On Thursday July 9, tmarri@xxxxxxxx wrote: > > >No. You don't have to way for raid-5 rebuilt to complete. > > >Can you try disabling the ADMA and run exactly the same test. > >If it works without ADMA, but fails with ADMA, then you know where > >the problem is. > > > Working: > 1. Using pure SW RAID without ADMA everything works fine. > 2. Using ADMA normal operations work fine > 3. Using ADMA installing file-system works fine while recover happening. > 4. Using ADMA with forcing raid(-f option to mdadm) and io to > file-system > mounted /dev/md0 while recovery happening works fine. > 5. Using ADMA wait till recovery finishes and then IO to file-system > mounted /dev/md0 works fine. None of these operations actually use the parity data that is created. They update it but would not notice if it was wrong. > > > Not Working: > 5. Using ADMA(without -f option to mdadm) and io(dd command) to > /dev/md0 which is mounted using > ext3 file-system causing file system errors. This creates the array in degraded mode so reads will often use the parity block to reconstruct data. > 6. Using ADMA with forcing (with -f to mdadm) and io to file-system > mounted /dev/md0 while rebuild happening(forced a disk failure) causes > file-system errors. This also involves degraded mode so the parity is needed. > 7. Sam as no-6 except without forcing(with out -f option to mdadm) cause > file-system errors. And this. So either the parity is not being calculated correctly, or the process for reconstructing a block from parity and other data is not working correctly. There is a very good chance that it is both as they are essentially the same operation. NeilBrown -- 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
