On Sat, Nov 9, 2013 at 6:13 PM, Jeff Liu <jeff.liu@xxxxxxxxxx> wrote: > Hi Harry, Hi Jeff, > As this inode has been marked to realtime, the subsequent write via > directIO will allocate > extents via it's specific allocation algorithm rather than the general > approach, and the > extent length is specified with fsx_extsize. > AFAICS, the realtime file data will be allocated on realtime volume > while the metadata will > still be allocated from the data volume. > I see. However, I think current real-time feature of XFS should be enhanced and could write/read specific data during specific time interval. I have no exact ideas about this. I would do if I could. Anyway, thanks for your replies. -- Thanks Weiwei Jia (Harry Wei) _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs
