Forum discussion: https://forums.mageia.org/en/viewtopic.php?f=15&t=2267 Possibly related bug: Bug 5489 - "dracut Warning: unable to process initqueue" on install startup when booting from USB3 https://bugs.mageia.org/show_bug.cgi?id=5489 Hardware: PC, Athlon II X2, 2GB RAM, WiFi: ZyXEL G-220, 1 HDD, 2 optical drives, Radeon 2100 graphics, bluetooth dongle I'll likely have access to this test system until this Sunday. Plan to test Dual Arch CD and Gnome Live CD on laptops (Sandy Bridge (tomorrow) and old one (next week)). Description of problem: ======================= 1st reboot from dual-arch install fails w/ dracut Warning: boot messages ... USB stuff comes up (the install medium is still inserted as sdb, but same error on next reboot if removed. I removed partitions and did same install again w/same result.) -> a short pause -> dracut Warning: Cancelling resume operation. Device not found. rm: cannot remove `/lib/dracut/hooks/initqueue/settled/resume.sh´: No such file or directory longer pause ~20sec -> [ 37.968699] dracut Warning: Unable to process initqueue dracut Warning: Unable to process initqueue dracut Warning: Cancelling resume operations. Device not dound. rm: cannot remove `/lib/dracut/hooks/initqueue/settled/resume.sh´: No such file or directory dracut Warning: Unable to process initqueue [ 38.205640] dracut Warning: "/dev/vg-mga/mgaroot" does not exist dracut Warning: "/dev/vg-mga/mgaroot" does not exist [ 38.257833] dracut Warning: "/dev/vg-mga/mgaroot" does not exist dracut Warning: "/dev/vg-mga/mgaroot" does not exist Dropping to debug shell. sh: 0: can't access tty: job conrol turned off (I copied these messages by typing; hopefully no typos.) There is no volume group in /dev. There is no password prompt. Steps to Reproduce: =================== 0. (Probably not part of minimal example, sorry) Already have Mdv w/ grub installed in /dev/sda; mga's entry: root (hd0,11) chainloader +1 1. Install from Dual-arch CD (sha1sum correct) to internal HDD via graphical installer, German, like this: /dev/sda12 /boot ext4 /dev/sda13 encrypted LLVM /dev/vg-mga mgaswap; mga-root, mga-home (both ext4) In mga's grub installed into /dev/sda12: kernel (hd0,11)/vmlinuz BOOT_IMAGE=linux root=/dev/vg-mga/mgaroot nokmsboot splash=verbose resume=UUID=blabla vga=794 initrd (hd0,11)/initrd.img This works fine in mga1 on other machine. (splash, vga and removal of quiet are manual changes.) 2. packet selection: the default selections, unchanged (Tells me I need radeon-firmware, zd1211-firmware.) 3. Reboot w/ or w/out removing USB install medium.
CC: (none) => callimera.42, mageia
Hi thanks for reporting. I've done similar setups (sans the chainloader) so things should be generally working, but there are clearly a couple edge cases left :s Firstly, when you boot, can you remove the resume= line from the command line. This will hopefully silence the errors from that bit such that we can concentrate on ensuring the real issue is addressed first and go back to it later. Incidentally, the splash=verbose syntax isn't really supported anymore. You should just have "splash quiet" for a splashy, quiet boot, and nothing if you want all the gory details. OK, so once you've booted (without the resume= arg), can you do: ls /etc/cmdline.d/ cat /etc/cmdline.d/* There should be two files in there. One to start the encrypted drive, and the other to activate the LVM inside it.
(In reply to comment #1) Well, thanks for dealing w/ bug reports! > I've done similar setups (sans the chainloader) so things should be generally > working, but there are clearly a couple edge cases left :s I'll see about a VM install to get rid of the chainloader if necessary. So now I booted with kernel (hd0,11)/vmlinuz BOOT_IMAGE=linux root=/dev/vg-mga/mgaroot nokmsboot vga=794 Exactely the same as before, but all lines referring to resume are missing. > ls /etc/cmdline.d/ 90lvm.conf > cat /etc/cmdline.d/* rd.lvm.lv=vg-mga/mgaroot (There is a leading space.) > There should be two files in there. One to start the encrypted drive, and the > other to activate the LVM inside it. Just the one. No hidden files, either.
OK, so it looks like the crypt module wasn't included in the initrd. In my tests VM tests it was included fine when creating or reusing an encrypted partition... so I guess we'll have to find out how this was installed and see if we can reproduce it. If you want to fix your system, you'll have to boot with the rescue mode and regenerate an initrd. I can give instructions here if needed.
It's just a test system, so I don't care. With your hint, I can probably figure it out for my further tests. I will want to use this function in the final release for sure, though. If you can use my help for testing, don't hesitate to ask.
Hi, This bug was filed against cauldron, but we do not have cauldron at the moment. Please report whether this bug is still valid for Mageia 2. Thanks :) Cheers, marja
Keywords: (none) => NEEDINFO
Installed Mageia 2 on test machine; no separate /home this time. Same result, except now: #cat /etc/cmdline.d/90lvm.conf rd.lvm.lv=vg0/swap rd.lvm.lv=vg0/root #cat /etc/cmdline.d/90kernel-modules.conf rd.driver.pre=ata_piix rd.driver.pre=ahci
Was this just a basic system without any encryption etc? Just LVM? Was it still chainloaded? Obviously, the volume groups should be activated. I presume just typing: "lvm vgchange -ay" in the emergency shell properly activates vg0?
Everything is like before, with encryption. Only this time the installation is alone on the machine (which is a physical, but different one, btw.: old centrino laptop w/ single harddisk). What I did do in the graphical installer was try to get a fairly minimal packet selection. I deselected everything (after choosing the "truly minimal" variant), then re-selected things like drakconf and lxde. I accepted all resulting dependencies (in which I had no choice anyway afair). #lvm vgchange -ay File descriptor 9 (/.console.lock) leaked on lvm invocation. Parent PID 688: sh No volume groups found.
Version: Cauldron => 2
Keywords: NEEDINFO => (none)CC: (none) => marja11Version: 2 => CauldronWhiteboard: (none) => (MGA2)
I actually suspect that this could be due to the fact that the initrd is generated (unintentionally) too early during the install of a given package (mageia-theme-Default) and as such not all packages are available to include in the initrd. This is the same issue that cause btrfs partitions not to be mounted. I need to fix mageia-theme to deal with this. i.e. possibly related to #6692. If you're still interested in debugging this issue, perhaps you could try the same fix/work around as mentioned there during installation?
(In reply to comment #9) > [...] i.e. possibly related to #6692. > If you're still interested in debugging this issue, perhaps you could try the > same fix/work around as mentioned there during installation? I did; no luck. I tried again installing from the dual-arch CD w/out network connection, this time with a standard package selection (just w/out Office and Multimedia). I tried both the simple rm of #6692 and the explicit initrd rebuild. For the latter case, this is the shorthand of the output (included modules and skipped rules): dash, i18n, rpmversion, plymouth, dm skip: device-mapper kernel-modules, lvm skip: device-mapper resume, rootfs-block, terminfo, udev-rules skip: udev, late, firmware usrmount, base, fs-lib, shutdown skip: kexec wrote initrd In both cases, I got the same results as before, including those of comment #8 and comment #6.
It works with 3.3.8-desktop-2.mga2, which I got with a network installation. Nice!
(In reply to comment #11) > It works with 3.3.8-desktop-2.mga2, which I got with a network installation. > Nice! Thanks for the feedback. :-D Would you have time to test Mageia 3 alpha 1 or do a cauldron network install? I suppose it is OK there, too now, but I'd rather be sure before closing this bug.
Well, that's complicated. I tried both, cauldron netinstall and 3alpha1 dual CD, in VMs. I run into bug 7412, so LVM is out. Test 1: boot-nonfree.iso: / on LUKS works. Test 2: Partitioned a VM's hard disk with Mageia 1 Live CD with / (ext4) on LVM on LUKS (rebooted after partitioning), then tried to install Mageia-3-alpha1-dual-CD.iso, using the existing partitions (custom partitioning, selecting mount points, setting noatime, no formatting, no swap). Works as expected, until after clicking "done" to finish the paritioning step, I get INTERNAL ERROR: unknown device vg0/lvroot. (See screenshots.) One curious detail: I had mga1 set up swap in the LVM too, but in the 3alpha1 installer, it showed as ext4, so I had the installer ignore it). I have saved snapshots; if you need anything from these systems, ask. However, I only have access occasionally. Usually I have only a machine to test on that's rather limited for various reasons (e.g. pretty much no VM), but I'll see what I can do if you let me know what you need (netinstall is most convenient).
Created attachment 2847 [details] Screenshot of INTERNAL ERROR: unknown device vg0/lvroot
Created attachment 2848 [details] For completeness' sake: screenshot of tty4 Seems irrelevant, but who knows. Taken immediately after the above screenshot.
(In reply to comment #13) > Well, that's complicated. I tried both, cauldron netinstall and 3alpha1 dual > CD, in VMs. I run into bug 7412, so LVM is out. > > > Test 1: boot-nonfree.iso: / on LUKS works. > > > Test 2: Partitioned a VM's hard disk with Mageia 1 Live CD with / (ext4) on LVM > on LUKS (rebooted after partitioning), then tried to install > Mageia-3-alpha1-dual-CD.iso, using the existing partitions (custom > partitioning, selecting mount points, setting noatime, no formatting, no swap). > Works as expected, until after clicking "done" to finish the paritioning step, > I get INTERNAL ERROR: unknown device vg0/lvroot. (See screenshots.) > One curious detail: I had mga1 set up swap in the LVM too, but in the > 3alpha1 installer, it showed as ext4, so I had the installer ignore it). > > > I have saved snapshots; if you need anything from these systems, ask. However, > I only have access occasionally. Usually I have only a machine to test on > that's rather limited for various reasons (e.g. pretty much no VM), but I'll > see what I can do if you let me know what you need (netinstall is most > convenient). Thanks a lot for all the testing, Joe! I don't have time to clone this report now (the new bugs you found should go into new reports). Can you attach /root/drakx/report.bug.gz too, please? @ pterjan Allready cc'ing you, because the new issues are both yours, IINM (the swap/ext4 one and the unknown device vg0/lvroot one) Oh, and bug 7412, too.
CC: (none) => pterjan
Always happy to make more work for ya'll. >I don't have time to clone this report now (the new bugs you found should go into new reports). Yeah, you're right. I'll have time to do that soon. >Can you attach /root/drakx/report.bug.gz too, please? I restored the snapshot and this file doesn't exist. non-chrooted /root is pretty much empty, and the installer seems not to have chrooted yet. /mnt is empty, cat /proc/mounts shows nothing of the hard disk. I managed to mount the new root, but I needed to run the partitioning step in the installer again with the same result as earlier. Please forgive me for making this a comic book for now (see screenshots); I'll clean this up and move it to appropriate bug reports.
Created attachment 2851 [details] cryptsetup status All screenshots today from same time as above ones. Being verbose because I'll have little access to this system; better safe than sorry.
Created attachment 2852 [details] lvm2 lvdisplay
Created attachment 2853 [details] fdisk -l
Created attachment 2854 [details] lsmod
Created attachment 2855 [details] vgchange and mount
Summary: Installed into encrypted LLVM from dual-arch; 1st reboot fails. (dracut Warning: Cancelling resume operation. Device not found. Unable to process initqueue) => Installed into encrypted LVM from dual-arch; 1st reboot fails. (dracut Warning: Cancelling resume operation. Device not found. Unable to process initqueue)
Created attachment 2867 [details] Empty dummy attachment to obsolete all earlier ones I cloned the two new bugs which didn't belong here: bug 7578 (swap/ext4) and bug 7603 (unknown device). All those screenshots pertained to the two new bugs.
Attachment 2847 is obsolete: 0 => 1 Attachment 2848 is obsolete: 0 => 1 Attachment 2851 is obsolete: 0 => 1 Attachment 2852 is obsolete: 0 => 1 Attachment 2853 is obsolete: 0 => 1 Attachment 2854 is obsolete: 0 => 1 Attachment 2855 is obsolete: 0 => 1
Depends on: (none) => 7382
a missing package was added in the dual for the alpha2 (bug 7382), so something is a little better (at least after the install)
(In reply to comment #24) Thanks for the heads-up. I've actually been following that, but since Dave was talking about various builds of alpha 2 and I could only find alpha 1 here, ftp://distrib-coffee.ipsl.jussieu.fr/pub/linux/Mageia/iso/cauldron/ , I assumed those were not builds accessible to the public. Please advise if I'm wrong.
oups, yes indeed, sorry it's still in QA, as I am little off I don't know if there is still 'blocker/annoying bugs' for make it public
I currently can't test any isos, sorry. Using the net installer, I run into the same LVM-related bugs with alpha2 as with alpha1.
Hi, I think I just hit the same bug when trying to install Mageia 2. The setup I attempted was /boot on xfs for primary partition 1 Then LVM for the rest of the disk. And put my /, swap, /home, etc on the LVM group. Unable to boot, same cannot resume wanring etc. I'm currently downloading mageia 3 alpha 2 to attempt the same install. This was with the 64bit dvd image dd pushing it onto a usb stick. Downloading 64bit mageia 3 alpha 2. Will report back afterwards. Cheers, Alan
CC: (none) => skyphyr
Could not reproduce with 3alpha3 dual medium (no online updates during installation). Testing of cauldron netinstall is blocked by Bug 8160.
thanks, closing then
Status: NEW => RESOLVEDResolution: (none) => FIXED