Installation hangs at 66% "Installing GRUB boot loader"

[Erwan Prioul]

When installing Ubuntu 17.04 (daily image) in ppc64el VM, after selecting the LVM guided installation, the installation hangs at 66% "Installing GRUB boot loader".

The same issue occured with Debian few days ago: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=853187
It appeared the issue came from os-prober: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=853277

The issue was fixed in os-prober 1.74. I've noticed that is os-prober 1.73ubuntu4 in the daily image.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Eric Desrochers (slashd)]

Hi Erwan,

The outcome of the Debian #853277 is the following commit :
--
commit 911b10ffb146f494f96e1584b1630d05a299adf7
Author: Cyril Brulebois <email address hidden>
Date: Wed Feb 1 00:01:14 2017 +0100

    Skip LVM2_member partitions (Closes: #853277).
--

thus not the one you mentioned above. The patch instruct os-prober to skip LVM2_member fstype.

I fixed Xenial release via LP: #1579609, and Zesty will eventually be fix by merging from Debian which includes a different fix[1]

[1] - https://bugs.launchpad.net/ubuntu/+source/os-prober/+bug/1579609/comments/34

Eric

[Petr Menzel (petrmenzel)]

Screenshot:
https://drive.google.com/file/d/0B6vr-kTcb3PKOUwtUDZnemZlLVU/view?usp=sharing

[Andy Chow (thegreatandychow)]

Same bug, Ubuntu server 17.10. Trying to install / on an SSD, and /home on a separate drive. Tried twice in a row. I have 3 drives connected, all SATA-3. An SSD (60 GB), and two HDD (5TB). One HDD I didn't touch during install, was going to set it up post installation. No removable drives or anything exotic.

[Andy Chow (thegreatandychow)]

Same bug, Ubuntu server 17.10. Trying to install / on an SSD, and /home on a separate drive. Tried twice in a row. I have 3 drives connected, all SATA-3. An SSD (60 GB), and two HDD (5TB). One HDD I didn't touch during install, was going to set it up post installation. No removable drives or anything exotic.

[Bjelleklang (bjelleklang)]

Same as Andy. No fancy setup; one disk with a Win 10 install, one for Xubuntu 17.10 from minimal CD containing /, and a separate disk with /home from the previous install. Also had issues with the 4.13.0.25 kernel and initramfs not behaving (messing up previously installed kernels), which led to reinstall.

[Andy Chow (thegreatandychow)]

I was able to "fix" this bug by installing everything into a single disk, and post-install moving /home to another disk. Not that hard, but there is definitely something broken trying to install / on one drive, that /somethingelse on another drive.

[totoro (ton-ami-totoro)]

Hi

Bug still present in 18.04.
I make my install from netboot(http://archive.ubuntu.com/ubuntu/dists/bionic/main/installer-amd64/current/images/netboot/)

os-prober version is 1.72 in the target.

And I use a preseed file: (https://pastebin.com/6jB5xVJT)

After few try and different partition and grub configuration, I have all the time the same problem.

The solution in the debian (https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=853187) help to pas the problem, but preseed is to a automatic install, I don't found a way to override the bug without do it manually. If you have any idea ?

[totoro (ton-ami-totoro)]

Hi

Problem is back after few days ... I don't understand why...

No change with my test.

[dh (dcharvey)]

Also suffering from broken pxe installs as a result of this. Can it be revisited please?

[Phillip Susi (psusi)]

Please file a new bug report and attach /var/log/syslog.

[Nic-t (nic-t)]

same as Petr Menzel (petrmenzel), when I install 18.04.1-server-amd64.iso with kinston datatraveler 2.0.

How to get /var/log/syslog when the screen stop at https://drive.google.com/file/d/0B6vr-kTcb3PKOUwtUDZnemZlLVU/view?usp=sharing

the host machine is Think server ts250 with 2 1TB SATA, host raid 1.

[Lenin (gagarin)]

Fix Released is wrong, this bug is still present.

[Lenin (gagarin)]

I had another two machines with the issue, just left them hang over night and after about 6-8 hours they continued the installation. FULLY AUTOMAGIC! no interaction required, just few more hours

[Yannick Croteau (eulogy)]

Got the problem with my personal mini server/mediacenter with 18.04.1-server-amd64.iso (not live, that one look to don't support the EUFI)

MSI B450I Gaming AC plus
Ryzen3 2200G
16Go Mem
Samsung M2 Drive (main installation)
Installed from Kingston USB Drive

I will try the "waiting" method..

Waiting in process...

(Before that, didn't get any problem on the Fedora 29 Server installation)

[Thomas (rundablage)]

I have the same problem on a Fujitsu TX150 S7 with Xeon X3450. Installing with mdadm on two SSDs. Going to leave it overnight as well and will see what happens. The last debug messages are

os-prober: debug: running /usr/lib/os-probes/50mounted-tests on /dev/sdb1 [this is the second member of the mdadm RAID1 array]
50mounted-tests: debug: creating device mapper device /dev/mapper/osprober-linux-sdb [cuts off]

[Thomas (rundablage)]

Follow-up: I was able to get around the problem using the reaction workaround described in this debian bug https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=853187#25

Had to kill two dmcreate processes.

[Lenin (gagarin)]

should be merged with? https://bugs.launchpad.net/ubuntu/+source/debian-installer/+bug/1769044

[dh (dcharvey)]

As this is marked fix released I've updated the bug rundablage links to above.

TL;DR
I've observed a pattern of it impacting a set of devices with KXG60ZNV512G NVMe TOSHIBA 512GB but not with Intel 760p 512GB

[Clint Armstrong (clint-f)]

This has happened to me on a variety of hardware installing 18.04. Killing the dmsetup process allows the installation to continue.

[Evan Hoffman (ehoffman24)]

Affecting Dell DSS1500, 100% of the time. SATA drive.

[Evan Hoffman (ehoffman24)]

I may have found a workaround/fix, I posted to my own bug report:

https://bugs.launchpad.net/ubuntu/+source/debian-installer/+bug/1854839/comments/2

[shangsong (shangsong)]

  I can reproduce the issue when install 18.04 with HWE, the workaround(Kill the dmsetup process or add GRUB_DISABLE_OS_PROBER=true to the kernel options) can works.

  Whether will build the new os-prober(1.75) for fixing this issue or not? If yes, what is the schedule?