Description of problem: GUI/VNC installs are not working. Version-Release number of selected component (if applicable): anaconda-10.1.1.89-1 tree: RHEL4-U7-re20080507.0 How reproducible: 100% with GUI/VNC Steps to Reproduce: 1. start the installation (GUI/VNC) 2. proceed to stage 2 3. Actual results: X server starts. You can see the mouse cursor. Right after that the installation shuts down. GUI/VNC installations fail. Expected results: X server starts and the installation proceeds with GUI or VNC. Additional info: Does not affect text mode installs. Same behavior on different arches, bare metal and xen guest.
Created attachment 305113 [details] screenshot of virtual machine
It is possible that this is not directly related to the Xserver as anaconda tells us that is started successfully. The report also states that the mouse cursor showed up, which is a good sign that the Xserver was started successfully.
update: 1) I saw the text messages indicating anaconda is starting 2) Then black screen for some time until X starts 3) Then I saw the welcome screen 4) Then (I think) follows the partitioning screen but before it the installation began to shut down
mzazrivec indicates that this could be related to: https://bugzilla.redhat.com/show_bug.cgi?id=222301 Anaconda team what's your opinion?
Doesn't seem too likely that the bugs are related, though I won't rule it out completely. Can you grab the /tmp/ramfs/X.log file and attach it to this bug? You may need to add the nokill command line option so you still have a shell on tty2 when we abort the install. If something in X is segfaulting, that file should contain the reason why.
ok, I debugged this and it is unquestionably gtk. the installer goes bum when gtk.main() is called. Going bum means something happens internally in gtk that makes the anaconda thread exit, in turn making the whole installation exit. I wonder if this has anything to do with 222301.
I'm not going to clone with 222301 as I am not sure that they are the same issue. I will however, change the component to gtk2.
Created attachment 305127 [details] /tmp/ramfs/X.log I don't think it'll help though. The X server starts fine with nice background and initialized mouse pointer and stays that way for a few seconds, you don't get any widgets though (everything goes south after that).
The stacktrace makes it pretty clear that it must be related to the patch that was committed in bug 222301. I'm at a loss why, though, and would like to talk to Behdad about it. The patch changes to the code to be identical to what we ship in current Fedora, and it works fine there... Have you verified that the crashes go away if you back down gtk2 ? If so, I would recommend that we back out the patch for 222301 until we have a clear understanding of what is going on here. That patch was a bit of a last-minute addition, anyway.
*** Bug 446474 has been marked as a duplicate of this bug. ***
I can confirm that with gtk2-2.4.13-24.el4 and anaconda-10.1.1.89-1 the GUI and VNC installs are working.
devel_ack for the rollback.
for clarification ... it looks like we fixed the root issue instead of rolling back the fix that introduced this bug * Sun May 13 2007 Behdad Esfahbod <besfahbo> - 2.4.13-24.el4 - Another try, fix serial overflow in x11-geometry. - Resolves: #222301
RHEL4-U7-re20080507.0 had gtk2-2.4.13-23.el4 which contained fix for bug #222301. The problem was that fix for bug #222301 also introduced regression (gtk2 applications randomly segfaulting). Anaconda sure did pick up this version of gtk2, which resulted into graphical installation being broken. The implications of this regression were much more serious though -- whole gnome-session and all gtk2 applications were unusable. Anyway the problem was being taken care of in bug #222301 already and we have a new version (2.4.13-24.el4), which works just fine (see https://bugzilla.redhat.com/show_bug.cgi?id=222301#c55). So to sum up, no rollback is needed. This bugzilla is just confusing.
*** Bug 446129 has been marked as a duplicate of this bug. ***
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on therefore solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2008-0773.html