>Hi Al,
>
>Does the Debian system use the Linux ext3 file system as its base. Red Hat uses the ext3 file system with journaling. This essentially eliminate the problem of file corruption. Even when a plug is pulled on a linux box, files should not become corrupted, and the time needed to perform a fsck when rebooting a system that was not normally shutdown is no longer necessary. I have been running a Red Hat samba share to serve a VFP 7 application. The VFP 7 tables are in a DBC. The app uses buffering 5 or no buffering depending on the users selection when starting the app. It also uses persistent relations, referential integrity checking and transactions. For me samba has performed its duties without any problems.
>
>In the case of your client do you know what was causing the corruption like maybe the connection was timing out leaving the user dangling, or maybe the user was shuting off their computer without doing a normal exit from the app, etc.

The specific problem I quoted was definitely caused by some sort of "point upgrade" of Samba - nothing else about their environment changed. I think it was a problem with that particular release - they went back to the previous one and the problem went away, and I believe they later upgraded about 3 point versions and that didn't have any problems either.

I'm not a Linux admin or user but I expect the main production box is not using a journaled file system because it takes about 25 minutes to reboot after an improper shutdown.

For some reason my client's sysadmin is switching the boxes from Debian to OpenBSD. The client is not *that* security conscious, and I was under the impression OpenBSD is pretty minimalist so I have no idea what the reasoning is behind that move.