CentOS yum update GUI no longer loads

LQParsons · 09-22-2017, 05:23 PM

Hi.
In a world of hurt, did nothing wrong, so there is nothing for me to undo or redo;
but lots to correct.
I have a large enough system running CentOS/7, QEMU as my KVM, and mostly run a CentOS Linux VM and a Win/7 VM.

Most Mondays for years, I start my work-week running in my Linux VM

Code:

yum update

and generally it's nothing, occasionally it's something, and every once in a while it's a huge update. Last Monday (18-Sep-2017) it was huge, so I busied myself doing other things, occasionally glancing at the terminal to see the progress. Right at the end it was a crash and burn like I've never seen--can't give details now, but the system crashed and I had a hard time rebooting -- couldn't mount disk, finally was able to

Code:

xfs_check
xfs_repair

etc. to be able to boot.

I mean XFS is so bullet-proof fsck is linked to fsck.xfs which returns 0 just to not break the start-up scripts. How often does one have to do the *_check/_repair!?!?

Took me a while to get it to reboot successfully, but it'd quickly crash; figured out how to single-user boot -- things have changed since I last did that with SuSE a decade ago.

My GNOME GUI no longer loads. This Is My Most Major Problem!

It seemed to be a gdm password problem, the solutions on the internet didn't work exactly -- evolution changed where files were etc. Maybe it's a PAM problem. I've tried this and that, nothing seems to work, nor is there any errors (save crash & core-dump), but it's making some progress.

ROOT gets an email when startx fails, but though it's over 100 lines long and tells what wonderful things it's done, it doesn't suggest where the problem lies. The X11 logs it refers, also it many lines long and tells what wonderful things it does before it gets a SIG abort. DMSG gives me loads of information but no clue. Etc.

I have done a

Code:

yum reinstall (mumble)

for this bright idea or anther to no avail. I can't re-install gnome, as it's a group, following the instructions to remove it so I can install-group it, gave me over 100-programs to delete, and I haven't that much courage to continue.

With customers etc. which to attend, it's taken me some calendar-time to work on my own issue here, and here I am at the end of my work-week hoping some of you out there can help suggest.
Yesterday, I did finally get XFCE installed and running so at least I can now get FIREFOX and Thunderbird again.

So, I did nothing wrong, so I don't know what to redo, undo, etc., but something got screwed up with the YUM update crash, and perhaps my poking and prodding has exacerbated things.

Oh, it seems my nautilus-desktop can't load -- my ~/.xinitrc looks like this:

Code:

# exec gnome-session
exec xfce4-session

I change the comment when I want to try gnome again, but my logs are saying my nautilus-desktop is failing to load (for no reason) when I do the 'gnome-session' (Who knew?)

Can anyone point me in better directions!?

zafar_dandoti · 09-23-2017, 03:22 AM

got the same issue earlier, the system failed to load GUI logoin screen after the update was interrupted.
it was just the gdm service which somehow disabled.

Code:

systemctl enable gdm.service

and

Code:

systemctl start gdm.service

worked.

LQParsons · 09-23-2017, 09:43 AM

Thanks. Tried it. The 'start' got it to come up for a moment, then crash again, looked at logs and email to root, tells me lots of stuff it did on the way up, and on the way down, without a hint of what to fix.

LQParsons · 09-24-2017, 08:03 AM

Hello:
Thanks for the help.

I'm on a virtual machine, I have access to all my files, just can't get GNOME working again, but have XCFE which sucks, so there's nothing I can't eventually, if painfully, do.

All solutions, both here and InternetSearching have sometimes done something, sometimes not, but I've learned lots of interesting things--such as we no longer use inittab which has been replaced with systemctl. ha ha ha

Something got screwed up during the inexplicable crash, perhaps I haven't found it, perhaps in my trying to fix it I screwed it up more.

So I had decided if it wasn't fixed by this week-end to just build myself another VM; you get so much crap over the years with things installed and deinstalled, things that "had I know then what I know now I'd've done it differently", it's probably a good thing to do once in a while.

It's fairly simple to 'scp' between two VM on the same physical machine, I've been good in keeping changes in 'local' and with so much now 'global' -- the mail's IMAP, dropbox etc. has much of the files. What the heck, eh?

Thanks again all!