If you use the Zarafa webaccess (6.40, but apparently also 6.30) on Ubuntu 9.10, apache2 will nicely segfault and you will not be able to access your webmail. Upside: the mobile server will keep functioning. Even Fedora has this problem and it was tracked down to an error in PHP. I do not want to compile my own software to work around this issue, since I am lazy and I use a package manager for a reason! I quickly found the least nasty workaround for this issue in this post on the Zarafa forums.I replace /usr/lib/apache2/modules/libphp5.so (from karmic’s 5.2.10) with the libphp5.so from the 5.2.6 libapache2-mod-php5 package. At least the file will still be managed by dpkg and thus updated/upgraded if neccessary. As Ubuntu 10.04 is just around the corner, I will suggest the owner to does not update libapache2-mod-php5 until he upgrades to 10.04.

Has it been 6 months already? Time flies when you are having fun or when you are very busy, it seems. The Dutch and English announcements can be found on the official Ubuntu wiki as well as the Dutch and English release notes. Another great upgrade, loaded with new stuff and changes. Here are some tips to get you through the potentially “rough” part of it

Fast upgrade

I gave a friend of mine a great tip. That tip enabled him to upgrade his system in less than an hour and downloading the updates took less than 10 minutes (at about 1.6 MB/s). I told him that the default is to use the official nl.archive.ubuntu.com as an update source and he should use an alternate local mirror if the main one is slow. Just add a line to your /etc/hosts :

ip.of.the.alternate.mirror nl.archive.ubuntu.com

Do not forget to undo this after the upgrade though

GNOME-RDP

Apparently the GNOME-RDP database “breaks” again if you upgrade. This bugreport also contained the “fix” (by Mick K), but make sure sqlite and sqlite3 are installed:

sudo apt-get install sqlite sqlite3
cp ~/.gnome-rdp.db ~/.gnome-rdp.db.backup
mv ~/.gnome-rdp.db ~/.gnome-rdp.db.old
# Open GNOME-RDP to create and initialize a new database and then close the application:
gnome-rdp
sqlite ~/.gnome-rdp.db.old “.dump session” | fgrep INSERT | sqlite3 ~/.gnome-rdp.db
# Keyring support slows it down a lot (and produces errors) so disable it:
sqlite3 ~/.gnome-rdp.db “INSERT INTO appOptions VALUES(‘UseKeyring’,'False’);”

Note: please keep your backup file in place (an extra backup is never a bad thing), since GNOME-RDP seems to have database issues when upgrading. I can not ensure this is ‘the’ best way to fix your current database, so use with caution.

Compiz and Intel

I suddenly found myself unable to enable compiz (“Blacklisted PCIID ’8086:2a02′ found”). After a quick Launchpad error search I found this bug which referenced the Dell site, which eventually contained the workaround. Just edit /etc/xdg/compiz/compiz-manager and add the following line:

SKIP_CHECKS=”yes”

CTRL+ALT+BACKSPACE

I really do NOT get this. This shortcut was disabled because some people accidentally triggered it and this restarts X. Which idiot accidentally presses CTRL+ALT+BACKSPACE? Anyway, this is how you get the famous shortcut back and working as intended:

sudo dontzap –disable

I found a screenshot from ‘top’ I ran in ‘xterm’ from quite a while ago (I guess at least 6 months). I was wondering why my machine became a bit sluggish and I suspected my dnetc client. It seemed that iptotald was using way too much CPU power, for the job it was doing (getting some ethernet traffic statistics). After removing iptotald from my system, it became faster again and my frown was turned upside down.

When I finally upgraded my desktop to Edgy today and I noticed tabs in fluxbox. I didn’t like them, so I wanted them gone. After I took a look at the fluxbox website and the fluxbox wiki, I got the solution. Since that didn’t work, I went on IRC to ask the people in #fluxbox if this was normal behavior. When I ruled out ‘user error’, I quickly filed a bug report.

Aangezien ik steeds meer de CLI ga gebruiken, is toch wel fijn als je geen 640*480 hebt. Als ik met mijn fluxbox bezig ben heb ik een relatief klein font en draai ik 1024*768, lekker overzichtelijk. Zodra ik dus de X server kill, ga ik terug naar een ontzettend onoverzichtelijk en lelijk scherm.

De oplossing blijkt in het gebruik van de framebuffer te liggen die je kan enablen door een simpele “vga=xxx” bootflag aan je kernel toe te voegen. Het werkt zeer fijn, totdat je hibernate gebruikt. Zodra ik uit hibernate kom en de X-server start, blijft deze zich continu opnieuw opstarten, tot deze het systeem vast laat lopen. Erg nasty. Dit probleem heb ik trouwens ook als ik handmatig het MAC adres van mijn netwerkkaart aanpas. Bug report link