I have been migrating data off my old drivepool server; first using rsync and then using teracopy.  With both methods I found that even though they passed checksum validation, after copying if I ran md5sum on unraid files would be wrong.

 

I disabled all of docker and it believe it's been fixed.  I found two bad mappings....

 

Unmanic had its cache mounted to /mnt/cache instead of /mnt/user/appdata (which has primary storage of cache).

 

Postgresql had database storage mounted to /mnt/disk1/appdata/postgresql13 instead of /mnt/appdata/postgresql13.

 

I know that writes between shares and disks is bad, and my cache is btrfs.

Can someone confirm (or deny) that this is a likely culprit?  I'm going one directory at a time now and validating each one, but would like a second opinion to help me rest easy or keep looking for a culprit.

 

Thanks,

C

The mappings should not cause data corruption.    The most likely issue is RAM problems.

Oh bummer, well thanks for the information.  I"ll run memtest on it tonight.

I ran 4 passes... no errors from memtest.  Now I have to put this issue on hold to figure out why I can no longer connect to SMB shares. "chdir_current_service: vfs_ChDir(/mnt/user/test) failed: Permission denied. Current token: uid=1000, gid=100, 4 groups: 100 3003 3004 3006".