Hello,

I have some systems that do video recording, with 4 disks each. 10% of
the disk is used for the base operating system (/), all mirrored in
software RAID. The rest of the space is mounted under 4 mountpoints and
unified under a rw meta mountpoint with unionfs. Something like this:

Filesystem            Size  Used Avail Use% Mounted on
/dev/sda3             275G  -   -       - /mnt/sda3
/dev/sdb3             275G  -   -       - /mnt/sdb3
/dev/sdc3             275G  -   -       - /mnt/sdc3
/dev/sdd3             275G  -   -       - /mnt/sdd3
unionfs               1100G -   1100G   - /storage

The application used for video recording produces big chunks of data,
this is why I need so much common storage. 

The problem is that when the first disk on the unionfs mount fills
(sda3) the application does not continue to write on the following disks
from the union, instead it crashes, being unable to work with no space
left on device (although the rest of the disks in the union are empty). 

Is there a way to have the common space used all at once by different
applications? It would be great if unionfs were able to detect when the
disk is almost full and continue n the remaining disks/space, or write
balanced info, on all the disks.

I'd rather use unionfs than RAID because all the data is not very
important and I can afford to have a bad disk replaced and loose all the
information on it, but it is not acceptable to have the whole RAID
matrix lost (and experience has proved this can happen, often, even with
RAID5).

Thank you.

Dragos 

_______________________________________________
unionfs mailing list
[email protected]
http://www.fsl.cs.sunysb.edu/mailman/listinfo/unionfs

Reply via email to