Thiago Damas <[email protected]> writes: > I tested now with 1Mb (2048 blocks) at the begining of disk, and same > behaviour: slow write speed and high disk latency.
It just occurred to me - if I'm not mistaken, when the disk block size is larger than either DFLTPHYS (64 kB), the kernel will do a read-modify-write no matter how much data you write. You should make sure the filesystem never sees a block size larger than that (e.g. with a four-disk stripe or a five-disk raid5, the individual disks' block size should not exceed DFLTPHYS / 4, because the array's block size will be 4 x the LCM of the individual disks' block sizes) DES -- Dag-Erling Smørgrav - [email protected] _______________________________________________ [email protected] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[email protected]"
