February 9, 200917 yr I have a 1TB samsung spinpoint HD103UJ that is showing some smart errors. I've run the short test now. » reallocated_sector_ct=63 » calibration_retry_count=1 » reported_uncorrect=189 » reallocated_event_count=3 » current_pending_sector=31 » udma_crc_error_count=3 » multi_zone_error_rate=78 » soft_read_error_rate=326 Should I be concerned?
February 9, 200917 yr Short answer: Yes. You have 63 sectors that have been reallocated. This is not bad in and of itself, but if this number is increasing over time, that is bad. Joe L. has a disk with 100 bad sectors that have been there for years and have not increased. So if your 63 bad sectors hold steady at 63 and don't increase even after a few parity checks, it is likely that the drive is not self-destructing. My guess, though, is that the number is going to be increasing. The current pending sectors are sectors that have been found to be bad but have not yet been remapped. Sectors that can't be read are NOT remapped until they are written to. (if you think about it this makes sense). The drive keeps track of these types of sectors and remaps them on the next write. But when unRAID is involved, it will, after a read error, immediately use parity to "recompute" the sector it was trying to read, and will rewrite the sector back to the source disk. That should cause the drive to remap the sector. So the question is why didn't that happen? I don't know the answer, but have seen this before when lots of bad sectors are getting detected. The other values are also consistant with problems, but truthfully I don't monitor those nearly as much as the bad sector numbers. I'd bet that SMART hasn't failed the disk yet, but I'd be prepared for it to go from bad to worse. Running a parity check on the array may not be such a good idea. If the drive is subjected to that level of activity, it might fail. There has been some evidence that failing drives can return some bad data (there is some disagreement here as to whether this is possible). The chances of a parity check helping you at this point is pretty much nil. I would do one of three things ... 1 - Pull the disk from the array and let it run simulating that bad disk until you can RMA it and get a new one. 2 - Leave it in the array a while longer until it gets bad enough to meet the RMA criteria (I've never had to RMA a drive so don't know what evidence they need). 3 - Put in a spare disk and let it reconstruct that disk on the the spare, and worry about getting the disk RMAed afterwards. I'd be leaning towards #3. You could leave the drive in the box and manually mount it with unmenu. You could then run some smartctl long tests on the drive. Hopefully it will fail this test and give you the ammunition you need to RMA it.
February 9, 200917 yr 4. Put in a new drive, copy data from the one going bad to the new one, then remove the old one.
February 9, 200917 yr Author Thanks for the advice...this drives only about 2 months old! Good news all the data on that disk is backed up on a usb drive and stored off site. I have an empty drive in the array I'm going to copy everything to and then RMA the drive. Thought about buying another drive but since I have space I'm waiting for the 2tb drives to drop in price. Again thanks for the input Erik
Archived
This topic is now archived and is closed to further replies.