Install Woes with F25

[Tim Evans]

I've had multiple failed installs on a brand new Dell XPS 8900 today.

During the boot from bootable thumb drive, several hundred messages of 
the following ilk showed up:

[nn] printk messages dropped [timestamp] pcieport 0000:00:1c.0 PCIebus 
Error: severity: corrected, type=physical layer, id.00e0 (Receiver ID)

Finally, after a couple of minutes of these messages, the installer came up.

I reached the 'create-a-user' stage (language, keyboard, filesystems, 
root password had been filled in.) The install stalled on the 
create-user screen after I entered just one letter of the user's name in 
the box. That is, the screen would not take any other input.  Mouse 
cursor was still active and I could access other features of the live 
environment--e.g., open a terminal.

After 20 minutes or so, a pop-up message stating the Anaconda installer 
was not responding, wait/exit.  Probably shouldn't have, but I clicked 
exit, then tried to restart the install-to-disk process. This generated 
a message about an existing anaconda.pid file.

So, I rebooted it to try again.  Now the above-referenced messages (or 
ones very similar to them) display indefinitely (I waited over half an 
hour).  The installer never starts. (I tried both F25 and F24 install 
media, with the same results.)

My WAG is that the filesystem creations had not completed by the time 
the Anaconda-is-not-responding pop-up appeared.  (Disk is 1TB, and I 
told it to delete with Windows partitions.) Possibly, exiting instead of 
waiting left one or more filesystems in some unusable (read-only?) state.

This tends to be confirmed by attempting to boot from Windows 10 DVD, 
which reports it can't find required partitions to "reset the PC."

Can someone provide recipe for booting F25 single user and doing 
something to restore sanity to the filesystems, or just clean up the 
mess so I can try the install again? At this point, with neither a 
working Fedora nor any Windows, the system is without an OS.

Thanks.
-- 
Tim Evans			|5 Chestnut Court
443-394-3864			|Owings Mills, MD 21117

[Joe Zeff]

On 12/03/2016 01:00 PM, Tim Evans wrote:
>  During the boot from bootable thumb drive, several hundred messages
> of
>  the following ilk showed up:
>
>  [nn] printk messages dropped [timestamp] pcieport 0000:00:1c.0 PCIebus
>  Error: severity: corrected, type=physical layer, id.00e0 (Receiver ID) 
This sounds like a hardware issue.  Try booting from a LiveUSB with 
GParted on it and see if you can use it to partition and format the 
disk.  If so, and the installer runs, just use the partitions you 
created as is.  If not, take it back to the dealer; that's exactly what 
warranties are for.

[spam?] Re: Install Woes with F25

[Tim Evans]

On 12/03/2016 04:41 PM, Joe Zeff wrote:
>  On 12/03/2016 01:00 PM, Tim Evans wrote:
> > During the boot from bootable thumb drive, several hundred messages of
> > the following ilk showed up:
> >
> > [nn] printk messages dropped [timestamp] pcieport 0000:00:1c.0 PCIebus
> > Error: severity: corrected, type=physical layer, id.00e0 (Receiver ID)
>
>  This sounds like a hardware issue.  Try booting from a LiveUSB with
>  GParted on it and see if you can use it to partition and format the
>  disk.  If so, and the installer runs, just use the partitions you
>  created as is.  If not, take it back to the dealer; that's exactly what
>  warranties are for. 
Thank you, Joe.  Do you agree these messages are disk-related?

These errors start before startup of the LiveUSB environment completes. 
There are a lot of normal boot messages, the last one of which is 
"starting Gnome Display Manager." At this point, the errors start 
streaming and never stop.  (Just tried again--they've been scrolling for 
an hour and a half.)


Can you get to single-user mode from LiveUSB? The menus it offers don't 
seem to indicate this is possible.
-- 
Tim Evans			|5 Chestnut Court
443-394-3864			|Owings Mills, MD 21117

[spam?] Re: Install Woes with F25

[Kenneth Marcy]

On 12/03/2016 03:35 PM, Tim Evans wrote:
>  On 12/03/2016 04:41 PM, Joe Zeff wrote:
> > On 12/03/2016 01:00 PM, Tim Evans wrote:
> >> During the boot from bootable thumb drive, several hundred messages of
> >> the following ilk showed up:
> >>
> >> [nn] printk messages dropped [timestamp] pcieport 0000:00:1c.0 PCIebus
> >> Error: severity: corrected, type=physical layer, id.00e0 (Receiver ID)
> >
> > This sounds like a hardware issue.  Try booting from a LiveUSB with
> > GParted on it and see if you can use it to partition and format the
> > disk.  If so, and the installer runs, just use the partitions you
> > created as is.  If not, take it back to the dealer; that's exactly what
> > warranties are for. 
Ay ay ay, oh no!  The "bad hardware" response to brand new gear from a 
nationally reputable manufacturer ,,,

>  Thank you, Joe.  Do you agree these messages are disk-related?
>
>  These errors start before startup of the LiveUSB environment 
>  completes. There are a lot of normal boot messages, the last one of 
>  which is "starting Gnome Display Manager." At this point, the errors 
>  start streaming and never stop.  (Just tried again--they've been 
>  scrolling for an hour and a half.)
>
>
>  Can you get to single-user mode from LiveUSB? The menus it offers 
>  don't seem to indicate this is possible. 
I don't have a Dell XPS 8900, but I have seen reports on reddit.com of 
people with Dell XPS 8900 with Fedora 23 and 24 problems after having 
had several other Linux distributions running just fine on the same 
hardware. I'd suggest that, at this preliminary stage, blaming hardware 
is a tech support convenience rather than the preferable way forward. 
Persevere with this hardware; there is a solution toward Linux luxury to 
be found.


Ken

[Greg Woods]

On Sat, Dec 3, 2016 at 2:41 PM, Joe Zeff <joe(a)zeff.us&gt; wrote:

>  Try booting from a LiveUSB with GParted on it and see if you can use
> it to
>  partition and format the disk. 

As an aside: you can also boot from a LiveUSB generated from a standard
Fedora Workstation image, then "dnf install gparted" and run it that way.
That's how I shrank the Dell-provided Windoze partition on my new work
laptop to make room for Linux without trashing Windoze.

--Greg

[Samuel Sieb]

On 12/03/2016 01:00 PM, Tim Evans wrote:
>  I've had multiple failed installs on a brand new Dell XPS 8900
> today.
>  
I'm guessing it's an Intel Skylake chipset inside.

>  During the boot from bootable thumb drive, several hundred messages
> of
>  the following ilk showed up:
>
>  [nn] printk messages dropped [timestamp] pcieport 0000:00:1c.0 PCIebus
>  Error: severity: corrected, type=physical layer, id.00e0 (Receiver ID)
>  
Try adding "pci=noaer" to the kernel command line when booting the 
installer.

[Tim Evans]

On 12/04/2016 12:36 AM, Samuel Sieb wrote:
>  On 12/03/2016 01:00 PM, Tim Evans wrote:
> > I've had multiple failed installs on a brand new Dell XPS 8900 today.
> >
>  I'm guessing it's an Intel Skylake chipset inside.
>
> > During the boot from bootable thumb drive, several hundred messages of
> > the following ilk showed up:
> >
> > [nn] printk messages dropped [timestamp] pcieport 0000:00:1c.0 PCIebus
> > Error: severity: corrected, type=physical layer, id.00e0 (Receiver ID)
> >
>  Try adding "pci=noaer" to the kernel command line when booting the
>  installer. 
Thank you, Samuel.  This did the trick.


-- 
Tim Evans			|5 Chestnut Court
443-394-3864			|Owings Mills, MD 21117

[Chris Murphy]

FWIW, GParted doesn't work on Wayland right now. And on Fedora 25 Live
media Wayland is the default with auto login. You'd have to give the
liveuser user a password and turn off auto login, log out, then at login
switch to GNOME on XOrg using the gear icon.

Or 'dnf install blivet-gui' and use it instead.


Chris Murphy