> -----Original Message----- > From: linux-raid-owner@xxxxxxxxxxxxxxx [mailto:linux-raid- > owner@xxxxxxxxxxxxxxx] On Behalf Of Carlos Carvalho > Sent: Saturday, November 19, 2005 2:30 PM > To: linux-raid@xxxxxxxxxxxxxxx > Subject: raid5 reliability (was raid5 write performance) > > Guy (bugzilla@xxxxxxxxxxxxxxxx) wrote on 19 November 2005 00:56: > >Assume a single stripe has data for 2 different files (A and B). A disk > has > >failed. The file system writes a 4K chunk of data to file A. The > parity > >gets updated, but not the data. Or the data gets updated but not the > >parity. The system crashes or power fails. The system recovers, but > can't > >do anything about the parity with a failed disk. > > But this is just the case Neil mentioned: an unclean degraded array. > This is the only case where data corruption can appear. Ok, so Neil hit send 1 minute before me, sorry. :) > > >Also, after a power failure, I have seen the system come back with a > single > >disk failed. I guess that 1 disk had an expired superblock. > > You mean the disk was fine and operating in the array before the power > failure? If so the superblocks were all in sync. If one of the > superblocks was different the array was already degraded and after the > power failure it'll also be dirty, and we're back at the case above. Yes, everything was fine, but while the superblocks were being updated, BANG! Now 1 is out of date. I think on reboot that 1 disk will be failed. But I am not sure of this. I do know I had 1 disk dropped from my array after a loss of power. I did not have any emails telling me a disk had failed. So, I think the power failure led to the disk being dropped from the array. I did not pursue how or why, I just added the disk back. I don't really know if this can lead to data corruption. I guess the stripes would be updated before the superblock updates start. But not sure. But if so, then I don't see how this case can corrupt data. > > >When you add that disk back, any stripes that were not up to date > >will be re-constructed with invalid data. > > If the array was not degraded the reconstruction data will not be > invalid. Only the parity will be rebuilt and no data corruption will > happen. No, if you add a disk, it is re-built 100%. At least with my version. The intent log may resolve this, but I am not sure. > > >I don't know if the intent logging will help here or not. > > It won't change anything, it only helps to accelerate the resync by > not re-writing stripes that were up to date. > - > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html To the best of my knowledge, I have never had corrupt data. Guy - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
