nvidia fun

nvidia giving me problems again

trying delete /usr/lib64/xorg/modules/extensions/libglx.so /usr/lib64/xorg/modules/extensions/libglx.so.290.10

to see if it is because of conflicting versions

that didn't do it so next rm
libnvidia-wfb.so.290.10
libnvidia-wfb.so.1
in /usr/lib64/xorg/modules

still no luck
more to expunge

./usr/lib64/libnvidia-cfg.so.290.10
./usr/lib64/libnvidia-tls.so.290.10
./usr/lib64/libnvidia-glcore.so.290.10
./usr/lib64/libXvMCNVIDIA.so.290.10
./usr/lib64/tls/libnvidia-tls.so.290.10
./usr/lib64/libvdpau.so.290.10
./usr/lib64/libnvcuvid.so.290.10
./usr/lib64/libGL.so.290.10
./usr/lib64/libnvidia-ml.so.290.10
./usr/lib64/libnvidia-compiler.so.290.10
./usr/lib64/libcuda.so.290.10
./usr/lib64/vdpau/libvdpau_nvidia.so.290.10
./usr/lib64/vdpau/libvdpau_trace.so.290.10
./usr/lib/libnvidia-tls.so.290.10
./usr/lib/libnvidia-glcore.so.290.10
./usr/lib/tls/libnvidia-tls.so.290.10
./usr/lib/libvdpau.so.290.10
./usr/lib/libGL.so.290.10
./usr/lib/libnvidia-ml.so.290.10
./usr/lib/libnvidia-compiler.so.290.10
./usr/lib/libcuda.so.290.10
./usr/lib/vdpau/libvdpau_nvidia.so.290.10
./usr/lib/vdpau/libvdpau_trace.so.290.10

no luck
a bunch of stuff in
/etc/X11

well, instead of crashing it crashes messily...

....

well, changing the video driver to vesa works well enough

Fall 2012 Lab

Got OpenSuse 12.2 RC2

set up authorized keys

fix /etc/fstab

Next, verify the repositories are correct. This is a little iffy since 12.2 is not fully out yet.

turn off autorefresh since it was bugging me
zypper mr -R --all

Retrieved list of installed packages in 12.1 using

rpm -qa --qf "%{name}\n"

Used

zypper in `cat packages.txt`

to try to install them all. It seems there was an API change in libcdio, so I'll need to redo the packages that depend on it later. Some of the repositories are based on 12.1


--- nfs mounts are not mounting during bootup...
- initially I had edited the fstab by hand, undid that and edited it via yast in case it ran some other configuration
- no difference
- next try see what systemd says after a fresh boot
- systemctl lists it as failing, and NetworkManager and virtualbox too
- removing NetworkManager and rebooting
- still no change
- adding exception to the firewall, it shouldn't matter since it does not get started, but it is worth a shot
- no change
- it is late, so trying a sledgehammer, editing /etc/init.d/boot.local to include mount -a
- no change



added guest users

installed virtual box
--- need to test it

--- need to install eclipse
--- need to install android development kit
--- need to install mathematica

onto authentication
- copy krb5.conf, ldap.conf, nsswitch.conf, pam.d,

change group
change hosts

--- check on the need to change mime types

--- check ssh server configuration

+++ redo yast boot loader config to use /dev/sda drive ids
- changed sysconfig/bootloader - it was interested in the /dev/sda5 partition though which is weird because it is swap

--- check firewall config, possibly add exception for nfs-client

www mechanize shell

This is a nice module to get started on scraping

 perl -MWWW::Mechanize::Shell -eshell

After navigating and doing a dry run you use the
script new_script.pl

command to save what you have done.

insserv

Created a startup script on one of the redcay machines to invoke ntp on startup since the battery is not reliable.

Used insserv to setup the proper sequencing for when the script should run, I like the prerequisite checking, it seems more reliable than randomly guessing a number.

automation

reviewed puppet and cfengine again, I'm not satisfied with how well it matches what I want to do.  Perhaps I don't understand the problem well....

Looking for Perl modules to help out on managing the packages, and not finding anything that looks good.  Many rpm modules are out of date, and the rpm2 module does not look useful for what I want.  So, Plan B is to just use system.

reboot

the lab machines are hanging on reboot, trying to fix

- tried adding flag to kernel bios=reboot|acpi, but neither helps
- it seems to reboot ok sometimes
- going through messages to get rid of warnings,
   - uninstalling openvm-tools
   - cleaned up fstab a bit, still having problems with double mounting /usr/local over nfs
   - changed the kernel options to bios=reboot

It seems better, but I'm not real satisfied, I'll have to test it more.  The bios=reboot was not working reliably before...

migrating mediawiki

first the faculty mediawiki
the plan:
   - set up new virtual machine
   - backup data into xml
   - restore and use sqlite instead of mysql

- need to upgrade os on files in order to install php5-cli


- not sure how well the upgrade went, php is still broken with respect to dumping

- installing postgres on faculty, will restore the db, migrate, then remove postgres

- some annoyances getting apache and the new media wiki configured, partly changes in configuration files, partly missing packages.

- now dump the database and try putting it into sqlite

it took forever to reload it, and this one was relatively small, need to reconsider for the other one.

virtual machine

trying to create a new virtual machine, here goes...

try yast2

Open Suse migration

kerberos

 - trouble with gui tools, at least remotely, so trying to modify the /etc files directly

running ntpdate pool.ntp.org manually

modifying
   - krb5.conf
   - pam config files
   - nsswitch.conf

having trouble contacting the ldap server now
    rechecking the ldap client config
    somehow the host line got removed...

for some reason ssh did not generate host dsa keys, copying the old ones

disabling sssd for now

credentials seem mostly ok