Re: root/boot on softraid in 5.0
Just a question: running on -current and, similar to the example, having the following disklabel: /boot on both sd0a and sd1a / on sd2a which is the raid result of sd0d and sd1d would the snapshot upgrade process be consistent through the standard installer or should one always go to shell, make manual adjustments and then run install? Thanks Pasha On Sat, Dec 24, 2011 at 2:06 PM, Josh Grosse j...@jggimi.homeip.net wrote: On Sat, Dec 24, 2011 at 02:26:26PM +0200, Ivo Chutkin wrote: Sorry for hijacking this tread. Can someone take time to clarify install process on softraid. I read the article on undeadly.org, preform the install on sd2, my softraid0, copy kernels on sd0a and sd1a, everything went well, and after reboot, it does not find bootable media. I put install cd, drop to shell, mount sd0a, then sd1a and then sd2a, everything is there. Do I miss some obvious step here... Can you point out what is wrong? The kernels are loaded from outside the softraid array (sd0a, or sd1a) by the second stage bootloader. But the second stage bootloader is loaded from the array. So, assuming sd2 is your softraid array, and you have an sd2a root partition, reboot the install cd and try: # mount /dev/sd2a /mnt # cp -p /usr/mdec/boot /mnt/boot # /usr/mdec/installboot -v /mnt/boot /usr/mdec/biosboot sd2 It is AMD64 20 December snapshot on HP microserver. I cannot take dmesg at this point. I can take one from standard install if you need it. Both amd64 and i386 use the same installboot(8) program.
Re: root/boot on softraid in 5.0
On Tue, Dec 27, 2011 at 12:32:59AM +0100, Paolo Aglialoro wrote: running on -current and, similar to the example, having the following disklabel: /boot on both sd0a and sd1a / on sd2a which is the raid result of sd0d and sd1d For i386 and amd64, architectures which have a two-stage boot, the second stage bootloader (/usr/mdec/boot) is installed into the RAID array. This program cannot read kernels from the array, so the kernels must be stored on one or more non-RAID FFS filesystems. In my example, I stored it on the a partition of two drives, wd0 and wd1. would the snapshot upgrade process be consistent through the standard installer or should one always go to shell, make manual adjustments and then run install? The installer installs kernels into the root partition. You must copy them to non-RAID partition(s) in order for the second stage bootloader to be able to load them. Using your example of a binary upgrade such as a snapshot, copy the ramdisk kernel to a non-RAID partition, boot it, conduct the upgrage, then copy the kernels to non-RAID storage, so that the second stage bootloader can find them.
Re: root/boot on softraid in 5.0
So this means that on sparc64 life would be way easier ;) Thanks! On Tue, Dec 27, 2011 at 1:30 AM, Josh Grosse j...@jggimi.homeip.net wrote: On Tue, Dec 27, 2011 at 12:32:59AM +0100, Paolo Aglialoro wrote: running on -current and, similar to the example, having the following disklabel: /boot on both sd0a and sd1a / on sd2a which is the raid result of sd0d and sd1d For i386 and amd64, architectures which have a two-stage boot, the second stage bootloader (/usr/mdec/boot) is installed into the RAID array. This program cannot read kernels from the array, so the kernels must be stored on one or more non-RAID FFS filesystems. In my example, I stored it on the a partition of two drives, wd0 and wd1. would the snapshot upgrade process be consistent through the standard installer or should one always go to shell, make manual adjustments and then run install? The installer installs kernels into the root partition. You must copy them to non-RAID partition(s) in order for the second stage bootloader to be able to load them. Using your example of a binary upgrade such as a snapshot, copy the ramdisk kernel to a non-RAID partition, boot it, conduct the upgrage, then copy the kernels to non-RAID storage, so that the second stage bootloader can find them.
Re: root/boot on softraid in 5.0
On Tue, Dec 27, 2011 at 01:49:59AM +0100, Paolo Aglialoro wrote: So this means that on sparc64 life would be way easier ;) Thanks! I never tested root-on-softraid on single-stage booting architectures. I don't have any handy. You'll have to test that yourself!
Re: root/boot on softraid in 5.0
On 23.12.2011 P3. 08:01 Q., Raymond Lillard wrote: On 12/22/2011 09:07 PM, Paul B. Henson wrote: On Wed, Dec 21, 2011 at 08:08:08PM -0800, Josh Grosse wrote: Woops. I misread your post. The commits were September 19, which is -current, beyond 5.0-release. You must either migrate to -current, or await 5.1-release. Ah, ok, thanks for the clarification. The installboot piece that lets you install bootblocks on softraid is in 5.0, so when that part worked I assumed it all was. The dates (commits in Sept, your post in Oct, 5.0 release in Nov) also led me to misbelieve it was in 5.0. But looking at the changelogs I see the bits that store boot info in softraid metadata and dynamically figure out the root happened after the 5.0 freeze. Something to look forward to in 5.1 :). Thanks again... Why wait? I have two amd64 servers in production on -current and all is humming along magnificently. One is heavily loaded by large image file manipulation over samba. Of course that has little to do with the boot block business. Both servers were booted more than a dozen times during provisioning so I feel good that they will come up after an extended power outage. I did boot with sd0 removed to force a boot from sd1. I then took a third disk and rebuilt sd0 while the machine was up and running. All went well. To me it was worth the trouble to pick a snapshot between commit storms by the developers, just so I didn't have to deal with the altroot thing any more. I have been using OBSD for Internet facing infrastructure for the last 10 years but always used RedHat/CentOS for internal servers. In Oct I installed the first of the two and don't see going back. I am so done with Linux servers. OBSD is getting close to the point where I can use it as a workstation/desktop. My biggest hindrance is no wine. I guess I am going to be forced to make friends with some VM system and cross mount the filesystems via samba. I assume smarter people than me have done that already. I would take suggestions of a preferred VM off-list so as not to hijack this tread. Regards all Sorry for hijacking this tread. Can someone take time to clarify install process on softraid. I read the article on undeadly.org, preform the install on sd2, my softraid0, copy kernels on sd0a and sd1a, everything went well, and after reboot, it does not find bootable media. I put install cd, drop to shell, mount sd0a, then sd1a and then sd2a, everything is there. Do I miss some obvious step here... Can you point out what is wrong? It is AMD64 20 December snapshot on HP microserver. I cannot take dmesg at this point. I can take one from standard install if you need it. Happy Holidays! Thanks for the help, Ivo
Re: root/boot on softraid in 5.0
On Sat, Dec 24, 2011 at 02:26:26PM +0200, Ivo Chutkin wrote: Sorry for hijacking this tread. Can someone take time to clarify install process on softraid. I read the article on undeadly.org, preform the install on sd2, my softraid0, copy kernels on sd0a and sd1a, everything went well, and after reboot, it does not find bootable media. I put install cd, drop to shell, mount sd0a, then sd1a and then sd2a, everything is there. Do I miss some obvious step here... Can you point out what is wrong? The kernels are loaded from outside the softraid array (sd0a, or sd1a) by the second stage bootloader. But the second stage bootloader is loaded from the array. So, assuming sd2 is your softraid array, and you have an sd2a root partition, reboot the install cd and try: # mount /dev/sd2a /mnt # cp -p /usr/mdec/boot /mnt/boot # /usr/mdec/installboot -v /mnt/boot /usr/mdec/biosboot sd2 It is AMD64 20 December snapshot on HP microserver. I cannot take dmesg at this point. I can take one from standard install if you need it. Both amd64 and i386 use the same installboot(8) program.
Re: root/boot on softraid in 5.0
On 24.12.2011 P3. 15:06 Q., Josh Grosse wrote: On Sat, Dec 24, 2011 at 02:26:26PM +0200, Ivo Chutkin wrote: Sorry for hijacking this tread. Can someone take time to clarify install process on softraid. I read the article on undeadly.org, preform the install on sd2, my softraid0, copy kernels on sd0a and sd1a, everything went well, and after reboot, it does not find bootable media. I put install cd, drop to shell, mount sd0a, then sd1a and then sd2a, everything is there. Do I miss some obvious step here... Can you point out what is wrong? The kernels are loaded from outside the softraid array (sd0a, or sd1a) by the second stage bootloader. But the second stage bootloader is loaded from the array. So, assuming sd2 is your softraid array, and you have an sd2a root partition, reboot the install cd and try: # mount /dev/sd2a /mnt # cp -p /usr/mdec/boot /mnt/boot # /usr/mdec/installboot -v /mnt/boot /usr/mdec/biosboot sd2 It is AMD64 20 December snapshot on HP microserver. I cannot take dmesg at this point. I can take one from standard install if you need it. Both amd64 and i386 use the same installboot(8) program. Hi Josh, I try it but it does not boot as well. Here is what I do step by step: 1.boot from cd 2.drop to shell 3.cd /dev, sh MAKEDEV sd1 4.fdisk -iy sd0 and sd1 5.disklabel -E sd0, add a, 100m, add d, the rest of disk, raid fail system 6. disklabel -R sd1 disklabel.sd1 7. bioctl -c 1 -l sd0d,sd1d softraid0 8.^D 9.install 10. when I am asked which one I wish to initiate i choose sd2 which is my raid partition. 11. continue with normal install and / is on sd2a 12. newfs sd0a and sd1a 13. copy kernels to sd0a and sd1a everything went well I tried installboot step before and after reboot, it does not boot. I will change hardware now to check if it could be some HP specific problem. Thanks, Ivo
Re: root/boot on softraid in 5.0
On 24.12.2011 P3. 15:42 Q., Ivo Chutkin wrote: On 24.12.2011 P3. 15:06 Q., Josh Grosse wrote: On Sat, Dec 24, 2011 at 02:26:26PM +0200, Ivo Chutkin wrote: Sorry for hijacking this tread. Can someone take time to clarify install process on softraid. I read the article on undeadly.org, preform the install on sd2, my softraid0, copy kernels on sd0a and sd1a, everything went well, and after reboot, it does not find bootable media. I put install cd, drop to shell, mount sd0a, then sd1a and then sd2a, everything is there. Do I miss some obvious step here... Can you point out what is wrong? The kernels are loaded from outside the softraid array (sd0a, or sd1a) by the second stage bootloader. But the second stage bootloader is loaded from the array. So, assuming sd2 is your softraid array, and you have an sd2a root partition, reboot the install cd and try: # mount /dev/sd2a /mnt # cp -p /usr/mdec/boot /mnt/boot # /usr/mdec/installboot -v /mnt/boot /usr/mdec/biosboot sd2 It is AMD64 20 December snapshot on HP microserver. I cannot take dmesg at this point. I can take one from standard install if you need it. Both amd64 and i386 use the same installboot(8) program. Hi Josh, I try it but it does not boot as well. Here is what I do step by step: 1.boot from cd 2.drop to shell 3.cd /dev, sh MAKEDEV sd1 4.fdisk -iy sd0 and sd1 5.disklabel -E sd0, add a, 100m, add d, the rest of disk, raid fail system 6. disklabel -R sd1 disklabel.sd1 7. bioctl -c 1 -l sd0d,sd1d softraid0 8.^D 9.install 10. when I am asked which one I wish to initiate i choose sd2 which is my raid partition. 11. continue with normal install and / is on sd2a 12. newfs sd0a and sd1a 13. copy kernels to sd0a and sd1a everything went well I tried installboot step before and after reboot, it does not boot. I will change hardware now to check if it could be some HP specific problem. Thanks, Ivo Hello again, The problem was in HP microserver bios. I have to change SATA mode from AHCI to IDE When I moved disks to some ordinary PC everything went well and I realise what could be the problem. Below is dmesg from it. Thanks for the help, Ivo Here is dmesg from HP microserver running from softraid :-) # dmesg OpenBSD 5.0-current (GENERIC.MP) #159: Tue Dec 20 11:02:37 MST 2011 dera...@amd64.openbsd.org:/usr/src/sys/arch/amd64/compile/GENERIC.MP real mem = 2011758592 (1918MB) avail mem = 1944121344 (1854MB) mainbus0 at root bios0 at mainbus0: SMBIOS rev. 2.6 @ 0xfb330 (35 entries) bios0: vendor HP version O41 date 04/02/2011 bios0: HP ProLiant MicroServer acpi0 at bios0: rev 2 acpi0: sleep states S0 S4 S5 acpi0: tables DSDT FACP APIC MCFG SPMI OEMB HPET EINJ BERT ERST HEST SSDT acpi0: wakeup devices PCE2(S4) PCE3(S4) PCE4(S4) PCE5(S4) PCE6(S4) PCE7(S4) PCE9(S4) PCEA(S4) PCEB(S4) PCEC(S4) SBAZ(S4) P0PC(S4) PE20(S4) PE21(S4) PE22(S4) PE23(S4) acpitimer0 at acpi0: 3579545 Hz, 32 bits acpimadt0 at acpi0 addr 0xfee0: PC-AT compat cpu0 at mainbus0: apid 0 (boot processor) cpu0: AMD Turion(tm) II Neo N40L Dual-Core Processor, 1497.77 MHz cpu0: FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CFLUSH,MMX,FXSR,SSE,SSE2,HTT,SSE3,MWAIT,CX16,POPCNT,NXE,MMXX,FFXSR,LONG,3DNOW2,3DNOW cpu0: 64KB 64b/line 2-way I-cache, 64KB 64b/line 2-way D-cache, 1MB 64b/line 16-way L2 cache cpu0: ITLB 32 4KB entries fully associative, 16 4MB entries fully associative cpu0: DTLB 48 4KB entries fully associative, 48 4MB entries fully associative cpu0: apic clock running at 199MHz cpu1 at mainbus0: apid 1 (application processor) cpu1: AMD Turion(tm) II Neo N40L Dual-Core Processor, 1497.52 MHz cpu1: FPU,VME,DE,PSE,TSC,MSR,PAE,MCE,CX8,APIC,SEP,MTRR,PGE,MCA,CMOV,PAT,PSE36,CFLUSH,MMX,FXSR,SSE,SSE2,HTT,SSE3,MWAIT,CX16,POPCNT,NXE,MMXX,FFXSR,LONG,3DNOW2,3DNOW cpu1: 64KB 64b/line 2-way I-cache, 64KB 64b/line 2-way D-cache, 1MB 64b/line 16-way L2 cache cpu1: ITLB 32 4KB entries fully associative, 16 4MB entries fully associative cpu1: DTLB 48 4KB entries fully associative, 48 4MB entries fully associative ioapic0 at mainbus0: apid 2 pa 0xfec0, version 21, 24 pins acpimcfg0 at acpi0 addr 0xe000, bus 0-255 acpihpet0 at acpi0: 14318180 Hz acpi0: unable to load \\_SB_._INI.EXH1 acpiprt0 at acpi0: bus 0 (PCI0) acpiprt1 at acpi0: bus 1 (P0P1) acpiprt2 at acpi0: bus -1 (PCE2) acpiprt3 at acpi0: bus 2 (PCE4) acpiprt4 at acpi0: bus 3 (PCE6) acpicpu0 at acpi0: PSS acpicpu1 at acpi0: PSS acpibtn0 at acpi0: PWRB ipmi at mainbus0 not configured cpu0: 1497 MHz: speeds: 1500 1300 1000 800 MHz pci0 at mainbus0 bus 0 pchb0 at pci0 dev 0 function 0 AMD RS880 Host rev 0x00 ppb0 at pci0 dev 1 function 0 vendor Hewlett-Packard, unknown product 0x9602 rev 0x00 pci1 at ppb0 bus 1 vga1 at pci1 dev 5 function 0 ATI Mobility Radeon HD 4200 rev 0x00 wsdisplay0 at vga1 mux 1: console (80x25, vt100 emulation) wsdisplay0: screen 1-5 added (80x25, vt100 emulation) radeondrm0 at vga1: apic 2 int 18 drm0 at radeondrm0 ppb1 at pci0 dev 4 function 0 AMD RS780 PCIE
Re: root/boot on softraid in 5.0
On Wed, Dec 21, 2011 at 08:08:08PM -0800, Josh Grosse wrote: Woops. I misread your post. The commits were September 19, which is -current, beyond 5.0-release. You must either migrate to -current, or await 5.1-release. Ah, ok, thanks for the clarification. The installboot piece that lets you install bootblocks on softraid is in 5.0, so when that part worked I assumed it all was. The dates (commits in Sept, your post in Oct, 5.0 release in Nov) also led me to misbelieve it was in 5.0. But looking at the changelogs I see the bits that store boot info in softraid metadata and dynamically figure out the root happened after the 5.0 freeze. Something to look forward to in 5.1 :). Thanks again... -- Paul B. Henson | (909) 979-6361 | http://www.csupomona.edu/~henson/ Operating Systems and Network Analyst | hen...@csupomona.edu California State Polytechnic University | Pomona CA 91768
Re: root/boot on softraid in 5.0
On 12/22/2011 09:07 PM, Paul B. Henson wrote: On Wed, Dec 21, 2011 at 08:08:08PM -0800, Josh Grosse wrote: Woops. I misread your post. The commits were September 19, which is -current, beyond 5.0-release. You must either migrate to -current, or await 5.1-release. Ah, ok, thanks for the clarification. The installboot piece that lets you install bootblocks on softraid is in 5.0, so when that part worked I assumed it all was. The dates (commits in Sept, your post in Oct, 5.0 release in Nov) also led me to misbelieve it was in 5.0. But looking at the changelogs I see the bits that store boot info in softraid metadata and dynamically figure out the root happened after the 5.0 freeze. Something to look forward to in 5.1 :). Thanks again... Why wait? I have two amd64 servers in production on -current and all is humming along magnificently. One is heavily loaded by large image file manipulation over samba. Of course that has little to do with the boot block business. Both servers were booted more than a dozen times during provisioning so I feel good that they will come up after an extended power outage. I did boot with sd0 removed to force a boot from sd1. I then took a third disk and rebuilt sd0 while the machine was up and running. All went well. To me it was worth the trouble to pick a snapshot between commit storms by the developers, just so I didn't have to deal with the altroot thing any more. I have been using OBSD for Internet facing infrastructure for the last 10 years but always used RedHat/CentOS for internal servers. In Oct I installed the first of the two and don't see going back. I am so done with Linux servers. OBSD is getting close to the point where I can use it as a workstation/desktop. My biggest hindrance is no wine. I guess I am going to be forced to make friends with some VM system and cross mount the filesystems via samba. I assume smarter people than me have done that already. I would take suggestions of a preferred VM off-list so as not to hijack this tread. Regards all
Re: root/boot on softraid in 5.0
Paul B. Henson henson at acm.org writes: I've been running openbsd 4.6 for a couple years now with root on softraid, booting off a CF card with a kernel compiled to hardcode root/swap on sd0. I read about official support for root on softraid Anyway, just to clarify my understanding, is it expected in 5.0 to be able to boot softraid root without a custom kernel or using -a? If so, what am I doing wrong? Thanks... See if http://www.undeadly.org/cgi?action=articlesid=20111002154251 provides what your looking for. Note in particular, the installboot(8) magic. The kernel still needs to be on a non-RAID partition, and only the kernel. The bootblocks point to a second stage bootloader within the RAID array.
Re: root/boot on softraid in 5.0
Josh Grosse josh at jggimi.homeip.net writes: Paul B. Henson henson at acm.org writes: I've been running openbsd 4.6 for a couple years now with root on softraid, booting off a CF card with a kernel compiled to hardcode root/swap on sd0. I read about official support for root on softraid Woops. I misread your post. The commits were September 19, which is -current, beyond 5.0-release. You must either migrate to -current, or await 5.1-release.
