Dan, Thanks for response. I ran the "echo check > /sys/block/md0/md/sync_action" and checked "/sys/block/md0/md/mismatch_cnt" which shows 0. I put a break point on async_xor_zero_sum( als on xxx_prep_adma_zero_sum() adma driver function) to check if indeed called. But during disk fail test I never seen async_xor_zero_sum() being called ? Is some thng I might be missing ? Regards, Marri -----Original Message----- From: dan.j.williams@xxxxxxxxx [mailto:dan.j.williams@xxxxxxxxx] On Behalf Of Dan Williams Sent: Tuesday, January 05, 2010 5:26 PM To: Tirumala Reddy Marri Cc: linux-raid@xxxxxxxxxxxxxxx Subject: Re: Question about ADMA + RAID5 in the event of disk failure On Mon, Jan 4, 2010 at 10:16 AM, Tirumala Reddy Marri <tmarri@xxxxxxxx> wrote: > > Hi Dan, > I have question about what ADMA + RAID-5 and disk failure. In the event of disk failure what would happen to the descriptors which are queued up the DMA engine for XOR operation ? > This is what I am trying. > 1. Create RAID-5 using 4 disks. > 2. Install xfs file system and mount it. > 3. Force the disk failure using mdadm -f /dev/md0 /dev/sdc > 4. Then hot remove the drive mdadm -r /dev/md0 /dev/sdc > 5. Un-mount the /dev/md0. Then run file system check using xfs_repair -L /dev/md0. > > > I see messages as shown below. These messages doesn't show up if I don't use ADMA. But there seem to be no data loss. But still I there seem to be a some kind of data inconsistency. Yes, presumably the adma driver as it does not look like parity is being correctly written to the disks. Does this driver survive extended testing with dmatest? Check that after creating the array you can run "echo check > /sys/block/md0/md/sync_action" and end up with zero errors in /sys/block/md0/md/mismatch_cnt. -- Dan -- 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
