File corruptions using NFS
 

This is a long shot, but I'm not sure where else to look. I'm hoping that someone here might have noticed this before.

I've been doing system backups of one of my servers weekly for about a year now using tar and gunzip. The file is then saved off onto an NFS share which is on a SAN. The problem that I just noticed is that about 40% of backup jobs are corrupt. When I try to un-gzip them it says that they are not recognized as gzip files.

I'm sure that it's something to do with NFS because I never once get this problem if I FTP the file to the NAS or if I save the file to a partition on a physical drive. Something in the NFS transfer changes the file because if I make a copy of the good gzip file then copy it to the SAN on the NFS mount I can do a 'diff' and it says that the file is different.

What could this be? Thanks!

What options do you use for the mount? In automounter I use: -rw,hard,intr,rsize=8192,wsize=8192,noatime. Besides this, do you use a hub or a switch?

Moloko, thanks for your response. What I have in my fstab file for that mount is


"192.168.1.2:/backup /mnt/backup nfs rw 0 0"


I'm not near that machine today, but I'm excited to use your options. Do you think that the
reason that I'm having the trouble is because of the way I had it set up?

I'm using a switch to connect the computers to the SAN. They are SMC 1000g switches.

Thanks again. :)

The rsize, wsize might help, but I'm not sure. I haven't experienced any problems with NFS, but it's the only option I can think of right now.

Thanks for the try... That didn't fix my problem. About 40% of the time I move any tar file over to my SAN and bring it back I get this error:


gunzip: testfile.tgz: Invalid data -- format violated


Hum... Not sure what to do at this point...

Are you sure it isn't due to memory corruption (faulty memory banks)?

Moloko,

I don't think it is because the data that is being moved is coming from about 44 other nodes of which all have the same problem. As far as the SAN that is exporting the NFS we don't see this problem on anything else. We're starting to think it might be an issue with the SAN client software, but not sure yet. :) Thanks for the continuted input!