On Monday February 25, matti.aarnio@zmailer.org wrote: > On Mon, Feb 25, 2002 at 08:58:17AM +1100, Neil Brown wrote: > > On Saturday February 23, asmith@umdgrb.umd.edu wrote: > > > > Not very.... but what filesystem are you usinf over your RAID5 array? > > > > XFS? > > > I use ext3. Do you know what causes these errors? > > > > Yes. > > The message (it isn't really an error) indicated that the raid5 layer > > got a request for a sector when it already had a pending request for > > that sector. This tends to suggest that something odd is going on > > with caching somewhere higher up. Normally if a particular block on > > disc had been requested and it was wanted again, the fact that a > > request was already pending would be noticed and it would not be > > repeated. > > EXT3 (and EXT2) do speed-optimization for "read ahead", which > works by asking for disk blocks to be read _without_ waiting > for it. IF the data read becomes real (vs. speculative) > soon enough (fast reader), the FS code will request the block > again, and then wait for it. But this "read ahead" is done through a cache, either the page cage (in 2.4) or the buffer cache (in 2.2). The cache may well receive two read requests for a particular block, but only would get passed down to the block layer. NeilBrown - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
