LiveCD: boot progress feedback (is it booting or is it hung up?)

[Chip Piller]

I tried the dapper-live-i386.iso version from March 7, 2006. While booting the horizontal progress bar display takes about 80% of the boot time to go about 20% of the way across the screen.
The system appears to get hung up on the item "mounting root filesystem", it takes a long time with absolutely no visual feedback that things are OK.
The boot process needs more visual feedback, something to show the user that the boot process is proceeding, that all is well, and that the system is not hung up.

[Martin Bergner (martin-bergner)]

I can confirm this for dapper flight 6.
The issue here is, that it drops out of usplash during "adding user", it then creates a locale, and after that it seems that it hangs. Maybe a message like: "This may take some time" would help the issue.

[Tollef Fog Heen (tfheen)]

The "mounting root file system" takes less than ten seconds here for me and it then moves on to "moving mount points."

As for the "dropping out of usplash", this was fixed a little bit ago and is presumably another bug.

[Chip Piller (piller)]

I downloaded the latest dapper-live (april 24) and tested it out on two computers and observed that a significant delay still occurs at the "mounting root file system" step.

Booting the liveCD on a P4 1.5 GHz machine resulted in a 45 second delay when it got to "mounting root file system". This is an eternity for someone booting a liveCD, there is absolutely no visual feedback during this time to indicate to the user that the system is not hung up.

I next tried the liveCD on an AMD64 3200+ machine, it still took a good 30 seconds at the "mounting root file system" step. This also is way too long.

I noticed some thing about hdc errors in dmesg output on both machines.

[Chip Piller (piller)]

dmesg output april 24 version of dapper-live on amd64 3200+ machine which shows boot time delay and some drive error messages.

[Chip Piller (piller)]

I tested the april 24 dapper-live on a P2-400 MHz machine. Total boot time from initial Ubuntu splash until the date/time appears on the desktop is 5 minutes and 32 seconds, including the 30 second default boot countdown.

This machine took about 34 seconds during the "mounting root file system" step, which is less time than on a P4-1.5GHz machine I tested.

When it got to the "configuring X" step the elapsed time was 2:53 and the horizontal progress bar had moved less than 10% of the way across the screen.

When one first boots the liveCD there is a quick text screen and then one gets the ubuntu splash screen. Then when the 30 second timeout elapses we see a couple of lines of text overwriting the top of the screen, loading vmlinuz and such. I don't think we should be seeing this.

Then we also see a graphical message box that overwrites the initial ubuntu splash screen as well, it looks like a "windows" message box. Anyway, I don't think this looks very good either and should be done differently.

dmesg output also shows drive error messages.

[Tollef Fog Heen (tfheen)]

* Chip Piller

| I downloaded the latest dapper-live (april 24) and tested it out on
| two computers and observed that a significant delay still occurs at
| the "mounting root file system" step.

Can you verify that the next step is "moving mount points»? If so,
there's no reason why that should take thirty seconds, unless you have
a really, really slow CD drive.

--
Tollef Fog Heen ,''`.
UNIX is user friendly, it's just picky about who its friends are : :' :
                                                                      `. `'
                                                                        `-

[Chip Piller (piller)]

Hi Tollef,
Yes, I have just verified this on an AMD64 3200+ machine and a P4 1.5 GHz machine. First I checked the iso md5sum and then burned a new CDR.

I inserted the CD, and restarted the PC. I did not press any keys, I just let the liveCD boot completely on it's own.

I recorded 4 times: when I first saw the initial ubuntu screen, when I saw mounting root file system, when I saw moving mount points, and when the time/date appeared in the top right corner of the desktop.

Time from mounting root file system to moving mount points: 59 seconds on AMD64 machine, 70 seconds on P4 machine.

Overall boot time including 30 second default timeout: 4:02 on AMD64, 3:48 on P4.

Lot's of kernel boot error messages (HDC errors mostly) listed in dmesg output, I think this is the real problem.

[Chip Piller (piller)]

Checked the dapper-live bootup again on my trusty old P2 400MHz machine, the time from mounting root file system to moving mount points is 36 seconds, which is slow but still a lot faster than the exact same liveCD on the P4 (70 s) and AMD64 (59 s) machines.

Total boot time from initial ubuntu screen to seeing the date/time on the desktop was 4:06, which is just slightly slower than the exact same liveCD on the AMD64 3200+ and P4 1.5GHz machines. So there is definitely a problem.

[Matt Zimmerman (mdz)]

It is usually fast on most systems I have tried, but on my amd64 test system I just saw this behaviour today on my amd64 test system.

By pressing F6 at the boot prompt and removing 'quiet' and 'splash' from the command line, I was able to see that the cause was I/O errors on the CD, presumably due to CD-RW media problems or potentially a dirty drive (I haven't investigated yet). Both of the attached dmesg outputs show the same problem.

This clearly isn't casper's fault. I suggest that anyone seeing these symptoms do an integrity check on their CD, and if that succeeds, boot without 'quiet' or 'splash' and see what's going on, and check dmesg.

[koftinoff (jeffk)]

I just ran into this bug. Actually, got here following bug 36667 which says that it is a duplicate of this bug, 34172. However I found a solution to the problem on my system. When I rebooted with out quiet and splash, I noticed the log scroll by saying that it recognized my cdrom drive as device hdf. Well, it always used to be hdc.

So the reason why it was hanging on 'waiting for root device' was because it was trying to load /dev/hda5 as my root device but for some reason the linux kernel after the upgrade decided to call my first hard drive "/dev/hde"...

So I modified grub/menu.lst and /etc/fstab and it boots fine...

FYI, before I upgraded to dapper, /proc/ide/ looked like:

drivers
hda -> ide0/hda
hdb -> ide0/hdb
hdc -> ide1/hdc
ide0
ide1

After booting the new kernel, /proc/ide looks like:

cmd64x
drivers
hdc -> ide1/hdc
hdd -> ide1/hdd
hde -> ide2/hde
hdf -> ide2/hdf
hdg -> ide3/hdg
ide1
ide2
ide3

I do have a scsi board and a promise ATA controller plugged in, but since these were in the system before, in my opinion the linux kernel should not have messed with my ide port ordering...

jeff koftinoff
<email address hidden>
www.jdkoftinoff.com

[Adam Conrad (adconrad)]

These were all either kernel bugs or (more likely) flaky CD media or drives. Nothing to do with initramfs-tools.

[Chip Piller (piller)]

This problem still exists on the May 19, 2006 ubuntu dapper-live-i386.iso.

As per Matt Zimmermans comment on May 3rd I did the "Check CD for Defects" thing on my freshly burned CDR and it passed. I also booted without quiet and splash. Checking dmesg shows the same ide/cd errors as before. Booting the breezy-live cd does not produce these errors, so I think this is a kernel bug.

I also noticed that during the liveCD integrity check that the progress bar cycles, it keeps moving and provides a heartbeat/visual feedback even when it gets to a lengthy step such as checking the squashfs filesystem. This progress bar behaviour differs from that which occurs during the liveCD boot process.

[Nando Vieira (fe-address)]

I'm trying to install (clear install) Dapper Final Release.

Here's the error:

hdc: ide_intr: huh? expected null handler on exit
Buffer I/O error on device hdc, logical block 357296

My config:

DVD-RW Samsung (bought 2 days ago)
ASUS A8N-SLI
AMD64 Venice 3700+
1 GB RAM Samsung
Radeon X300 256mb
HD 120GB Samsung
HD 20GB Seagate

[David Fraser (davidf)]

Had a similar error to above using ubuntu 6.0.6 desktop i386 install, with Samsung DVD/RW
Note that in my case the device name is correct and consistent (hdb)
The logical block given for the error is 357298 (then later 357297/6) - but the iso image only has 357297 blocks (as confirmed by the iso header) - and have checked the md5sum of the downloaded image
So it seems like this has to do with trying to read past the end of the disk and the failure been really slow ... after leaving it for about 5-10 minutes it seems to mount the loopback filesystem OK