raid1 running fine for years .. and then

[Justin Piszcz <jpiszcz@lucidpixels.com>]

Hello,

I have 2 disks + 1 spare (3 total) in a raid-1 configuration, over the past
3-4 weeks after 1-2 year of raid-1 (md), this is the second time this has
happened:

[1538654.702201] ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6 frozen
[1538654.702209] ata1.00: cmd ea/00:00:00:00:00/00:00:00:00:00/a0 tag 0
[1538654.702210]          res 40/00:00:3c:4f:c2/00:00:00:00:00/00 Emask 0x4 (timeout)
[1538654.702213] ata1.00: status: { DRDY }
[1538654.702217] ata1: hard resetting link
[1538660.012018] ata1: link is slow to respond, please be patient (ready=0)
[1538664.704019] ata1: COMRESET failed (errno=-16)
[1538664.704024] ata1: hard resetting link
[1538670.058008] ata1: link is slow to respond, please be patient (ready=0)
[1538674.750009] ata1: COMRESET failed (errno=-16)
[1538674.750015] ata1: hard resetting link
[1538680.104031] ata1: link is slow to respond, please be patient (ready=0)
[1538692.344014] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[1538692.348906] ata1.00: configured for UDMA/133
[1538692.348929] ata1: EH complete
[1538692.362731] end_request: I/O error, dev sdb, sector 293041531
[1538692.362737] md: super_written gets error=-5, uptodate=0
[1538692.362741] raid1: Disk failure on sdb3, disabling device.

The smart statistics on the disk all show as OK and the short+long tests
run successfully.

Anyone know what is going on here?

rm --read-sector 293041531 /dev/sdb

/dev/sdb:
reading sector 293041531: succeeded
..
..
< the data / numbers >

The last time this happened, I wrote the disk with zeroes/badblocks/etc 
and failed/removed then re-added it to the array without any issues, now, 
3-4 weeks later, it has happened again.

Justin.