Hey guys- got a little samba server running for a small personal file share, and was having some trouble with a very specific scenario. So let's take it from the top:
My samba server is a CentOS 5.3 box, and is running a large share for simply dumping files. The main partition works fine- ext3, internal disk. That's not the problem. The problem was introduced when I thought to help a colleague who had an external disk he simply used as a dump as well. I offered to add a share on the server and mount his disk, so he wouldn't need to move it around the office, which is all he would do with it. Here's where I start asking for trouble. It's an NTFS partitioned disk, mounted over USB. When it works, it's great, but every so often (in regrettably undocumented conditions, but NOT upon reboot- that works fine) it corrupts the mount folder. So that leads to something like this:
ls -l /mnt
drwxrwxrwx 11 root smb_user 4096 Jun 22 15:28 bucket
?--------- ? ? ? ? ? matt_external
Weird, yeah? Connecting it to his machine again, all of the data is fine. It's only the mount folder getting clobbered. This has happened multiple times, and remounting to a different folder makes it work again for a time. Anyone know what's causing this? I know I'm asking for trouble, but this is a low-risk scenario, and it's an interesting scenario. It'd be cool to figure out how to make this setup sustainable.
Best Answer
The Linux in-kernel NTFS support is known more or less broken in write mode. It's OK in read-only mode, though. If you want to use an NTFS drive under Linux, use ntfsmount to mount (with FUSE) your drive, NOT mount -t ntfs. I suppose this wouldn't prevent sharing it with samba, because samba is in userland too.