warmflatsprite

Hello,

I work for a small IT company that runs a CVS server on Windows XP that's formatted to NTFS. Since we started with this server configuration (about 6 months ago) it has corrupted four drives. At first we thought it was hardware issues, so we've swapped out drives and eventually the machine, but over the course of last weekend, it happened again. We're beginning to look into software issues. As far as I know, CVS doesn't do any low-level disk access, so that'd lead me to believe that there's some limitation in the interface between the OS and CVS that's causing our discs to crash. The first thing that comes to mind would be directory depth. The files in our repository go to a VERY deep directory structure, and there's over 22,000 files stored in the repository. Could this be the source of our frustrations? Is there anything else we should take a look at?

The corruption is bad enough that a full NTFS rollback was not possible, and each time we lost changes to our codebase. Needless to say, any help would be GREATLY appreciated.

Thanks,
Ben

albybum

http://www.windowsforumz.com/Securit...ict545082.html

---

Geffy

are you using a dedicated machine to host the cvs server? an alternative if you cant get NTFS to host it stabley then try either Unix or Linux. I havent tried to run cvs on anything I have always used SVN which I run on FreeBSD servers.

how much space does the whole repository take up? how many files are there? what is the current deepest path? are you syncing this with other systems running on NTFS? does it corrupt right after a commit or update? during a commit or update? or just seemingly randomly during the night when no one is really using it?

---