From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4.1)
Description of problem:
Firstboot attempts to run, but fails with a window manager error of
being unable to open :1. There are no errors in the XFree86.1.log
and it looks like there are only some minor avc's involved.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. install from development/rawhide
2. reboot as normal
3. watch firstboot fail
Actual Results: firstboot fails to open the X-server started on vt02
claiming it can't open :1
Expected Results: normal firstboot processing
Computer: AMD-K6-2 @ 333MHz ATI Mach64 Video card 320MB RAM etc.
Attached is XFree log.
Created attachment 99017 [details]
X server log for firstboot run
Created attachment 99018 [details]
Messages abstract from firstboot time of system
Fresh development install from snapshot of Sat Apr 03,2004:
Still doesn't run firstboot correctly. This version of anaconda seems
to be very confused about SELinux, according to the install, SEL is
"disabled" and there are lots of kernel: security_context_to_sid
messages (probably unrelated to this bug)
Attaching XFree86.1.log for viewing
Created attachment 99095 [details]
XFree86.1.log related to firstboot cannot open :1 (Apr 03,2004)
as of 4/17/2004 the bug still occurs, BUT, if rhgb is allowed to run
it doesn't happen. Should add that bug occurs when rhgb is NOT used.
I have not been able to reproduce this bug. Can you provide more info
about how you disabed rhgb?
I simply didn't install it. Unselect rhgb from the packages list
using "custom" install.
I wonder if this wasn't an SELinux bug. I am unable to reproduce it
with test3. Can you try with test3 and see if it still happens?
I just did a full install from rawhide/test3 and the error occurred.
There was the usual lack of errors in the X server log and a flash of
"RunTime Error: could not open screen :1" as firstboot failed. I did
an install with "rhgb" in the mix, and firstboot works. Installing
with rhgb excluded makes firstboot fail.
Additional possibility, the video card is and ATI Rage XL/Mach 64
older card. Works well, but is a bit aged.
SELinux is not involved in this one. If you want to come over to
Durham, I can show you the setup. :-)
Full install of FC2rawhide as of 2005.05.04 (May 4) without RHGB and
firstboot *succeeded* to run. Seems the version of kernel and other
pieces let it work right on my hardware this time.
I'm going to close the bug.
FC2rawhide of 2005.05.09 *fails* to start firstboot correctly without
RHGB in the mix. ReOpening bug.
I just installed FC2 (final), and disabled rhgb (by editing the kernel
line in grub on the first boot of the new system). I watched the
system attempt to bring up the firstboot X server, which failed, and
then the regular X server fired up w/ GDM just fine.
I switched to vt0 (1, whichever I get from C-A-1), ran firstboot
manually and that worked okay.
This is a Dell Latitude laptop, NM256AV video.
Hmm, I'm just not able to reproduce this on my test machines.
ggw: can you try again with FC2 final and see if you still have the
I did an FC2 install both with and without rhgb included. If rhgb is
included, firstboot works fine; without rhgb it starts up the x server
on screen :1 and then fails with "Runtime error: cannot open display"
This is a 400MHz K6-2 with PCI ATI Rage XL (Mach 64 driver)
Installing from rawhide as of 2004-06-25 yeilds the same results.
Are there any xorg logs in /var/logs?
Created attachment 101572 [details]
Xorg.1.log for install of FC2 final
this is the requested xorg.1.log
I understand you are having trouble reproducing this bug. It seems to
be a problem with attaching the window manager to the :1 screen. The
display comes up with a root X cursor, and moving it works; but very
quickly the server exits and firstboot reports its failure.
Selinux is NOT enabled, so it is not a problem with that. There might
be a permission problem on the vt02 device, but since it is running in
root context that seems to be a non-cause. The biggest hint to me is
that having rhgb in the mix allows firstboot to work; there is
something not being properly initialized if rhgb isn't running.
A rawhide/development install of 2004-07-17 does NOT exhibit the problem!
Rawhide/Development/FC3t1 of 2004-07-23 still does NOT have this bug.
Unless some kernel change re-introduces it, I'd say it been swatted.
Ok, closing as Rawhide. Thanks for your help on this report.
Rawhide of 2004-08-06 re-introduced the problem
rawhide (development) of 2004-08-18 still exhibits the problem, same
no errors in xorg logs. Installing RHGB is still the workaround.
Firstboot is now failing differently on development of 2004-09-10
RHGB runs, then firstboot starts, but hangs forever (10 hours is
forever to a computer!) Running firstboot from root user login works
as a workaround. There are no exceptions or error messages in the
logs for this failure.
I am seeing this error too. It occurs on a fairly default
workstation install of FC3T2. It is not a SE Linux issue, it
happens when booted with enforcing=0.
I can provide administrative access to a machine that has this
problem with serial console if it helps. Contact me with my
redhat.com email address if interested.
Rawhide of 2004-09-27 shows the problem.
also happens on fresh install of rawhide of 2004-09-28
Me too, with 2004-09-27 rawhide, albeit with different symptoms than
what's being reported here (so maybe I need a different bug, although
I don't know for sure): firstboot seems to quickly fail (I didn't time
it -- I was away from the computer for several minutes -- but it sure
wasn't 10 hours) and the only evidence I see is:
a) "firstboot: failed" in /var/log/messages
b) /etc/sysconfig/firstboot is missing
fresh install of rawhide 2004-10-01 fails. Exact same symptoms.
rawhide install of 2004-10-15 has the problem. If RHGB is not
installed firstboot will not properly open the display for running...
If RHGB *is* installed it works okay.
rawhide fresh install from 2005.01.09 does NOT exhibit the problem.
Installing without RHGB and firstboot did run fine.
Please test with FC4 and report if it's still off-and-on broken or not.
Seems to be working with rawhide 2005-09-13, FC5. Also the FC4 final worked fine.