Upgrade to karmic koala

10/31/2009 11:16am

I started the upgrade of my server with the alternate cd, but I fell for the trick when it asked "by the way, do you want the install to get the latest packages from the internet?" I thought, fine that's a good idea. 90% of the packages are probably going to be found from the CD. How much longer could it take?

Bad idea.

I'm on the 3rd hour now.

10/31/2009 11:34am

Nice suggestion about optimizing my default server for downloads. But I can't switch it now without canceling. Huh, maybe canceling is a good idea.

It just seems to be ignoring the cd.

ooo, that cancel is sweet. It actually works.

OK, it is picking an optimum server now....

Huh, it likes university of chicago. I don't like that so much. I'm picking the Oregon State open source lab.

I'm starting the upgrade again from the alternate CD. I think I have about ten more seconds to work on this before the kids destroy the TV room downstairs....

10/31/2009 12:02pm

wow, that worked great. The upgrade really did ignore the internet when I told it to. All packages were quickly found on the CD. Now it is installing. I wonder if I still have internet connectivity to send this now...

10/31/2009 1:32pm

After boot, won't go into X and the text flashes like crazy during startup.

But most other things seem to be working... Network and firewall seem to be working. And mail server.

10/31/2009 2:54pm

I tried to ps aux | grep gdm and found it running and tried to kill it. But it was already dead. It is just looping trying to start and failing over and over.

I tried to find out what nvidia packages are installed and it looks like a mess:

root@weasel:/var/log# dpkg -l *nvidia*
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Cfg-files/Unpacked/Failed-cfg/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name           Version        Description
+++-==============-==============-============================================
un  nvidia-173-ker <none>         (no description available)
ii  nvidia-173-mod 173.14.20-0ubu Modaliases for the NVIDIA binary X.Org drive
un  nvidia-177-ker <none>         (no description available)
un  nvidia-177-mod <none>         (no description available)
ii  nvidia-180-ker 180.44-0ubuntu NVIDIA binary kernel module source
ii  nvidia-180-lib 180.44-0ubuntu Video Decode and Presentation API for Unix
un  nvidia-180-mod <none>         (no description available)
ii  nvidia-185-mod 185.18.36-0ubu Modaliases for the NVIDIA binary X.Org drive
un  nvidia-71-kern <none>         (no description available)
un  nvidia-71-moda <none>         (no description available)
un  nvidia-96-kern <none>         (no description available)
ii  nvidia-96-moda 96.43.13-0ubun Modaliases for the NVIDIA binary X.Org drive
ii  nvidia-common  0.2.15         Find obsolete NVIDIA drivers
un  nvidia-glx <none>         (no description available)
un  nvidia-glx-173 <none>         (no description available)
rc  nvidia-glx-177 177.82-0ubuntu NVIDIA binary Xorg driver
ii  nvidia-glx-180 180.44-0ubuntu NVIDIA binary Xorg driver
un  nvidia-glx-71 <none>         (no description available)
un  nvidia-glx-96 <none>         (no description available)
un  nvidia-glx-env <none>         (no description available)
un  nvidia-glx-ia3 <none>         (no description available)
un  nvidia-glx-leg <none>         (no description available)
un  nvidia-glx-leg <none>         (no description available)
rc  nvidia-glx-new 169.12+2.6.24. NVIDIA binary XFree86 4.x/X.Org 'new' driver
un  nvidia-glx-new <none>         (no description available)
un  nvidia-glx-src <none>         (no description available)
un  nvidia-kernel- <none>         (no description available)
un  nvidia-kernel- <none>         (no description available)
un  nvidia-kernel- <none>         (no description available)
ii  nvidia-kernel- 20080825+1ubun NVIDIA binary kernel module common files
un  nvidia-kernel- <none>         (no description available)
un  nvidia-kernel- <none>         (no description available)
un  nvidia-kernel- <none>         (no description available)
un  nvidia-legacy- <none>         (no description available)
un  nvidia-legacy- <none>         (no description available)
un  nvidia-new-ker <none>         (no description available)
un  nvidia-new-ker <none>         (no description available)
ii  nvidia-setting 180.25-0ubuntu Tool of configuring the NVIDIA graphics driv
un  nvidia-xconfig <none>         (no description available)

Well, whichever packages are installed, I should be able to fix my xorg.conf and tell it to use the nv drivers instead of nvidia...

I verified that xorg.conf was using nvidia drivers. But instead of trying to hack together an nv file, I think I can just completely remove it and let it use sensible defaults. Hey, it worked. That was easy.

Wow, very slick color scheme....

Now I think the first thing I have to repair is the temperature sensors.... I think lmsensors got whacked.

Uh oh, evolution is chewing up 60% of my cpu. I hate evolution.

10/31/2009 3:31pm

I spent a few minutes in Evolution settings trying to figure out how to disable it... couldn't find anything, so I removed it and the 4 things that rely on it. That fixed it.

Now my system load is back down to 0.06% where it belongs.

10/31/2009 10:20pm

The first phase of the upgrade was pretty painless. It wasn't until after trick or treating that I asked it to check via update manager for more updates and it found the tricky bits. Hobbit, imap, sql, mediawiki... all the stuff that I have heavily customized. So I had to think hard about pages full of differences and decide if I should keep or reject them.

Now that's done and I can't get mysql to start. I don't think I've run into this particular problem before.

root@weasel:~# /etc/init.d/mysql start

* Starting MySQL database server mysqld                                 [fail]

Weird. I hate silent failures. When I do a "sh -x mysql start" it seems to be trying multiple times and failing to find a clear socket or something....

+ ping_output=/usr/bin/mysqladmin: connect to server at 'localhost' failed
error: 'Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' (2)'
Check that mysqld is running and that the socket: '/var/run/mysqld/mysqld.sock' exists!

I created the missing directory /var/run/mysqld and tried again, but no go. This is all that appeared in daemon.log:

Oct 31 21:54:19 localhost /etc/init.d/mysql[25434]: 0 processes alive and '/usr/bin/mysqladmin --defaults-file=/etc/mysql/debian.cnf ping' resulted in
Oct 31 21:54:19 localhost /etc/init.d/mysql[25434]: #007/usr/bin/mysqladmin: connect to server at 'localhost' failed
Oct 31 21:54:19 localhost /etc/init.d/mysql[25434]: error: 'Can't connect to local MySQL server through socket '/var/run/mysqld/mysqld.sock' (2)'
Oct 31 21:54:19 localhost /etc/init.d/mysql[25434]: Check that mysqld is running and that the socket: '/var/run/mysqld/mysqld.sock' exists!
Oct 31 21:54:19 localhost /etc/init.d/mysql[25434]:

Time for a reboot after all those upgrades...

After a reboot, mysql still fails to start, but now, trackerd is running amok. Oh look, it calmed down. I guess they fixed that problem from last time.

But still, what's up with mysql

Hobbit is working, that's cool. And it tells me that everything is green except for mysql and slapd. Of those two, mysql is the higher priority.

The ubuntu software installer tool told me that mysql server was not installed. Odd. So I installed it. Installing it sure made it start up a lot better.

10/31/2009 10:39pm

Why doesn't ldap start?

The ubuntu software store thingy doesn't have anything to say about ldap or openldap or slapd. When I asked synaptic the same questions, it looked like it was actually going to have some good answers, but then it crashed. The new bug buddy was launched but then found the the type of crash I got was not something that it could report. Oh well.

After launching synaptic again, it was better behaved and told me that I don't have openldap installed. That's dumb. I used to have it installed. And it isn't obsolete. So I installed it again. That mostly fixed ldap. That is, it starts now, and my contacts seem to be intact, but not quite.

Actually ldap is fine now. The problem is with the flaky version of GQ, my ldap browser. If I use some other tool to view my ldap contacts, it works fine.

So I think completes my upgrade. That was weird to have it throw out perfectly good packages in the upgrade this time around...

11/10/2009 11:59am

Why doesn't sensors report the same information?

Before the upgrade, sensors -Af | grep -i temp had this output:

k8temp-pci-00c3
Core0 Temp:  +86.0°F
Core1 Temp:  +89.6°F
M/B Temp:    +91.4°F  (high = +33.8°F, hyst = -184.0°F)  sensor = thermistor
CPU Temp:    +95.0°F  (high = +176.0°F, hyst = +167.0°F)  sensor = thermistor
temp3:       +66.2°F  (high = +176.0°F, hyst = +167.0°F)  sensor = diode

After the upgrade, it has this output:

temp1:       +71.2°F  (crit = +206.2°F)                  
k8temp-pci-00c3
Core0 Temp:  +80.6°F                                    
Core1 Temp:  +80.6°F       

How do I get my motherboard and case temperature back?

It doesn't appear that there are any differences between /etc/sensors.conf and my backup sensors.conf which is dated 7/4/2008. Nor is there any difference between /etc/sensors3.conf and the backup.

The answer is probably to run

sensors-detect

and accept all the suggested defaults. It determined that I need to add the following to my /etc/modules

#----cut here----
# Chip drivers
w83627hf
k8temp
#----cut here----

Unfortunately, those were already in my /etc/modules as of May 6, 2008, so that wasn't the problem.

I think the k8temp module is properly installed:

root@weasel:/var/log# lsmod | grep k8temp
k8temp                  5504  0 

But the w83627hf is not:

root@weasel:/var/log# lsmod | grep w83627hf
root@weasel:/var/log# 

And when I try to load it, I get this:

root@weasel:/var/log# modprobe w83627hf
FATAL: Error inserting w83627hf (/lib/modules/2.6.31-14-generic/kernel/drivers/hwmon/w83627hf.ko): Device or resource busy

Is this related to the new strictness of kernel 2.6.31? http://www.lm-sensors.org/wiki/FAQ/Chapter3#Mysensorshavestoppedworkinginkernel2.6.31 If so, I guess I'm hosed.