Occasionally my Ubuntu 10.04 PC won't boot properly. It gets past Grub and then stops at a blank screen and blinking cursor. From what I've read, this blinking cursor screen is presented by Ubuntu itself and not Grub, so I assume the boot process gets halted for some reason. Has anyone any guidance on how to diagnose this issue or what the cause is likely to be? Normally I need to press the reset button to reboot the PC and often it will reboot fine. The fact that it is intermittent is what confuses me.
Any pointers on diagnosing the problem would be much appreciated.
It's been a while, mainly because my server has been up for a long time. It looks like I've captured a recurrence of this issue, I copied the messages
file and the dmesg
file and had a look where processing seems to have stopped and found the messages below. I'm going to do some research on Google etc. but figured I'd put it up here in case anyone can help and wants to earn themselves some points. I should mention that the ondemand governor failed
message happens on successful boots but the other two don't appear to.
Oct 11 23:17:21 linux kernel: [ 98.905370] ondemand governor failed, too long transition latency of HW, fallback to performance governor
Oct 11 23:21:48 linux kernel: Kernel logging (proc) stopped.
Oct 11 23:21:48 linux rsyslogd: [origin software="rsyslogd" swVersion="4.2.0" x-pid="697" x-info="http://www.rsyslog.com"] exiting on signal 15.
I found a few vague references to rolling over of logfiles at boot time being the cause.
Hold shift during boot, then hit E to edit the GRUB entry. Remove the part that says
quiet splash
and replace it withtext
to see what's happening during boot.I encountered this issue and it turned out the problem was my hard drive being 100% full. Steps I took to troubleshoot and finally fix this were as follows:
Boot to blinking cursor
Press Ctrl+Alt+Fx to enter Ubuntu's tty virtual console screen, where x is anything between 2 and 6. In my case I did Ctrl+Alt+F2
Log in using your username and password.
Type in
df
to check the storage and if indeed there is no more free spaceIf storage is the issue, clear some space by deleting unnecessary files.
To ensure kernel/grub settings are also not an issue, edit them settings by going to the grub settings file:
Edit the
GRUB_CMDLINE_LINUX_DEFAULT
line in the grub settings file to sayUpdate grub settings
sudo update-grub
Reboot by typing in
reboot
Boot to blinking cursor and wait, it should now bring you to the login page! :)
In my case, the blinking cursor was all I would ever get. No boot. It was upon installing a fresh Ubuntu Minimal. I figured out that during the GRUB installation step, it was installing GRUB onto the wrong drive, the "first" drive (/dev/sda).
My system has 3 drives. Two 500GB drives in RAID, that I didn't want to touch during installation, and a 120GB SSD that I use for the OS. For whatever reason, the "first drive" (/dev/sda) is one of my 500GB drives. /dev/sdb is my 120GB drive and /dev/sdc is the other 500GB drive.
So, when formatting with a partition table of "mbr" on my 120GB drive, I did the normal 117GB of bootable ext4 and 3GB of swap. On the GRUB installation step, DO NOT choose Yes to put GRUB onto the "first" drive. Choose NO. This will bring up another screen that allows you to input /dev/sdX. In my case, I tried /dev/sdb and /dev/sdb1, but the installer would give me a fatal error every time, which still makes no sense.
Finally, I had to format my 120GB drive with a partition table of "gpt". With GPT, you have to manually create a GRUB partition. That's the way things are done with GPT. So, the first partition I made for GRUB was 32.0 MB formatted for "boot or something (forget wording)". Second partition was my 3.0 GB formatted for "swap", at the "end". Third partition was the remaining space formatted as "ext4".
Now, when choosing NO during the GRUB installation step, manually input /dev/sdb, not /dev/sdb1 surprisingly, and it then works. GRUB installs into the 32MB boot partition on the correct drive and the system boots normally. YAY!
BTW, you have to choose Expert install from the menu at the beginning of the installation to do all this and format your HDD "manually" not "guided". Guided will always choose /dev/sda as the first drive and blinking cursor/no boot will result if /dev/sda isn't your OS drive.
I have had this problem in the past, and found that it appears to happen on some kernels and not others although I have not had this issue since upgrading to Meerkat. But often times I found I would have to select a prior Kernel to load into Ubuntu properly.
Another thing I'd check is that your hard drive is healthy. Check in System > Administration > Disk Utilities, look at the SMART status, it should be Disk is healthy, otherwise your drive might be failing.
I've had that problem quite a few times now, but I could distinguish at least three different variants:
This has led me to believe that when you see what you're describing, the actual problem is that the
quiet
boot option is hiding something from you. For instance, I could track down one of my incidents to the system recovering (not the usual routine check) my harddisk. I have since removed thequiet
option from my grub entries.I had this problem with a new install of 11.10 server.
I was able to switch to a VT with alt-F1, so the machine was alive but had switched to vt7, despite no X being enabled.
I fixed it by altering
GRUB_CMDLINE_LINUX_DEFAULT
options fromquiet splash
tonomodeset
which meant I got the proper boot information, followed by VT1.Verify your disk drive is seated correctly (especially if you are fooling around with your hardware). We just solved this problem because my admin came into my lab and popped out the disk drives and popped them back in and everything booted up nicely! I was swapping the drives between machines and being way too delicate when re-seating them, so they weren't fully connected to the machine, hence no Operating System to load.
In my case for the above description (sometimes getting a black screen with a blinking cursor), lightgdm having a race condition and not being able to start properly was the issue; see my full answer to this other related question.
See the details of the solution here: http://www.webupd8.org/2013/01/ubuntu-lightdm-black-screen-when-using.html (see also this bug report).
The gist of it: Use gdm and not lightgdm (i.e.
sudo apt-get install gdm
, and choosegdm
as default login manager when asked).Hope this helps someone.
I had a similar issue in the past using the recommended version of a proprietary nVidia driver with a certain video card. The solution was to boot into recovery mode, run the xfix option, then boot into the desktop. Once in the desktop, I would go into the hardware drivers screen and select an older version of the driver.