[SOLVED] GRUB2: Timed out waiting for device (was anciently a swap partition)

JZL240I-U · 04-03-2014, 01:54 AM

This is a little complicated, so please bear with me.

On my netbook I boot 6 Linuxes and FreeBSD (chainloaded) with GRUB2 from a dedicated /boot partition. GRUB on the MBR comes from openSuSE 12.3. This is my layout as seen by "blkid":

Code:

/dev/sda1: LABEL="boot" UUID="e8b9e1af-e836-4332-8446-d149a0c68d40" TYPE="ext4" 
/dev/sda2: LABEL="swap" UUID="2184cf1b-154a-4ebf-bda8-f1ca3a53de3b" TYPE="swap" 
/dev/sda3: UUID="533130b21005f3f5" TYPE="ufs" 
/dev/sda5: LABEL="suse-12.3" UUID="a13e1c0f-c157-4ee5-a1c3-67a39ebe01d1" TYPE="ext4" 
/dev/sda6: LABEL="fedora" UUID="f583b5ff-7a87-44de-bc3b-d1e5f5041b9f" TYPE="ext4" 
/dev/sda7: LABEL="ubuntu" UUID="2b9b3bd5-1c29-4035-b86a-1258616a7094" TYPE="ext4" 
/dev/sda8: LABEL="mint" UUID="a6bc25ed-94ca-40ba-929a-ff44fe076113" TYPE="ext4" 
/dev/sda9: LABEL="chakra" UUID="3231a646-16b7-447c-b540-be9a0f2f94ea" TYPE="ext4" 
/dev/sda10: LABEL="suse-13.1" UUID="9e553e3d-3c7f-48cd-ae7b-a629ff340ede" TYPE="ext4" 
/dev/sda11: LABEL="home" UUID="4091f3e8-1191-47a2-8c16-0ecbf3c8b06c" TYPE="ext4" 
/dev/sda12: LABEL="backup" UUID="e4b40d27-d78d-48bc-8e9f-2f2afda2eb62" TYPE="ext4"

Last install was fedora 20 "Heisenbug" which sort of bricked the netbook (error: symbol 'grub_dl_get' not found). Fedora also formatted the swap partition, which changes its UUID. Anyway, I chrooted into openSuSE 12.3 and re-installed GRUB2.

During boot (only of openSuSE12.3) I get this output on the screen and in boot.log (excerpt):

Code:

Trying manual resume from /dev/disk/by-id/ata-WDC_WD3200BPVT-22JJ5T0_WD-WXU1CB1S5967-part3
Invoking userspace resume from /dev/disk/by-id/ata-WDC_WD3200BPVT-22JJ5T0_WD-WXU1CB1S5967-part3
resume: libgcrypt version: 1.5.3
Trying manual resume from /dev/disk/by-id/ata-WDC_WD3200BPVT-22JJ5T0_WD-WXU1CB1S5967-part3
Invoking in-kernel resume from /dev/disk/by-id/ata-WDC_WD3200BPVT-22JJ5T0_WD-WXU1CB1S5967-part3
Waiting for device /dev/root to appear:  ok
...
[ OK  ] Reached target Encrypted Volumes.
[ OK  ] Listening on udev Kernel Socket.
[ OK  ] Listening on udev Control Socket.
         Expecting device dev-disk-by\x2duuid-c8250322\x2d34af\x2d4d18\x2db58d\x2dd2f6b5e4887e.device...
         Expecting device dev-disk-by\x2duuid-e8b9e1af\x2de836\x2d4332\x2d8446\x2dd149a0c68d40.device...
         Expecting device dev-disk-by\x2duuid-4091f3e8\x2d1191\x2d47a2\x2d8c16\x2d0ecbf3c8b06c.device...
         Expecting device dev-disk-by\x2duuid-e4b40d27\x2dd78d\x2d48bc\x2d8e9f\x2d2f2afda2eb62.device...
[ OK  ] Listening on Journal Socket.
         Starting LSB: Set default boot entry if called...
...
[  OK  [0m] Started LSB: AppArmor initialization.
[ TIME ] Timed out waiting for device dev-disk-by\x2duuid-c8250322\x2d34af\x2d4d18\x2db58d\x2dd2f6b5e4887e.device.
[DEPEND] Dependency failed for /dev/disk/by-uuid/c8250322-34af-4d18-b58d-d2f6b5e4887e.
...
[ OK  ] Started LSB: Console mouse support.
        Expecting device dev-disk-by\x2duuid-c8250322\x2d34af\x2d4d18\x2db58d\x2dd2f6b5e4887e.device...
         Starting Authorization Manager...
[ OK  ] Started LSB: X Display Manager.
...

And this is in /var/log/messages (excerpt):

Code:

...
2014-04-03T07:54:13.103372+02:00 netbook systemd[1]: Job dev-disk-by\x2duuid-c8250322\x2d34af\x2d4d18\x2db58d\x2dd2f6b5e4887e.device/start timed out.
2014-04-03T07:54:15.097339+02:00 netbook kernel: [  199.177151] fuse init (API version 7.20)
2014-04-03T07:54:15.226326+02:00 netbook systemd[1]: Timed out waiting for device dev-disk-by\x2duuid-c8250322\x2d34af\x2d4d18\x2db58d\x2dd2f6b5e4887e.device.
2014-04-03T07:54:15.226396+02:00 netbook systemd[1]: Dependency failed for /dev/disk/by-uuid/c8250322-34af-4d18-b58d-d2f6b5e4887e.
2014-04-03T07:54:15.228064+02:00 netbook systemd[1]: Job dev-disk-by\x2duuid-c8250322\x2d34af\x2d4d18\x2db58d\x2dd2f6b5e4887e.swap/start failed with result 'dependency'.
...

I found in the file configuring GRUB2 for chakra linux that the expected device was /swap before fedora came and did its foul deeds.

Now my questions: How can I find out which program wants the old swap? How can I configure it so that it uses the new UUID?

Phew. Sorry for this contorted description.

bigrigdriver · 04-03-2014, 04:13 PM

My guess, look at /etc/fstab for any distro you boot into which gives the error messages.

You should see an entry like this:

Code:

# swap was on /dev/sda2 during installation
UUID=24f30eef-6968-494e-b1ff-6f7b0c6623f5 none     swap    sw         0       0

You said that Fedora reformated swap. What did Fedora format it to?

If you don't mind reinstalling Fedora, I suggest that you reformat /dev/sda2 to swap, then reinstall Fedora. Take special care when selecting partition(s) to install Fedora into so that the problem does not recur. Fedora should see the swap and use it as swap during installation.

syg00 · 04-03-2014, 06:11 PM

I would expect you'll find a "resume=" keyword in the opensuse grub - delete it or change it to noresume.
For multiple installs on the same system, I've found UUID a PITA - I never use it for swap. The installers typically go looking for a pre-existing swap to use - and sometimes use that in the final build as well. After a mkswap as you found.

JZL240I-U · 04-04-2014, 01:51 AM

Quote:

Originally Posted by bigrigdriver

...You should see an entry like this:

Code:

# swap was on /dev/sda2 during installation
UUID=24f30eef-6968-494e-b1ff-6f7b0c6623f5 none     swap    sw         0       0

I see an entry like this:

Code:

UUID=24f30eef-6968-494e-b1ff-6f7b0c6623f5 swap swap    defaults  0       0

Quote:

Originally Posted by bigrigdriver

You said that Fedora reformated swap. What did Fedora format it to?

To swap and quite unnecessarily so

. That anaconda installer is an unspeakable abomination. Aside from the fact, that it was literally impossible to install f17 through f19 on my given configuration it refuses to accept the existing swap as such without re-formatting it.

Quote:

Originally Posted by bigrigdriver

If you don't mind reinstalling Fedora, I suggest that you reformat /dev/sda2 to swap, then reinstall Fedora.

This will be my last installation of fedora. That anaconda is simply unacceptable and since the maintainers chose to ignore the obvious I'll just look for an other distro with Gnome 3.

Quote:

Originally Posted by bigrigdriver

Take special care when selecting partition(s) to install Fedora into so that the problem does not recur. Fedora should see the swap and use it as swap during installation.

It doesn't see swap (as I wrote above) and I do take care, believe me, since I do not want to destroy my setup. Ah, well, I just want to understand where openSuSE got its problem from, fedora can go and rot. So this is not solved yet...

JZL240I-U · 04-04-2014, 02:02 AM

Quote:

Originally Posted by syg00

I would expect you'll find a "resume=" keyword in the opensuse grub - delete it or change it to noresume.

I do but it points to the actual swap with the correct "/dev/disk/by-id/ata-WDC_WD...67-part3" not UUID. If you look at my first post in this thread you'll note the first few lines in the second "code" box. They show that the resume option worked as it is supposet to do.

And it is not GRUB2 anymore. The error message comes from systemd, i.e. the linux kernel is already active and running.

Quote:

Originally Posted by syg00

For multiple installs on the same system, I've found UUID a PITA - I never use it for swap.

I tend to agree, though I recognize the reason it's done for. The disk by-id scheme seems somewhat more suitable here.

Quote:

Originally Posted by syg00

The installers typically go looking for a pre-existing swap to use - and sometimes use that in the final build as well. After a mkswap as you found.

Yeah, well, see my answer to bigrigdriver's post. Anaconda is {censored}.

syg00 · 04-04-2014, 02:09 AM

Quote:

Originally Posted by JZL240I-U

And it is not GRUB2 anymore. The error message comes from systemd,

... missed that small detail ...

JZL240I-U · 04-04-2014, 02:11 AM

Quote:

Originally Posted by syg00

... missed that small detail ...

Happens

. But do you have an idea what's causing it?

syg00 · 04-04-2014, 02:43 AM

suspend.target ?. The swap entry(s) in fstab get used to generate swap unit(s) at early start - might be something there too.
Just a guess as I haven't looked at opensuse in years - and with systemd we're probably all stumbling around in the dark somewhat.

JZL240I-U · 04-04-2014, 02:46 AM

But how are "suspend.target" or the swap units parameterized? Would there be any .config files?

JZL240I-U · 04-07-2014, 07:23 AM

Hmmm. No other takes then?

I could also change the UUID (this is on ext4) like so:

Code:

tune2fs /dev/{device} -U {uuid}

and repair the other GRUB2 stanza(s). But I still would like to know the how and why of the currently active mechanism.

JZL240I-U · 04-08-2014, 09:05 AM

Gaah. How stupid can one be? "Fedora also formatted the swap partition, which changes its UUID." Yes. But it does not edit openSuSE's /etc/fstab which still expects the old /swap. *sigh*. Solved.

Remark: This is one of the cases, where the UUID has its definite drawbacks. This would not have happened with "/dev/sda3".

syg00 · 04-08-2014, 08:26 PM

Happens

ciyo.ma · 05-22-2015, 10:14 AM

I encountered the same problem but in openSUSE 13.2.
The reason is that the dracut turn on --hostonly option by default. You can comment out this option in the files under /etc/dracut.d/
Then rebuild the initramfs by dracut -f.
Issue solved.

JZL240I-U · 05-26-2015, 07:51 AM

You lost me there. "dracut" sounds fedora-ish and I jettisoned fedora anyhow. And in my case it was simply the obsolete entry in openSUSE's /etc/fstab, no need to rebuild initramfs.