Must reboot to get to GDM...
 

Hi all,

I am a very happy F7 user, was using FC6 before (and FC5, FC3 also...)

But since about 2 months, in FC6 and now in F7, whenever I first turn on my computer in the morning, instead of getting at once to the GDM (login screen) quickly after the regular Fedora boot routine, I get a blue screen with the little arrow of the mouse, and that's it !

I have to reboot with either the power button or the reset button, sometimes up to 4 times, and then at last, the GDM shows up at once, after the Fedora boot routine...

I am wondering what can be the cause of that (small) nuisance ???

As a matter of fact, I upgraded to F7 from FC6 with a complete clean install, so it is like a new OS I have now, since about a week.

If any of you has an idea for me, it will be very welcome.

Thanks in advance for your time !

That's happened to me a few time, but a <Ctl-Alt-F7> usually gets me to the correct screen. I think that there may be a bug in the X11 server code that's failing -- sometimes -- to switch to the X display after the boot gets to the startx. I tried the <Ctl-Alt-F7> because <Ctl-Alt-F1> worked to get me to a TTY session, and, when I saw X running, I tried switching back to the display, and there it was. Weird, eh?

By the way, I use kdm in preference to gdm, That's why I'm pointing a finger at X rather than the display manager, (I prefer kdm because the user switching is much more mature then the support available in gdm.)

Thank you for your tip, I'll try it tomorrow morning, and post my feedback here !

Ok, so this morning, I got stuck as usual with the "blue screen". So I did a <Ctl-Alt-F7>. It lead me to a black screen with only a flashing dash. So then I tried a <Ctl-Alt-F1>. That lead me to the Fedora booting routine, and it was stuck at that step :

enabling /etc/fstab swaps

-How strange !

Happened to me this morning too. But the system was not stuck at "enabling /etc/fstab swaps." That's just the last line displayed on that screen. I did a <Ctl-Alt-F2> to open another TTY window, which did open, but then startx finished whatever was blocking it, and my usuall log-in screen was displayed (at <Ctl-Alt-F7>, but I didn't need to press that).

Interestingly, I was booting to a GNOME desktop instead of my usual KDE one (because I'd switched to a GNOME desktop during a KDE session yesterday, so my default was set to "GNOME," which I'd forgotten this morning when I powered up my laptop). Perhaps it's a GNOME thing, but I don't really think so, since I first noticed the problem booting to KDE.

I just looked at my /var/log/Xorg.0.log file, but I didn't see anything strange. I do wonder, though, about my video driver. I've got an ATI O.E.M. clone in the laptop, and I'm using the FOSS "ati" driver from xorg. Perhaps the fglrx driver from livna might work better?

Last night I let the PC on because of a download. But while reading your post, I am starting to wonder if I shall have to disable the GUI login manager, and use a more "Slackware-like" approach with a CLI and a "startx"... That would make my wife miserable, as she has an account in that box, and she is very unfamiliar with anything CLI...

Anyway, I thought about posting my /var/log/Xorg.0.log here, but it is too long, and I can't do it ;)

I am thinking of setting up a dual-boot with another distro, maybe Debian, maybe Arch... I don't know yet, so I could compare... I installed Debian in my (brand new) VMware yesterday but it is just a toy, and I shall have it directly on another partition...

Thank you for your feedback, as all the tricks you are providing are useful !

I hope you understood from my post that I didn't do the startx. What I said was that the startx that the boot process was running had finished (successfully) before I got past the login prompt on the second TTY screen.

Fedora 7 just put an update to X11 in updates which I installed yesterday. I haven't had the problem since that install. (Of course, I didn't have the problem on every boot before the update, so I can't tell if there's any relationship.)

Good morning !

It has been a real challenge to boot to my GDM today ! I guess I had to reboot around 10 times !

So I thought about booting in that long forgotten winxp partition, and it gave me a BSOD mentioning ATI...

That leads me to think it is a hardware problem : I live in a very humid and dusty environment, close to the ocean (tropical climate). When it gets really warm, I have to leave my PC's case open so it doesn't over heat... So I shut down the PC, moved a bit my old ATI 9200 video card, and rebooted directly to the GDM without a problem...

Tomorrow I will try to move the video card BEFORE trying to boot... I'll post my feedback then...

The morning after...

I moved my video card before turning on my PC... Still I got that blue screen instead of the GDM... Rebooted in my winxp partition, worked fine... Rebooted in my F7 partition : blue screen ! Turned off the PC, moved a bit more my video card, rebooted, and F7's GDM showed up...

I am not sure anymore it is a hardware problem... It could be some sort of Fedora issue...

Will disable GUI login tonight, and boot with "startx" tomorrow, and I'll post my feedback...

Ok, now I know where to stand with this problem :

I had some time available yesterday afternoon, so I set a multiboot on my main box with F7 and PCLinuxos (this is the hype distro of the moment, so why not ?).

Today, wake up, coffee, first boot, and I am directly in the KDM (it is a KDE distro this time...) without any need to fight with reboots or touch my video card !

That helps me a great deal to pinpoint the problem to Fedora...

I am thinking of wiping my F7 partition to install Debian Etch...

But now I know this weird issue is not a hardware problem, it is a Fedora issue.

You may be correct, although, for example, it might be a kernel issue since Fedora is running 2.6.21, and PCLinuxOS may be on an earlier kernel release. (Note: I don't actually know what kernel version PCLinuxOS is using.)

Indeed... I am on F7 right now, but i believe PCLOS is with 2.6.18.something...

Since I am in my multiboot mode, and I have some free time this afternoon, I just deleted some movies to make some space to create (yet another) partition and to install Debian Etch as well...

I just can't wipe F7 because it is such a nice release, and as much as PCLOS is efficient, I am a Gnome guy at heart...

I just noticed something that may be germane to this discussion:

My access point disappeared from my Network Manager display, and I rebooted using the "interactive" boot option so I could see any error messages generated during the boot. (Sometimes the messages go off the top of the screen before I can understand them if I just display the messages in graphical boot mode.)

Anyhow, about the last service started was the Network Manager, and a message was displayed that "eth0 could not be found so activation of eth0 was aborted." Them the system seemed to "hang" for several minutes and, finally, a second, identical, message was displayed about eth0, and then eth1 was connected.

Now, on my laptop system, eth0 is a Ethernet port not usually plugged in, and eth1 is the wireless connection. When I reviewed my settings in system-config-network, I saw that starting eth0 during boot was not checked.

Then I remembered that, when I was first trying to get my network to configured, I'd done an ifup eth1 and got the same message about eth0 not being connected. I'd looked at the ifup script but I could find no mention of eth0 hard-coded in it. I didn't pursue the problem then, but perhaps I'll submit a bug report now.

Anyhow, my point is that perhaps you're having the same problem: that when the Network Manager starts, it attempts to start all the network devices that can be found, and it has to wait for the pings to the disconnected devices to time-out before it will let the boot proceed. (By the way, I run the ifplugd service which is supposed to detect unplugged devices and keep other programs from trying to use them, but the Network Manager does not seem to be aware of the fact that the device is not connected.)

I don't know... Maybe... On my side, since I now have a multiboot system, the only way to go is to boot into another distro (now I have PCLOS & OpenSUSE), surf the web a bit, and then reboot into Fedora... Otherwise, I still get that annoying blue screen...

Hum.:scratch: Well, surfing with the other distributions might be setting cached values at your ISP (or in your firewall appliance -- e.g., ADSL modem -- if you're using one) so that network response time is somewhat faster when you boot Fedora.

If that's at all plausible, it would mesh with my suggestion that the NetworkManager service started at the end of the Fedora boot was initializing network access, and -- as is well known -- the system may seem to be hung while the kernel is waiting for a network response.

You could test this conjecture by using the system-config-services command to turn off starting the NetworkManager service during boot, and start it by hand with a ./etc/init.d/NetworkManager start command after the boot finishes.
<edit>
Oh, also turn off NetworkManagerDispacher since it may start NetworkManager if it's started when NM isn't running.
</edit>

F.Y.I., I did post a bugzilla for the ifup (and ifdown) problem I noticed. If you're interested, here's the strange responses I get for those commands:There is about a two minute delay before the "done" is printed and the ifup eth1 command finishes (with eth1 started).

On this laptop, eth0 is the standard wired port and eth1 is the wireless connection. When I ran the commands, eth0 had nothing plugged into it. I ran grep on the ifup script (and the other if scripts), and found no explicit mention of eth0, so the problem may be in the programs invoked by those scripts.