If you are using guest access and getting failures due to CNID DB
Update:
It seems that a crashed appletalk daemon can corrupt the cdb database in the folder .AppleDB in the share, so change the share scheme to dbd
Hint on how to accomplish this (Ubuntu assumed - if you run another distro, you did not need this section anyway - ha):
Stop /etc/init.d/netatalk stop Edit /etc/default/netatalk to have CNID_METAD_RUN = yes Edit /etc/netatalk/AppleVolumes.default and change your share to have cnidscheme:dbd at the end Move the .AppleDB dir out of the way in the share /etc/init.d/netatalk start
-- This info is old (below), and just a stopgap fix for those using cdb db scheme back ends ---
You might try moving the .AppleDB folder in the share and restarting you appletalk daemons
It seems like it can get corrupted.
If you have processes stuck in state D (i/o wait), and -9 kill is not working, you might need a reboot as well.
Here is some info on that subject
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=239346
You also might want to have the clients run this in a command file (dotfiles-fix.command) or something (chmod a+x) and run - stops clients from writing .DS_Store files on network shares:
/usr/bin/defaults write com.apple.desktopservices DSDontWriteNetworkStores true