FC6 release. Tested on:
Tyan 2882 w/2 Opteron and 1G memory
Tyan 2882-D w/2 Dual dual core Opterons and 2G memory
Both systems using 3ware 9xxxx series raid controllers for afp storage.
Netatalk database files appear to get corrupted and can end up not only locking
up the share, but on occation also cause the afp client thread to 'jam'
consuming a large amount of process time and effectivly making it impossible
for the user to login a second time by locking up the Mac computer on login
The only way to clear this error is to kill the afp thread if hung and delete
the .AppleDB files related to that share. Next login recreates the DB files and
all works correctly for a time. No data files appear to have corruption
Using the same hardware, FC5 x86-64 install (with the same major number
netatalk) works correctly even with loads beyond 50 users
This problem appears to be fixed in FC7 release
I agree. I have been testing for three weeks without incident.