Comment 7 for bug 271246

After attempting upgrade today I had several errors with mysql, cups, and dpkg in general.
It all seems to be caused by the following error:

debconf: DbDriver "config": /var/cache/debconf/config.dat is locked by another process: Resource temporarily unavailable

I also noticed the same error in syslog (or .xsession-errors) earlier, while using startupmanager.
Upon opening synaptic:

E: dpkg was interrupted, you must manually run 'sudo dpkg --configure -a' to correct the problem.
E: _cache->open() failed, please report.

Prior to doing that, the recommended 'fuser -v /var/cache/debconf/config.dat' returned nothing.
Returning to startupmanager...it seems that su-to-root (and other related programs) was active, running something that gave the following in my .xsession-errors log:
---
A new version of /boot/grub/menu.lst is available, but the version installed
currently has been locally modified.

  1. install the package maintainer's version
  2. keep the local version currently installed
  3. show the differences between the versions
  4. show a side-by-side difference between the versions
  5. show a 3-way difference between available versions
  6. do a 3-way merge between available versions (experimental)
  7. start a new shell to examine the situation

What would you like to do about menu.lst?
---

So, in my case, this was a startupmanager problem - it has no facility to notify the user that this input is required...and waiting in a non-existent terminal. I added this as a bug a while ago, but no one is currently supporting startupmanager and the bug has gone without a fix. I dunno, maybe it's a problem with 'sudo update-initramfs -u'? To whomever this problem belongs, in my case initramfs noticed either the menu.lst~ or the menu.lst.backup and wanted user input to decide whether or not to update menu.lst - and startupmanager has no such facility to let the user know.

Upon killing ucf, this showed up in my .xsession-errors:

/usr/sbin/update-grub: line 1036: 11220 Terminated ucf --debconf-ok --debconf-template grub/update_grub_changeprompt_threeway --three-way "$buffer" $ucf_menu_file

So, maybe it's update-grub's fault? =)

Anyways, after killing ucf, 'sudo dpkg --configure -a' still didn't fix the packages, so I found su-to-root still running and End'd it; but still, update-grub & 'python /usr/sbin/startupmanager' (after attempting an End) are in a Zombie state and cannot be End'd or Kill'd. Even 'su kill -9' or -anything won't end them. ... OK, so 'sudo killall -g update-grub' finally killed them all.

Then I ran 'sudo dpkg --configure -a' and it told me that mysql-server-5.0 was in really bad shape and I should re-install it, the other 4-5 packages were configured OK. After a 'sudo apt-get install --reinstall mysql-server-5.0 mysql-server' everything seems OK and the 'sudo dpkg --configure -a' returns cleanly. And, most importantly, my LAMP server is functional again!

This may have been, and probably was, over-kill...but at least you know this error can be over-come (without a logout or restart). It is also not a bug of any one program, many different circumstances can cause this "debconf: DbDriver"config":" error to rear its nasty head.

Final Note: The requested 'fuser' command never once showed anything, it always returned cleanly. It does not appear to be a valid debug tool in this case.

*** Firefox/Launchpad won't let me delete the attachment, so you guys get to see my apt-get update terminal logs...how fun!