind posing your > > lilo.conf(s) along with the version of lilo that you are using. > > > > lilo.conf.hda > > > > disk=/dev/md0 > > bios=0x80 > > sectors=63 > > heads=15 > > cylinders=13328 > > partition=/dev/md1 > > start=1 > > boot=/dev/hda > > map=/boot/map > > install=/boot/boot.b > > vga=normal > > default=vmlinux > > keytable=/boot/us.klt > > prompt > > timeout=50 > > message=/boot/message > > menu-scheme=wb:bw:wb:bw > > image=/boot/vmlinuz > > label=vmlinuz > > root=/dev/md0 > > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > > read-only > > image=/boot/vmlinux > > label=vmlinux > > root=/dev/md0 > > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > > read-only > > > > lilo.conf.hdc > > > > disk=/dev/md0 > > bios=0x81 > > sectors=63 > > heads=16 > > cylinders=16278 > > partition=/dev/md1 > > start=1 > > boot=/dev/hdc > > map=/boot/map > > install=/boot/boot.b > > vga=normal > > default=vmlinux > > keytable=/boot/us.klt > > prompt > > timeout=50 > > message=/boot/message > > menu-scheme=wb:bw:wb:bw > > image=/boot/vmlinuz > > label=vmlinuz > > root=/dev/md0 > > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > > read-only > > image=/boot/vmlinux > > label=vmlinux > > root=/dev/md0 > > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > > read-only > > > > - Original Message - > > From: "Luca Berra" <[EMAIL PROTECTED]> > > To: <[EMAIL PROTECTED]> > > Sent: Monday, December 18, 2000 12:08 PM > > Subject: Re: bootable raid 1 > > > > > On Mon, Dec 18, 2000 at 09:15:51AM -0500, Charles Wilkins wrote: > > > > I am currently doing bootable raid 1 with all of the latest patches for > > > > 2.2.18 and then some. > > > > When I try to boot the system with either drive removed, I get a bunch > > of > > > > 010101010's in an endless loop. > > > > Yet, when both drives are in, the system comes up fine. What am I doing > > > > wrong? > > > > > > > > Using lilo 21.5 > > > > > > With Lilo 21.5 you don't need any geometry junk, nor to use > > > more than one lilo.conf > > > just use md0 as if you had only a normal disk. > > > remember to include a 'default' entry. > > > > > > L. > > > > > > > lilo.conf.hda > > > > > > > > disk=/dev/md0 > > > > bios=0x80 > > > > sectors=63 > > > > heads=15 > > > > cylinders=13328 > > > -- > > > Luca Berra -- [EMAIL PROTECTED] > > > Communication Media & Services S.r.l. > > > /"\ > > > \ / ASCII RIBBON CAMPAIGN > > > XAGAINST HTML MAIL > > > / \ > > > - > > > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > > > the body of a message to [EMAIL PROTECTED] > > > > - > > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > > the body of a message to [EMAIL PROTECTED] > > - > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > the body of a message to [EMAIL PROTECTED] - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [EMAIL PROTECTED]
Re: bootable raid 1
Try the following with lilo 21.2 it works for me Neither 21.6 nor 21.5 do. boot=/dev/md1 map=/boot/map install=/boot/boot.b prompt timeout=50 default=linux image=/boot/vmlinux label=linux read-only root=/dev/md0 append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" Charles Wilkins wrote: > Let me try this again. > > I have now tried lilo 21.5 and 21.6 > I have tried this wth 1 lilo.conf and with 2. > > My /boot filesystem exists on device /dev/md1. > My root filesystem exists on device /dev/md0. > > I do have a motherboard that will attempt to boot the secondary master if > the primary is absent. > > When I try to use 1 lilo.conf with /dev/md1 as the boot and /dev/md0 as the > root, when the system tries to boot, it comes up with hieroglyphics and > needless to say, does not boot. > > When I try to use 2 lilo.confs (listed below), the system boots, but only > with both drives physically connected. This somewhat defeats the purpose of > raid1 as inevitably a drive will eventually fail. > > Is anyone successfully booting raid1 with ide that has tested booting the > system with either drive removed? If so, would you mind posing your > lilo.conf(s) along with the version of lilo that you are using. > > lilo.conf.hda > > disk=/dev/md0 > bios=0x80 > sectors=63 > heads=15 > cylinders=13328 > partition=/dev/md1 > start=1 > boot=/dev/hda > map=/boot/map > install=/boot/boot.b > vga=normal > default=vmlinux > keytable=/boot/us.klt > prompt > timeout=50 > message=/boot/message > menu-scheme=wb:bw:wb:bw > image=/boot/vmlinuz > label=vmlinuz > root=/dev/md0 > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > read-only > image=/boot/vmlinux > label=vmlinux > root=/dev/md0 > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > read-only > > lilo.conf.hdc > > disk=/dev/md0 > bios=0x81 > sectors=63 > heads=16 > cylinders=16278 > partition=/dev/md1 > start=1 > boot=/dev/hdc > map=/boot/map > install=/boot/boot.b > vga=normal > default=vmlinux > keytable=/boot/us.klt > prompt > timeout=50 > message=/boot/message > menu-scheme=wb:bw:wb:bw > image=/boot/vmlinuz > label=vmlinuz > root=/dev/md0 > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > read-only > image=/boot/vmlinux > label=vmlinux > root=/dev/md0 > append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" > read-only > > - Original Message - > From: "Luca Berra" <[EMAIL PROTECTED]> > To: <[EMAIL PROTECTED]> > Sent: Monday, December 18, 2000 12:08 PM > Subject: Re: bootable raid 1 > > > On Mon, Dec 18, 2000 at 09:15:51AM -0500, Charles Wilkins wrote: > > > I am currently doing bootable raid 1 with all of the latest patches for > > > 2.2.18 and then some. > > > When I try to boot the system with either drive removed, I get a bunch > of > > > 010101010's in an endless loop. > > > Yet, when both drives are in, the system comes up fine. What am I doing > > > wrong? > > > > > > Using lilo 21.5 > > > > With Lilo 21.5 you don't need any geometry junk, nor to use > > more than one lilo.conf > > just use md0 as if you had only a normal disk. > > remember to include a 'default' entry. > > > > L. > > > > > lilo.conf.hda > > > > > > disk=/dev/md0 > > > bios=0x80 > > > sectors=63 > > > heads=15 > > > cylinders=13328 > > -- > > Luca Berra -- [EMAIL PROTECTED] > > Communication Media & Services S.r.l. > > /"\ > > \ / ASCII RIBBON CAMPAIGN > > XAGAINST HTML MAIL > > / \ > > - > > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > > the body of a message to [EMAIL PROTECTED] > > - > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > the body of a message to [EMAIL PROTECTED] - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [EMAIL PROTECTED]
Re: bootable raid 1
Let me try this again. I have now tried lilo 21.5 and 21.6 I have tried this wth 1 lilo.conf and with 2. My /boot filesystem exists on device /dev/md1. My root filesystem exists on device /dev/md0. I do have a motherboard that will attempt to boot the secondary master if the primary is absent. When I try to use 1 lilo.conf with /dev/md1 as the boot and /dev/md0 as the root, when the system tries to boot, it comes up with hieroglyphics and needless to say, does not boot. When I try to use 2 lilo.confs (listed below), the system boots, but only with both drives physically connected. This somewhat defeats the purpose of raid1 as inevitably a drive will eventually fail. Is anyone successfully booting raid1 with ide that has tested booting the system with either drive removed? If so, would you mind posing your lilo.conf(s) along with the version of lilo that you are using. lilo.conf.hda disk=/dev/md0 bios=0x80 sectors=63 heads=15 cylinders=13328 partition=/dev/md1 start=1 boot=/dev/hda map=/boot/map install=/boot/boot.b vga=normal default=vmlinux keytable=/boot/us.klt prompt timeout=50 message=/boot/message menu-scheme=wb:bw:wb:bw image=/boot/vmlinuz label=vmlinuz root=/dev/md0 append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" read-only image=/boot/vmlinux label=vmlinux root=/dev/md0 append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" read-only lilo.conf.hdc disk=/dev/md0 bios=0x81 sectors=63 heads=16 cylinders=16278 partition=/dev/md1 start=1 boot=/dev/hdc map=/boot/map install=/boot/boot.b vga=normal default=vmlinux keytable=/boot/us.klt prompt timeout=50 message=/boot/message menu-scheme=wb:bw:wb:bw image=/boot/vmlinuz label=vmlinuz root=/dev/md0 append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" read-only image=/boot/vmlinux label=vmlinux root=/dev/md0 append=" ide3=autotune ide1=autotune ide2=autotune hdh=ide-scsi" read-only - Original Message - From: "Luca Berra" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Monday, December 18, 2000 12:08 PM Subject: Re: bootable raid 1 > On Mon, Dec 18, 2000 at 09:15:51AM -0500, Charles Wilkins wrote: > > I am currently doing bootable raid 1 with all of the latest patches for > > 2.2.18 and then some. > > When I try to boot the system with either drive removed, I get a bunch of > > 010101010's in an endless loop. > > Yet, when both drives are in, the system comes up fine. What am I doing > > wrong? > > > > Using lilo 21.5 > > With Lilo 21.5 you don't need any geometry junk, nor to use > more than one lilo.conf > just use md0 as if you had only a normal disk. > remember to include a 'default' entry. > > L. > > > lilo.conf.hda > > > > disk=/dev/md0 > > bios=0x80 > > sectors=63 > > heads=15 > > cylinders=13328 > -- > Luca Berra -- [EMAIL PROTECTED] > Communication Media & Services S.r.l. > /"\ > \ / ASCII RIBBON CAMPAIGN > XAGAINST HTML MAIL > / \ > - > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > the body of a message to [EMAIL PROTECTED] - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [EMAIL PROTECTED]
Re: bootable raid 1
On Mon, Dec 18, 2000 at 09:15:51AM -0500, Charles Wilkins wrote: > I am currently doing bootable raid 1 with all of the latest patches for > 2.2.18 and then some. > When I try to boot the system with either drive removed, I get a bunch of > 010101010's in an endless loop. > Yet, when both drives are in, the system comes up fine. What am I doing > wrong? > > Using lilo 21.5 With Lilo 21.5 you don't need any geometry junk, nor to use more than one lilo.conf just use md0 as if you had only a normal disk. remember to include a 'default' entry. L. > lilo.conf.hda > > disk=/dev/md0 > bios=0x80 > sectors=63 > heads=15 > cylinders=13328 -- Luca Berra -- [EMAIL PROTECTED] Communication Media & Services S.r.l. /"\ \ / ASCII RIBBON CAMPAIGN XAGAINST HTML MAIL / \ - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to [EMAIL PROTECTED]