SERIOUS BUG regarding fsck/RAID and HFS+/netatalk or samba


I want to let people know there are two issues with Mac OS X Server 1.2. There is no good reason to believe that the Mac OS X Beta is not similarly affected.

The first problem occurs when attemping to fsck a UFS volume. fsck will hang when executed. The problem occurs when using the Infortrend IFT-3102U2G RAID controller on a 400GB volume. This hang will occur even on a freshly-initialized disk in single-user mode. This problem may be related to the volume size and not the RAID controller (as I had previously believed). If you are using a relatively large disk or any type of hardware RAID, you should backup your data and then test whether a forced fsck works.

The second problem concerns HFS+. It happens when using netatalk or samba to copy very large files (>> 100 MB) to the server, which will cause the server (and hence the client) to stall. The larger the file the longer the stall. For example, a 760MB file stalls for about 46 seconds. For files about 1 GB in size, the copy can fail and the connection to the server lost! Under Apple File Services, this stall occurs only when the file copy is cancelled during mid-copy. This issue appears to occur on all HFS+ formatted volumes.