Re: [gentoo-user] Re: remote installation, dual boot
On 24.04.2015 13:56, J. Roeleveld wrote: dracut has to assemble /dev/md3 at first (this is the single PV in the VG VG01) and /dev/VG01/genroot then is the filesystem with the new gentoo-rootfs I only added stuff like rd.md=1 etc ... I think I got the assembling wrong. And maybe even the root= option. This is one of the reasons why I have given up on genkernel and dracut for initramfs creation and now simply build my own and have it integrated into the kernel: $ zcat /proc/config.gz | grep INITRAMFS_SOURCE CONFIG_INITRAMFS_SOURCE=/usr/src/initramfs/config If you're interested I will send you the required files. Yes, I am. I looked at generating it via dracut and get confused a bit. I try to generate it on another gentoo-host by editing a specific dracut.conf whic contains the md-related lines: kernel_cmdline+= rd.md.uuid=6a6226ed:87d41201:76269125:1a17f6a4 kernel_cmdline+= rd.md.uuid=2d6cd278:5f966c0f:ac5ed5c3:5a0bb8f5 kernel_cmdline+= rd.md.uuid=b492f31d:96bfca88:8cd97590:ad997d2c kernel_cmdline+= rd.md.uuid=e848b637:ca2bde73:9f92f3cc:128cdbad (to assemble the raid-arrays at boot) I run dracut -m lvm mdraid bash -c dracut.conf -f init2.img 3.18.9-gentoo and it ends with: *** Store current command line parameters *** Stored kernel commandline: rd.md.uuid=6a6226ed:87d41201:76269125:1a17f6a4 rd.md.uuid=2d6cd278:5f966c0f:ac5ed5c3:5a0bb8f5 rd.md.uuid=b492f31d:96bfca88:8cd97590:ad997d2c rd.md.uuid=e848b637:ca2bde73:9f92f3cc:128cdbad root=UUID=6eafd21c-c5f4-496d-bb90-ab4dc0a2c93c rootflags=rw,noatime,stripe=64,data=ordered rootfstype=ext4 *** Creating image file *** *** Creating image file done *** ... so the root= is looked up from the host I run the command on and not fitting the target machine ... etc etc hmm Does root= inside the grub-config override these lines? Should I use dracut --no-hostonly-cmdline ? Way too much variables in here ... especially when I can't just press Reset and try again. Stefan
Re: [gentoo-user] Re: remote installation, dual boot
On 28 April 2015 10:33:32 CEST, Stefan G. Weichinger li...@xunil.at wrote: On 24.04.2015 13:56, J. Roeleveld wrote: dracut has to assemble /dev/md3 at first (this is the single PV in the VG VG01) and /dev/VG01/genroot then is the filesystem with the new gentoo-rootfs I only added stuff like rd.md=1 etc ... I think I got the assembling wrong. And maybe even the root= option. This is one of the reasons why I have given up on genkernel and dracut for initramfs creation and now simply build my own and have it integrated into the kernel: $ zcat /proc/config.gz | grep INITRAMFS_SOURCE CONFIG_INITRAMFS_SOURCE=/usr/src/initramfs/config If you're interested I will send you the required files. Yes, I am. I looked at generating it via dracut and get confused a bit. I try to generate it on another gentoo-host by editing a specific dracut.conf whic contains the md-related lines: kernel_cmdline+= rd.md.uuid=6a6226ed:87d41201:76269125:1a17f6a4 kernel_cmdline+= rd.md.uuid=2d6cd278:5f966c0f:ac5ed5c3:5a0bb8f5 kernel_cmdline+= rd.md.uuid=b492f31d:96bfca88:8cd97590:ad997d2c kernel_cmdline+= rd.md.uuid=e848b637:ca2bde73:9f92f3cc:128cdbad (to assemble the raid-arrays at boot) I run dracut -m lvm mdraid bash -c dracut.conf -f init2.img 3.18.9-gentoo and it ends with: *** Store current command line parameters *** Stored kernel commandline: rd.md.uuid=6a6226ed:87d41201:76269125:1a17f6a4 rd.md.uuid=2d6cd278:5f966c0f:ac5ed5c3:5a0bb8f5 rd.md.uuid=b492f31d:96bfca88:8cd97590:ad997d2c rd.md.uuid=e848b637:ca2bde73:9f92f3cc:128cdbad root=UUID=6eafd21c-c5f4-496d-bb90-ab4dc0a2c93c rootflags=rw,noatime,stripe=64,data=ordered rootfstype=ext4 *** Creating image file *** *** Creating image file done *** ... so the root= is looked up from the host I run the command on and not fitting the target machine ... etc etc hmm Does root= inside the grub-config override these lines? Should I use dracut --no-hostonly-cmdline ? Way too much variables in here ... especially when I can't just press Reset and try again. Stefan I sent my files on a seperate thread last weekend. Did you see those? -- Joost -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
Re: [gentoo-user] Re: remote installation, dual boot
On 28.04.2015 17:30, J. Roeleveld wrote: I sent my files on a seperate thread last weekend. Did you see those? no ... got to dig, thanks
Re: [gentoo-user] Re: remote installation, dual boot
On 28 April 2015 18:03:28 CEST, Stefan G. Weichinger li...@xunil.at wrote: On 28.04.2015 17:42, J. Roeleveld wrote: On 28 April 2015 17:37:06 CEST, Stefan G. Weichinger li...@xunil.at wrote: On 28.04.2015 17:30, J. Roeleveld wrote: I sent my files on a seperate thread last weekend. Did you see those? no ... got to dig, thanks Sent on 25th. Initramfs in subject. Happy hunting. :) found, but too tired today to give that a try ... thanks S Oki. Comments and questions about that please in the other thread. With sufficient demand, I will collate the information and post it on the wiki. -- Joost -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
Re: [gentoo-user] Re: remote installation, dual boot
On 28.04.2015 16:53, Stefan G. Weichinger wrote: I now booted from a recent sysresccd ... and it didn't assemble/detect the LVM-PV ... *sigh* So I couldn't even mount the LV to chroot in there. The mdadm-RAID-devices were there and active. I didn't want to screw around with the LVM too much as there is data on there (yes, there are backups, but ...). pvscan always told me something like: No device found for PV hMMm0m-w4Ds-SU77-ATjF-iF1A-0qfq-V2IcXb ok, one step further: (no exact cut and paste because of java-based KVM access) pvscan --cache tells me: Found duplicate PV hMMm0m-w4Ds-SU77-ATjF-iF1A-0qfq-V2IcXb: using /dev/sdj1 not /dev/md3 -- md3 is the correct array to form the PV ... /dev/sdj1 is actually a partition that is an active *member* of the array /dev/md3 I tried to filter this partition by editing lvm.conf without success. Under suse I don't see /dev/sdj1 as pv: # pvdisplay /dev/sdj1 Failed to read physical volume /dev/sdj1 - Should I * pvremove /dev/sdj1 from sysresccd * in suse: remove sdj1 from array md3 ... overwrite header and re-add ? Oh my. !! BUT - I now found another nice fact: /dev/md2 (NO LVM) is unused since the initial installation. I even *documented* that in the project wiki years ago. 40 GB of unused space on a RAID1 of SAS hdds. Nice! Sounds like my new rootfs ;-) dracut plus these md-lines should work then. Stefan
Re: [gentoo-user] Re: remote installation, dual boot
On 28.04.2015 18:43, Stefan G. Weichinger wrote: 40 GB of unused space on a RAID1 of SAS hdds. Nice! Sounds like my new rootfs ;-) dracut plus these md-lines should work then. call me coward or defensive: I used genkernel ... and now the server is up and running with gentoo linux ;-) OK, way too much modules for now etc ... but booted. btw: I learned about the kernel-option panic=N ... maybe this would help in case the kernel does not detect the rootfs? thanks all, Stefan
Re: [gentoo-user] Re: remote installation, dual boot
I now booted from a recent sysresccd ... and it didn't assemble/detect the LVM-PV ... *sigh* So I couldn't even mount the LV to chroot in there. The mdadm-RAID-devices were there and active. I didn't want to screw around with the LVM too much as there is data on there (yes, there are backups, but ...). pvscan always told me something like: No device found for PV hMMm0m-w4Ds-SU77-ATjF-iF1A-0qfq-V2IcXb When I boot into Suse again, it shows up fine: # pvdisplay --- Physical volume --- PV Name /dev/md3 VG Name VG01 PV Size 4,09 TB / not usable 4,00 TB Allocatable yes PE Size (KByte) 4096 Total PE 1072103 Free PE 242919 Allocated PE 829184 PV UUID hMMm0m-w4Ds-SU77-ATjF-iF1A-0qfq-V2IcXb Do I hit some compatibility-issue? Stefan
Re: [gentoo-user] Re: remote installation, dual boot
On 28 April 2015 17:37:06 CEST, Stefan G. Weichinger li...@xunil.at wrote: On 28.04.2015 17:30, J. Roeleveld wrote: I sent my files on a seperate thread last weekend. Did you see those? no ... got to dig, thanks Sent on 25th. Initramfs in subject. Happy hunting. :) -- Sent from my Android device with K-9 Mail. Please excuse my brevity.
Re: [gentoo-user] Re: remote installation, dual boot
On Thursday, April 23, 2015 07:21:12 PM Stefan G. Weichinger wrote: On 23.04.2015 18:57, james wrote: Stefan G. Weichinger lists at xunil.at writes: On 23.04.2015 17:43, Stefan G. Weichinger wrote: I will see if I get that figured out ... got to get the kernel-options and dracut modules right. first try started booting but couldn't mount rootfs ... and now the box hangs there and I can't press the power button (only PS/2 attached keyboard/mouse) got to wait for the folks there to come to work tmrw There are methods to get the console terminal session, remotely. One, is called netconsole [1]. I use to use this sort of thing decades ago. How it is currently supported and the associated codes (ssh) to use it remotely, I have not kept up with. Some headless embedded system still use this sort of thing; with holes punched in the routers/firewalls. I've also heard of folks setting up a serial console over serial-2-usb converter, like ftdi, and then making a terminal session connection form a remote cell phone to another cell phone, to tunnel an ascii terminal session. I've never done this, but I'd be really curious how somebody makes that sort of 9600 baud terminal connection to a remote router or server, via a serial/console port using cell phones as modems. Thanks for the hint but for now I have to wait. When I think about I have to add several kernel-options to menu.lst: dracut has to assemble /dev/md3 at first (this is the single PV in the VG VG01) and /dev/VG01/genroot then is the filesystem with the new gentoo-rootfs I only added stuff like rd.md=1 etc ... I think I got the assembling wrong. And maybe even the root= option. This is one of the reasons why I have given up on genkernel and dracut for initramfs creation and now simply build my own and have it integrated into the kernel: $ zcat /proc/config.gz | grep INITRAMFS_SOURCE CONFIG_INITRAMFS_SOURCE=/usr/src/initramfs/config If you're interested I will send you the required files. We will see tmrw if I get in early from other jobs. Good luck :) -- Joost
Re: [gentoo-user] Re: remote installation, dual boot
On 23.04.2015 18:57, james wrote: Stefan G. Weichinger lists at xunil.at writes: On 23.04.2015 17:43, Stefan G. Weichinger wrote: I will see if I get that figured out ... got to get the kernel-options and dracut modules right. first try started booting but couldn't mount rootfs ... and now the box hangs there and I can't press the power button (only PS/2 attached keyboard/mouse) got to wait for the folks there to come to work tmrw There are methods to get the console terminal session, remotely. One, is called netconsole [1]. I use to use this sort of thing decades ago. How it is currently supported and the associated codes (ssh) to use it remotely, I have not kept up with. Some headless embedded system still use this sort of thing; with holes punched in the routers/firewalls. I've also heard of folks setting up a serial console over serial-2-usb converter, like ftdi, and then making a terminal session connection form a remote cell phone to another cell phone, to tunnel an ascii terminal session. I've never done this, but I'd be really curious how somebody makes that sort of 9600 baud terminal connection to a remote router or server, via a serial/console port using cell phones as modems. Thanks for the hint but for now I have to wait. When I think about I have to add several kernel-options to menu.lst: dracut has to assemble /dev/md3 at first (this is the single PV in the VG VG01) and /dev/VG01/genroot then is the filesystem with the new gentoo-rootfs I only added stuff like rd.md=1 etc ... I think I got the assembling wrong. And maybe even the root= option. We will see tmrw if I get in early from other jobs. Stefan