[SOLVED] Archiso and mkinitcpio's timestamp hook

Strike · 2013-09-21 02:20:23

Yesterday, I ran:

# pacman -Syyu

I then proceeded to create a live media using Archiso in the usual fashion which promptly failed with the same error described here:

https://bbs.archlinux.org/viewtopic.php … 7#p1326827

So the fix was simple; I simply removed the timestamp hook from mkinitcpio.conf with:

sed -i 's/timestamp //' "$HOME/archlive/releng/mkinitcpio.conf"

This fixed the Archiso problem and I successfully built my installation media in the usual way.

However upon booting this media, I receive the following message:

Loading boot/x86_64/vmlinuz......
Loading boot/x86_64/archiso.img............ready.
Probing EDD (edd=off to tisable)... ok
early console in decompress_kernel

Decompressing Linux... Parsing ELF... done.
Booting the kernel.
:: running early hook [udev]
:: runnning hook [udev]
:: Triggering uevents...
:: running hook [memdisk]
:: running hook [archiso]
:: running hook [archiso_loop_mnt]
:: running hook [archiso_pxe_common]
:: running hook [archiso_pxe_nbd]
:: running hook [archiso_pxe_http]
:: running hook [archiso_pxe_nfs]
:: Mounting '/dev/disk/by-label/ARCH_201389' to '/run/archiso/bootmnt'
Waiting 30 seconds for device /dev/disk/by-label/ARCH_201309...
.
.
.
.
.
.
:: Device '/dev/disk/by-label/ARCH_201389' mounted successfully.
:: Mounting /run/archiso/cowspace (tmpfs) filesystem, size=75%...
losetup: /run/archiso/bootmnt/arch/x86_64/root-image.fs.sfs: failed to set up loop device: No such file or directory
:: Mounting '' to '/run/archiso/sfs/root-image'
Waiting 30 seconds for device...
.
.
ERROR: '' device did not show up after 30 seconds...
    Falling back to interactive prompt
    You can try to fix the problem manually, log out when you are finished
sh: can't access tty; job control turned off

What's going on here? Any ideas?

Last edited by Strike (2014-07-17 09:55:14)

jghodd · 2013-09-21 05:47:33

I'm having the same issue. If I just remove timestamp, my distro builds, but it won't boot - same error messaging you're seeing. If I replace udev and timestamp with systemd, it also builds but I get an infinite scrolling of "Welcome to emergency mode!" that moves so fast I can't see the actual error and never get to a prompt.

If I figure anything out, I'll let you know, otherwise... same question - anybody have any ideas?

qinohe · 2013-09-21 10:24:59

Hi Strike, and welcome to Arch, FYI, today I installed the archiso-git again, and I will stick to it, although the output stays the same, waiting 30 seconds..

Maybe later this afternoon, I will try some things, although I don't know what yet...

Although, to bad I can't build, I'm glad I'm not alone with this one.

falconindy · 2013-09-21 10:32:24

Probably fixed with systemd-207-5 (which is on its way to [core] currently).

qinohe · 2013-09-21 10:34:40

falconindy wrote:

Probably fixed with systemd-207-5 (which is on its way to [core] currently).

Oh, you're my hero

jghodd · 2013-09-21 16:56:54

That would be awesome, falconindy. Trying out some really heavy-handed workarounds at the moment by downgrading systemd and initcpio, but I'd much rather have a working solution. ;-)

jghodd · 2013-09-21 17:05:23

Are you sure you're talking about systemd-207-5? I'm showing I already upgraded to 207-5 on Sept. 19...

-rw-r--r-- 1 root root 2265752 Sep 19 18:20 /var/cache/pacman/pkg/systemd-207-5-i686.pkg.tar.xz

Yeah... had to wait for my build to get to point where pkglist was created to make sure it was actually using 207-5. And it is. So, I can say with a great degree of certainty that 207-5 doesn't fix the problem.

Back to square one.

If my heavy-handed downgrade works, I'll let you know.

Last edited by jghodd (2013-09-21 17:39:19)

falconindy · 2013-09-21 20:39:09

jghodd wrote:

Are you sure you're talking about systemd-207-5? I'm showing I already upgraded to 207-5 on Sept. 19...
-rw-r--r-- 1 root root 2265752 Sep 19 18:20 /var/cache/pacman/pkg/systemd-207-5-i686.pkg.tar.xz
Yeah... had to wait for my build to get to point where pkglist was created to make sure it was actually using 207-5. And it is. So, I can say with a great degree of certainty that 207-5 doesn't fix the problem.
Back to square one.
If my heavy-handed downgrade works, I'll let you know.

None of what you're saying here really makes sense. If 207-5 doesn't fix the boot error, then it's something unrelated to the OP.

Alternatively, if you upgraded systemd and didn't rebuild your initramfs, then you didn't really upgrade.

Last edited by falconindy (2013-09-21 20:42:04)

jghodd · 2013-09-21 21:27:27

Alternatively, if you upgraded systemd and didn't rebuild your initramfs, then you didn't really upgrade.

systemd was updated to 207-5 during an archiso build, which also rebuilds the initramfs, so yes, it was a full upgrade.

All of my distro builds have worked just fine for the past 9 months, up until the systemd-207 release, and I have about a 3 week turnaround on 12 separate builds - that's at least 144 separate builds/releases, and I know the number is much closer to 200 with interim and test builds included.

Since the systemd-207 release, I've tried building with just timestamp removed, and I've tried replacing udev and timestamp with systemd. Neither of these cases works. At the moment I'm dead in the water. With just timestamp removed, the boot fails the same way that Strike's fails (see 1st post). If I replace udev and timestamp with systemd, I get an endless looping of "Welcome to emergency mode!" along with the journalctl message - it never dumps me to a prompt and I have to hard power-off.

My distro is meant to be booted from a thumb drive, which is how I test it. Is there anything I need to know that changed in terms of timing, etc.?

At this point, I've removed the cached systemd update package to force a re-download, and am currently building with udev and timestamp replaced with systemd. I'm not holding my breath since I noted from the AUR package meta data that systemd-207-5 was in fact ready on the 19th, so I'm pretty sure nothing will have changed. I'll see if I can somehow freeze the screen early enough after the boot to catch the error message, but that's not going to be easy and may prove to be impossible. My last ditch effort will be to retry a build with just timestamp removed.

I'm open to any advice you might have to give...

jghodd · 2013-09-21 21:31:26

None of what you're saying here really makes sense. If 207-5 doesn't fix the boot error, then it's something unrelated to the OP.

Well, the other possibility is that 207-5 has a bug, or that something that's required has been "fixed" out of existence...

Last edited by jghodd (2013-09-21 21:32:19)

jghodd · 2013-09-21 21:49:55

Just thought I should add that I don;t see this issue with installations, just the archiso boot image when loaded onto a thumb drive. My arch disk installations work just fine with just udev and no timestamp. I don;t know exactly what timestamp used to do, but apparently that functionality is now missing vis-a-vis usb thumb drives. It might have been a wait that's no longer being done, or a change in timing...? I don;t know.

Also wanted to add that this build is simply a versioning upgrade - in other words, there are no file or directory changes between this one and the last working version I released on 09.03, other than dealing with timestamp in the mkinitcpio.conf file at the root of the releng directory used for the iso boot (and the package updates).

Perhaps this makes more sense...? Any ideas as to what I might try?

Last edited by jghodd (2013-09-21 22:18:17)

jasonwryan · 2013-09-21 21:51:17

Stop bumping this thread: there is an edit button, please learn how to use it.

Strike · 2013-09-22 02:14:10

Thanks falconindy; it's playing nicely again. Cheers!

Please mark as solved.

Last edited by Strike (2013-09-22 02:15:36)

jghodd · 2013-09-22 04:29:05

Well, I have no idea why forcing a new download of 207-5 should've worked, but the udev + no-timestamp scenario is now booting without any issues.

The systemd replacement of udev + timestamp is still not working though - it starts booting just fine, gets to "Reached target Initrd Default Target" without any failures, then loops with the "Welcome to emergency mode!" message until I power down or enter ctl-alt-del. It never does get to a prompt.

In any case, I'm happy with the udev + no-timestamp scenario, so my issue appears to be solved too.

Thanks for the help, falconindy.

cookies · 2013-09-23 00:49:56

Strike wrote:

Please mark as solved.

You opened the thread, you have the power to mark it as solved (if that's the case for you). https://bbs.archlinux.org/viewtopic.php?id=130309 tells you how to do it.

ewaller · 2013-09-23 02:08:47

... And we ask the the original poster do that, because only they know when their thread is solved

Arch Linux

#1 2013-09-21 02:20:23

[SOLVED] Archiso and mkinitcpio's timestamp hook

#2 2013-09-21 05:47:33

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#3 2013-09-21 10:24:59

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#4 2013-09-21 10:32:24

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#5 2013-09-21 10:34:40

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#6 2013-09-21 16:56:54

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#7 2013-09-21 17:05:23

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#8 2013-09-21 20:39:09

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#9 2013-09-21 21:27:27

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#10 2013-09-21 21:31:26

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#11 2013-09-21 21:49:55

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#12 2013-09-21 21:51:17

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#13 2013-09-22 02:14:10

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#14 2013-09-22 04:29:05

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#15 2013-09-23 00:49:56

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

#16 2013-09-23 02:08:47

Re: [SOLVED] Archiso and mkinitcpio's timestamp hook

Board footer