[00:35] I came up with a workaround [00:35] if [ -s "$LOGDIR/additional.var" ]; then [00:35] . $LOGDIR/additional.var [00:35] fi [00:35] in the top of the 50 script [00:35] crude, but effective [00:36] DavidDreezer (Adium@c-71-227-156-146.hsd1.wa.comcast.net) left #fai. [06:35] pav5088 (~Mark@120.153.210.32) joined #fai. [06:54] pav5088 (~Mark@120.153.210.32) left irc: Ping timeout: 480 seconds [10:50] ErKa (~keryell@keryell.pck.nerim.net) joined #fai. [11:30] sep_oslo (~sep@87.238.46.238) left irc: Quit: Leaving [11:46] al (d@niel.cx) left irc: Ping timeout: 480 seconds [11:47] al (d@niel.cx) joined #fai. [13:25] i tried daily squeeze iso and installed it on the raid created by fai. this workde well.... But i still fail installing FAI system using grub-pc on raid1 with fai 3.4.5 on squeeze... [13:32] alekibango, could try manually executing grub commands to install it? [13:32] i did once, but it failed to boot :) [13:33] will try now again [13:33] so just let it install and don't reboot at the end [13:33] first see which errors are shown during execution of scripts [13:33] and then let's do the manual tries [13:33] mt: i already pasted those yesterday imho [13:33] its 'unknown device' [13:34] or somthing like that [13:34] yes, ok [13:34] http://pastebin.com/uW2dw195 [13:34] have you already got to the point that the install is complete (but probably broken)? [13:34] you mean end of fai -- pres [enter] to reboot? [13:35] yep [13:35] but please don't reboot and get a shell instead [13:35] I suppose chroot /target grub-mkdevicemap --no-floppy will work fine [13:36] please paste /tmp/fai/disk_var.sh [13:37] Action: mt is hoping to finally have a confirmed-to-work solution for sw-raid+grub-pc [13:55] Sp4rKy_ (~Sp4rKy@rennes1.dunnewind.net) left irc: Quit: Changing server [13:56] alekibango, I'll go and prepare lunch; please be patient in case I don't reply rightaway [14:12] k' i was talkingto jehova whitnesses for 30 minutes.... poor souls they will not sleep tonight... [14:26] mt are u here? [14:39] alekibango, back now [16:18] siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) joined #fai. [19:05] justleen (~leen@s5591f74f.adsl.wanadoo.nl) joined #fai. [19:58] justleen (~leen@s5591f74f.adsl.wanadoo.nl) left irc: Quit: Leaving. [21:00] DavidDreezer (~Adium@c-71-227-156-146.hsd1.wa.comcast.net) joined #fai. [21:01] MT, are you there by chance? [21:01] yes :-) [21:01] Good afternoon [21:02] it is afternoon there, no? [21:02] well, for me it's rather evening (10pm) [21:02] oh, well, good evening then [21:02] thank you for the email that you sent to me, I appreciate it. [21:02] you're welcome! [21:02] I got it working! [21:02] sorry for leaving you all alone on this [21:03] so how did you fix it? [21:03] I have an end to end trial run this afternoon [21:03] mt i wrote some debian bug. about the grub [21:03] for today I put this in the head of the 50 script [21:03] if [ -s "$LOGDIR/additional.var" ]; then [21:03] . $LOGDIR/additional.var [21:03] fi [21:03] ok [21:03] Ultimately, Micheal, I will merge my 45 and the 50 into on script [21:03] so they will be in the same shell [21:04] but then, really, renaming the 45 script to 45-mac.source should do the trick [21:04] oh, you're right [21:04] and merging will, of course, do as well :-) [21:04] alekibango, which bug# is it? [21:04] do not know yet :ú [21:04] I prefer to keep my modifications in a different script for future upgrade purposes, making it into a .source wouldbe excellent [21:05] and i forgot to include some info... but the core is there [21:05] now, I have an end to end FAI install on software raid. [21:05] I believe this should work like a charme, unless fai-class is buggy [21:05] oh, software raid [21:05] cool [21:05] alekibango just had problems with his sw raid :-) [21:05] well, so I tried the 10-setup in ~45 as requested [21:06] and it did not work :( [21:06] there was no error logged [21:06] yes, kind of expected [21:06] alekibango and me debugged this all afternoon [21:06] however, upon booting the machine, error 15 then recovery prompt [21:06] ah! [21:06] we believe to have found a working solution [21:06] it would actually be great if you could try to confirm that [21:07] i substuted the 10-setup that I had posted and that worked, but of course that is not a scalable solution [21:07] i would love to [21:07] alekibango, may I copy the link to the pastebin in here? [21:07] y [21:07] for the working 10-setup? [21:07] np, dont ask all the time :) you are welcome [21:07] haha :) [21:08] http://pastebin.com/4qQWGXec [21:08] I believe it can actually be simplified [21:08] DavidDreezer: he asked me like this 10 times today already :) [21:09] bah, i've been fiddling with your system [21:09] i'm comparing that to mine Michael. [21:09] hmm [21:09] you can, most probably, remove the mdraid [21:09] an essential part is the sed line [21:09] grub guys thought i should use grub from sources [21:09] to test it [21:10] omg [21:10] ok, sure, that's kind of expected [21:10] well, there was only 1 of them alive and he was not a grub devel [21:10] is grub-setup required? [21:10] seems like weekend [21:11] most probably it suffices if you put that on a running system, compile it, and run the grub-probe command to see what you get [21:11] grub-setup? [21:11] DavidDreezer: grub-install calls grub-setup [21:11] thx [21:11] if you use probes and grub setup you can avoid grub-install bugs [21:11] :D [21:12] who says that? [21:12] well, just me, making fun a bit [21:12] grub-install /dev/md0 should be the way to use it [21:12] could easily be that this is true :-) [21:12] I also follow it with $ROOTCMD update-initramfs -k all -u however I don't think it to be necessary [21:13] not for grub, no [21:13] but it might be necessary for other reasons [21:13] let me try this script, it only takes but a minute for me to copy/paste it into my server then reboot my client. [21:13] if you get in trouble after grub has successfully loaded the kernel [21:14] mt: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=606035 [21:14] the update-initramfs could be necessary if the new kernel was installed before mdadm was present, then the initrd lacks mdadm support and hence your system might fail [21:14] but grub will work nicely without the update-initramfs [21:14] cool, thx [21:15] ok, script installed, let me reboot my client and we shall see what we see. [21:15] mt: look near end [21:15] heh i used czech locales, darn [21:16] well, I guess people can translate that [21:17] even though it might be nice to post a follow-up with that last bit translated [21:17] it might be easier for grub-devs to grep through the code [21:17] so actually grub-setup must be broken!? [21:17] btw, could you also do a run using (md0) and post just the last bit [21:18] mt: its explained in text... [21:18] that there is no /dev/md/0 [21:18] would that mean that (md0) is passed on to grub-setup? [21:18] mt: no. (md/0) is passed [21:18] and it doesnt work [21:18] (md0) should be passed - works [21:18] explained, yes, but the very precise wording helps so much when using grep [21:18] :-) [21:19] i believe debian should upgrade its bug reporting system [21:19] launchpad is so cool compared to bug reporting in debian [21:20] (still it sucks inside) [21:20] really? [21:20] well, installing it is pain [21:20] I never quite got how to use launchpad [21:20] it touches files all over the machine [21:20] the debian BTS is so simple :-) [21:20] mt: 1) i hate mails. i have almost 40k unread mails in my box [21:20] 2) i hate sending mails [21:21] OMG! [21:21] lol, 40k! [21:21] and its all based on mails! [21:21] ah, ok, so launchpad allows web bug filing? [21:21] LD [21:21] :)) [21:21] mt: right [21:21] well, ok, so /me is just different [21:21] i still read mails in mutt btw :) [21:21] I strictly prefer email (as you already noticed...) [21:21] so do I [21:22] i hate even administering mail servers... :) its wrongly created service, should be memoryholed [21:23] I don't have the luxury of telling my customers "no email" :( [21:24] for me, email mostly is a form of instant messaging [21:24] alekibango, is grub-setup a script or compiled code? [21:25] compiled [21:25] ok, then I'll leave this to grub devs to debug :-) [21:26] ok, at faiend with this script, let me check for errors then reboot [21:26] even though it would probably be considered a nice service to the grub-devs if you could indeed compile their latest version and probably just try calling grub-setup [21:26] no need to install anything, just compile [21:26] we can even do this in live system [21:26] this version errored [21:26] if aufs will be able to keep it ok [21:26] :D [21:26] fai.log:GRUB_PC/10-setup FAILED with exit code 1. [21:26] shell.log:grub-mkimage: error: cannot stat /usr/lib/grub/i386-pc/mdraid.mod [21:26] shell.log:++ error=1 [21:26] shell.log:GRUB_PC/10-setup FAILED with exit code 1. [21:26] status.log:GRUB_PC/10-setup FAILED with exit code 1. [21:27] DavidDreezer, which version of grub are you using [21:27] grub2 [21:27] ah, you're running lenny [21:27] is that right [21:27] ? [21:27] this server is open to the world, by the way, if you want access to it [21:27] that would make the second one today :-) [21:28] :) [21:28] DavidDreezer: which grub2? [21:28] well, maybe, yes [21:28] got an IP for me? [21:28] which version of grub-pc ? last squeeze ? [21:28] our office is not, of course, but my house, i have opened so that I can reach it from the office [21:28] root@faiserver:/var/log/fai/db099.dh01.groupee-inc.net/last# dpkg --list | grep grub [21:28] ii grub 0.97-47lenny2 GRand Unified Bootloader (Legacy version) [21:28] ii grub-common 1.96+20080724-16 GRand Unified Bootloader, version 2 (common files) [21:28] 1.96+20080724-16 - this is somewhat old [21:29] :) [21:29] well, maybe it's not quite as broken as yours :-) [21:29] i did an apt-get update, upgrade yesterday [21:29] is it lenny? [21:29] yes [21:29] ah :) this script might need to have different solution for lenny [21:29] :) [21:30] I think all it takes should be dropping the mdraid [21:30] DavidDreezer, could you just go into a console (instead of rebooting) [21:30] i'd been unable to get sw raid working under fai 3.x or old grub [21:30] certainly can [21:30] it is at a consoile now [21:30] and type the commands one-by one [21:30] so well, [21:30] export ROOTCMD="chroot /target" [21:31] so you can copy them off from the pastebin or from the script [21:31] # [21:31] $ROOTCMD grub-mkdevicemap --no-floppy [21:31] well, I'll do the copying in here (with slight modifications) [21:32] $ROOTCMD grub-probe -tdrive -d /dev/md0 [21:32] so run first [21:32] $ROOTCMD grub-mkdevicemap --no-floppy [21:32] (assuming that BOOT_DEVICE is /dev/md0) [21:32] yes? [21:32] yes [21:32] it is [21:32] or should be [21:32] cat /tmp/fai/disk_var.sh [21:32] should tell [21:32] SWAPLIST=${SWAPLIST:-"/dev/md1"} [21:32] BOOT_DEVICE=${BOOT_DEVICE:-"/dev/md0"} [21:32] ROOT_PARTITION=${ROOT_PARTITION:-/dev/md2} [21:33] should be fine [21:33] what does grub-probe say? [21:33] when i ran ROOTCMD grub-mkdevicemap no-floppy i got the prompt, with no response of any kind [21:33] just back to prompt [21:33] that's fine [21:34] no news is good news :-) [21:34] but grub-probe should give some news :-) [21:34] db099:~# $ROOTCMD grub-probe -tdrive -d /dev/md0 [21:34] (md0) [21:34] db099:~# [21:34] which is correct, no? [21:35] alekibango, this "old" version is much better than yours :-) [21:35] yes, that's great [21:35] $ROOTCMD grub-install --no-floppy --modules="lvm raid" "(md0)" [21:35] would be the next step [21:35] oh, wait though [21:36] ? [21:36] this is the script [21:36] $ROOTCMD grub-install --no-floppy --modules="lvm raid mdraid" "$GROOT" [21:36] and the error was could not find module mdraid [21:36] yes, but I believe mdraid isn't necessary [21:36] so use the above command [21:36] shell.log:grub-mkimage: error: cannot stat /usr/lib/grub/i386-pc/mdraid.mod [21:37] without mdraid [21:37] i think you're right [21:37] it appears to have been successful [21:37] db099:~# $ROOTCMD grub-install --no-floppy --modules="lvm raid" "(md0)" [21:37] Installation finished. No error reported. [21:37] This is the contents of the device map /boot/grub/device.map. [21:37] Check if this is correct or not. If any of the lines is incorrect, [21:37] fix it and re-run the script `grub-install'. [21:37] (hd0) /dev/hda [21:37] (hd1) /dev/hdb [21:37] sounds good [21:38] shall I try to reboot? [21:38] nope [21:38] $ROOTCMD update-grub [21:38] oops [21:38] you already did? [21:38] no [21:38] ok :-) [21:38] oops as in "i forgot that step" [21:38] it wouldn't have been too bad [21:38] success [21:38] db099:~# $ROOTCMD update-grub [21:38] Updating /boot/grub/grub.cfg ... [21:38] Found linux image: /boot/vmlinuz-2.6.26-2-686-bigmem [21:38] Found initrd image: /boot/initrd.img-2.6.26-2-686-bigmem [21:38] Found memtest86+ image: /boot/memtest86+.bin [21:38] done [21:38] just no menu of any kind [21:39] well, then please give it a try ... [21:39] will do, cross fingers [21:39] with alekibango we afterwards even verified that corrupting either of the two disks still left the system in a bootable state [21:41] grub rescue mode. :( [21:41] is this due to my partition topography? [21:42] hmm [21:43] can you do "ls" [21:44] yes, i have to bring my laptop down to where i have the computer physically. I might disconnect momentarily [21:44] I'll learn about rescue mode meanwhile [21:45] ls works. i have all the hd0 and 1 partitions. no md of course [21:46] grub manual says [21:47] cat /grub/grub.cfg [21:47] no such disc [21:48] type "help" [21:48] is there something about "root" [21:48] makes some sense. grub is on md0, and we have no md0 yet, no? [21:48] I believe it *should* have mapped that [21:48] i can try root (hd0,1) [21:48] but that might be a wrong assumption [21:48] pls do [21:49] Action: mt thinks he might understand [21:49] dine. (hd0,1): Filesystem is ext2 [21:49] now i need a boot line, i htink [21:50] how long does a reinstall take? [21:51] 5 minutes [21:51] my takes some 300 seconds [21:51] I can do them all day [21:51] might be the quicker option then :-) [21:51] works for me [21:51] so what am I after: [21:52] grub-pc in lenny seems not to have properly working support for raid [21:52] so must be installed in each device separately [21:53] grub-install line must be run with /dev/hda1, /dev/hdb1 [21:53] (two times) [21:53] well, but this works: [21:53] $ROOTCMD /usr/bin/grub-mkimage --output=/boot/grub/core.img ext2 pc biosdisk raid [21:53] $ROOTCMD /usr/sbin/grub-install '(md0)' [21:53] $ROOTCMD /usr/sbin/grub-setup '(md0)' [21:53] $ROOTCMD /usr/sbin/update-grub [21:53] $ROOTCMD update-initramfs -k all -u [21:53] though it is not to the device [21:54] i've not got it to work installing to the device [21:54] hmm [21:54] i've tried a for look installing to each device in device map [21:54] always gives an error 15 and rescue boot [21:54] loop* [21:54] meaning this fully works including a booting system? [21:55] oh yes [21:55] remember I said wit went end to end this morning? [21:55] that was with the script i just quoted [21:55] however! [21:55] I have not tested pulling one drive to see what it does [21:55] but with both drives it boots perfectky [21:56] well, I suppose it will work just fine then [21:56] ok, could we try the following: [21:56] i would think because mdo is both drives [21:56] both or either, [21:56] do you have a script that executes the above commands? [21:56] (as part of your install) [21:56] yes, I do [21:57] want me to paste it for you? [21:57] no, it's fine [21:57] could you comment out the mkimage and grub-setup lines? [21:57] and see whether it still workds? [21:57] sorry, no, won't [21:57] my 10-setup has a case switch for HW_RAID or SW_RAID, then run the above commands for SW [21:58] pls add --modules="lvm raid" to the grub-install line [21:58] for HW your default script works perfectly [21:58] meaning we only run grub-install and update-grub [21:58] siggg (~jean@tra78-5-88-172-128-32.fbx.proxad.net) left irc: Ping timeout: 480 seconds [21:59] maybe it's just that grub-mkimage must be told about the modules [21:59] ok, this is where I am [21:59] which won't happen if run as I told you before [21:59] $ROOTCMD /usr/bin/grub-mkimage --output=/boot/grub/core.img ext2 pc biosdisk raid [21:59] $ROOTCMD /usr/sbin/grub-install --modules="lvm raid" '(md0)' [21:59] $ROOTCMD /usr/sbin/grub-setup '(md0)' [21:59] $ROOTCMD /usr/sbin/update-grub [21:59] $ROOTCMD update-initramfs -k all -u [21:59] now you want me to remove which part? [21:59] just comment out, don't remove :-) [21:59] ok [21:59] grub-mkimage and grub-setup lines [21:59] ok, just a sec [22:00] both of which should be called by grub-install anyway [22:00] now, it is hard coded to md0. do you want to make a change to that as of yet? [22:01] we'll leave this for a second round :-) [22:01] very well, let me place this and reboot the client [22:08] < wish list: someday faimond will timestamp entries [22:08] :) [22:10] faimond has the optopn -T which add a timestamp to the log. [22:10] (quoting Mrfai) [22:11] from http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=290539 [22:12] oh my so it does! [22:12] ok, faiend [22:12] let's see [22:12] any errors? [22:12] yes [22:12] fai.log:GRUB_PC/10-setup.test FAILED with exit code 1. [22:12] shell.log:grub-mkimage: error: cannot stat /usr/lib/grub/i386-pc/mdraid.mod [22:12] shell.log:++ error=1 [22:13] mind pasting 10-setup.test? [22:13] .test? [22:13] that's what you quoted [22:13] it is, and yet [22:14] there is none [22:14] damn sure? [22:14] oh my gosh I see [22:15] I moved the test script we ran earlier to test, but left it executable [22:15] which of course, it did [22:15] I'm an idiot [22:15] no, surely not [22:16] it might just mean that things have to be restarted [22:16] either of the scripts might have left something semi-broken behind [22:16] well, the actual 10 setup that we want to try did not error [22:16] it was the previous iteration that errored [22:17] as expected [22:17] so I chmod -x that [22:17] and shall reboot yet again. sorry for wasting your time [22:17] oh, np, I just keep doing other things meanwhile [22:27] no errors, now let me go see if it boots [22:30] alas. error: no such disk, then rescue mode. doing an LS shows only the hd0 and1 [22:32] hmm [22:32] so let's see what has to be added back [22:32] maybe try the mkimage first [22:33] does grub-install call grub-setup? [22:33] yes, at least it's supposed to :-) [22:34] ok, let me uncomment the mkimage and go from there [22:35] btw, you might want to do an explicit ping to make me respond faster, I'm usually acting on other desktops [22:36] ok, i'm in no hurry, it is Sunday and I feel guilty as though I am taking advantage of your time. [22:46] same problem, so I shall try it without mkimage but with grub-setup [22:47] ok [22:47] I think this will work [22:47] this would make a debuggable case [22:47] my script line is this [22:47] $ROOTCMD /usr/sbin/grub-setup '(md0)' [22:48] because grub-install is a script [22:48] and it will work, I thinkj, because I am telling it mdo [22:48]