Systemd came in - here's how to boot again. :)

GoinEasy9 · #1 29th July 2010, 02:39 AM

I had to add "init=/sbin/upstart" to the kernel line at boot. That's the temporary fix. I would like to see how it works, so I'm using the temporary fix.

It's failing to load distcache.service & default.service then it locks up. At least there's a way to boot the machine again.

smr54 · #2 31st July 2010, 04:00 AM

Actually, that one didn't work for me. Not sure why, and no real inclination to troubleshoot.

Wish they had left it as optional till it was tested more thoroughly. Not blaming the developer really, he did mention that he tested extensively on his own machines, and did leave the option for upstart. Not sure why upstart didn't work for me, nor the suggested fix (which I did by booting from puppy and then doing a chroot.)

Seems like it will be fixed relatively quickly though, they seem to have pinpointed the issue, that it's not creating targets.

https://bugzilla.redhat.com/show_bug.cgi?id=618315
(Figure it's good to get the bugzilla link in here, many folks will probably search the forum first if they get bitten.)

GoinEasy9 · #3 31st July 2010, 04:11 AM

I wrote that original post before I found, and added comment to that bugzilla. I'm surprised that init=/sbin/upstart did not work for you though, you're the first I've heard that wasn't able to recover.

#4 31st July 2010, 04:42 AM

Quote:

Originally Posted by smr54

Actually, that one didn't work for me. Not sure why, and no real inclination to troubleshoot.

Wish they had left it as optional till it was tested more thoroughly. Not blaming the developer really, he did mention that he tested extensively on his own machines, and did leave the option for upstart. Not sure why upstart didn't work for me, nor the suggested fix (which I did by booting from puppy and then doing a chroot.)

Seems like it will be fixed relatively quickly though, they seem to have pinpointed the issue, that it's not creating targets.

https://bugzilla.redhat.com/show_bug.cgi?id=618315
(Figure it's good to get the bugzilla link in here, many folks will probably search the forum first if they get bitten.)

i believe i think 2 are working on systemd for Fedora anyway, Rahul an that dude that did PulseAudio

---------- Post added at 01:42 PM CDT ---------- Previous post was at 01:41 PM CDT ----------

https://fedoraproject.org/wiki/Features/systemd

smr54 · #5 31st July 2010, 05:03 AM

That's Adam. We can blame him, because he's in QA.

(It's always fun to blame him.)

Seriously, I'm a bit surprised too--upstart was left in as a safety net, and somehow, I managed to find a hole in it. I'm not sure where I erred, and as mentioned, am not in a troubleshooting mood. My rawhide installs are always rather minimal, and as spare time is errm, spare, these days, I don't know if I'll find the time--it's probably some minor error I made somewhere (perhaps in trying the fix in comment 3 of the bug report), but regardless, by the time I find the time to try to figure out what I did wrong, it'll probably be fixed.

GoinEasy9 · #6 31st July 2010, 05:37 AM

Let's see, with the dependency problem waiting to be solved with the python 2.7 and boost rebuilds, the systemd/upstart problems, the AVC denials, the nouveau/clutter/mutter problems, the missing and/or non-functioning icons due to missing gnome apps, etc.etc.. I may just wait for a bootable F14 alpha iso and do a reinstall of rawhide. All the work arounds and fixes that I tried applying over the past month or so has probably put the install into such a state that cleaning it up really sounds like a good idea. I hate wondering if the apps I'm testing are being affected by real bugs, or modifications I've made to work around old bugs. Besides, the delay of Gnome 3 might cause some apps to revert back to gtk2 until it gets closer to Gnome 3 release.
A working F14 alpha with systemd working might be the best way to start a new rawhide install.

---------- Post added at 12:37 AM CDT ---------- Previous post was at 12:34 AM CDT ----------

hey, now that F14 and rawhide have been branched, we might see some stability on the horizon.

smr54 · #7 31st July 2010, 12:09 PM

I think I"m with you, will probably wait a bit. I enjoy testing, but these days, have no time, and no time to help troubleshoot, meaning all I can do is point out problems if I run into them.

(I'd actually forgotten about the systemd thing--I was planning to wait a few days till they fixed it, but ran into X suddenly seeming to have trouble with my Asus1000HE touchpad--movement of the mouse was painfully slow, so figured I'd run an update.)

GoinEasy9 · #8 31st July 2010, 03:51 PM

One of the errors I read due to the systemd install was that of slow mouse movements. Probably due to a "cpu gone wild" ... LOL ... you need something like gkrellm running to actually witness that. Every report/thread I read, said that at least one core of whatever processor running was up to 99%, that'll slow down a mouse.

nimnull22 · #9 31st July 2010, 04:59 PM

Why don't you compare "active" servises with "upstart" and with "systemd"
They both use different config.
chkconfig - for upstart
systemctl - for systemd

dd_wizard · #10 31st July 2010, 07:35 PM

On my machine, it's Xorg that's keeping one cpu 100% busy. I've no idea what it'd doing though.

dd_wizard

nimnull22 · #11 31st July 2010, 07:38 PM

Does it happen with upstart or with systemd?
Powertop will help you. But on my i686 it segfaults.

dd_wizard · #12 31st July 2010, 07:56 PM

That's with systemd. I used "sudo htop" to find the culprit. The new upstart seems to be fine, except I'm becoming convinced it always does a shut down, even when I click restart.

dd_wizard

nimnull22 · #13 31st July 2010, 08:31 PM

If the difference lies between systemd and upstart you can also use "ps -ef" and compare. Something should be different.
And what do you mean "The new upstart", that upstart was from 2010-07-24 01:54:53. I use it, and it perfectly does reboot and shutdown.

dd_wizard · #14 31st July 2010, 08:55 PM

The difference is that Xorg itself uses 100% of one cpu. So something is slightly broken when booting with systemd. Several users mentioned it in the bug thread. As for upstart, I'm using:

Code:

upstart.x86_64        0.6.5-7.fc14         @rawhide

dd_wizard

GoinEasy9 · #15 31st July 2010, 11:08 PM

Yes I've noticed the Xorg causing it. And it's true, the "CPU gone wild" stops if you log out and then log back in again. But with systemd the desktop is useless, I can't bring up a browser using it, so waiting for fixes.
I'm also using upstart 0.6.5-7 but the 32 bit flavor.

---------- Post added at 06:08 PM CDT ---------- Previous post was at 06:06 PM CDT ----------

Oh, BTW dd, hitting restart does shutdown with upstart, you not hallucinating. Not yet anyway.