Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

An SSD just over a month old is producing read errors in my syslog whenever rsync backs it up:

[276877.360221] ata1.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[276877.360226] ata1.00: irq_stat 0x40000008
[276877.360229] ata1.00: failed command: READ FPDMA QUEUED
[276877.360234] ata1.00: cmd 60/10:00:78:a2:27/00:00:07:00:00/40 tag 0 ncq 8192 in
[276877.360234]          res 41/40:00:78:a2:27/00:00:07:00:00/40 Emask 0x409 (media error) <F>
[276877.360238] ata1.00: status: { DRDY ERR }
[276877.360240] ata1.00: error: { UNC }
[276877.360686] ata1.00: configured for UDMA/100
[276877.360697] sd 0:0:0:0: [sda] Unhandled sense code
[276877.360699] sd 0:0:0:0: [sda]  
[276877.360701] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[276877.360703] sd 0:0:0:0: [sda]  
[276877.360705] Sense Key : Medium Error [current] [descriptor]
[276877.360708] Descriptor sense data with sense descriptors (in hex):
[276877.360709]         72 03 11 04 00 00 00 0c 00 0a 80 00 00 00 00 00 
[276877.360717]         07 27 a2 78 
[276877.360721] sd 0:0:0:0: [sda]  
[276877.360724] Add. Sense: Unrecovered read error - auto reallocate failed
[276877.360727] sd 0:0:0:0: [sda] CDB: 
[276877.360728] Read(10): 28 00 07 27 a2 78 00 00 10 00
[276877.360735] end_request: I/O error, dev sda, sector 120038008
[276877.360750] ata1: EH complete
[276877.361389] ata1.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[276877.361392] ata1.00: irq_stat 0x40000008
[276877.361395] ata1.00: failed command: READ FPDMA QUEUED
[276877.361400] ata1.00: cmd 60/08:00:78:a2:27/00:00:07:00:00/40 tag 0 ncq 4096 in
[276877.361400]          res 41/40:00:78:a2:27/00:00:07:00:00/40 Emask 0x409 (media error) <F>
[276877.361403] ata1.00: status: { DRDY ERR }
[276877.361405] ata1.00: error: { UNC }
[276877.361841] ata1.00: configured for UDMA/100
[276877.361847] sd 0:0:0:0: [sda] Unhandled sense code
[276877.361850] sd 0:0:0:0: [sda]  
[276877.361851] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[276877.361853] sd 0:0:0:0: [sda]  
[276877.361855] Sense Key : Medium Error [current] [descriptor]
[276877.361857] Descriptor sense data with sense descriptors (in hex):
[276877.361858]         72 03 11 04 00 00 00 0c 00 0a 80 00 00 00 00 00 
[276877.361867]         07 27 a2 78 
[276877.361871] sd 0:0:0:0: [sda]  
[276877.361873] Add. Sense: Unrecovered read error - auto reallocate failed
[276877.361875] sd 0:0:0:0: [sda] CDB: 
[276877.361876] Read(10): 28 00 07 27 a2 78 00 00 08 00
[276877.361883] end_request: I/O error, dev sda, sector 120038008
[276877.361893] ata1: EH complete
[276880.336132] ata1.00: exception Emask 0x0 SAct 0x1 SErr 0x0 action 0x0
[276880.336140] ata1.00: irq_stat 0x40000008
[276880.336147] ata1.00: failed command: READ FPDMA QUEUED
[276880.336157] ata1.00: cmd 60/08:00:78:a2:27/00:00:07:00:00/40 tag 0 ncq 4096 in
[276880.336157]          res 41/40:00:78:a2:27/00:00:07:00:00/40 Emask 0x409 (media error) <F>
[276880.336163] ata1.00: status: { DRDY ERR }
[276880.336167] ata1.00: error: { UNC }
[276880.336660] ata1.00: configured for UDMA/100
[276880.336670] sd 0:0:0:0: [sda] Unhandled sense code
[276880.336672] sd 0:0:0:0: [sda]  
[276880.336674] Result: hostbyte=DID_OK driverbyte=DRIVER_SENSE
[276880.336675] sd 0:0:0:0: [sda]  
[276880.336677] Sense Key : Medium Error [current] [descriptor]
[276880.336679] Descriptor sense data with sense descriptors (in hex):
[276880.336680]         72 03 11 04 00 00 00 0c 00 0a 80 00 00 00 00 00 
[276880.336687]         07 27 a2 78 
[276880.336690] sd 0:0:0:0: [sda]  
[276880.336692] Add. Sense: Unrecovered read error - auto reallocate failed
[276880.336694] sd 0:0:0:0: [sda] CDB: 
[276880.336695] Read(10): 28 00 07 27 a2 78 00 00 08 00
[276880.336701] end_request: I/O error, dev sda, sector 120038008
[276880.336709] ata1: EH complete

Other than this, the disk seems to be working fine. Is this a sign of a failing disk, or just harmless debugging information?

share|improve this question
    
Please provide more detail. Brand/model of SSD, which SATA chipset, which version of Linux kernel are you using ? –  Tonny Apr 6 '13 at 8:36
    
Kernel is Linux 3.5.0-26-generic on x64. Disk is ADATA ASP600S3-64GM-C, SATA III. Mainboard is MSI 970A-G46, that has AMD chipsets, not sure how to tell which SATA chipset specifically. Other disks are working fine. –  Rena Apr 6 '13 at 8:44
    
Adata... I wouldn't touch these with a 10-foot pole... They have a very bad reputation (RAM and SSD's). Follow Mike Sherill's suggestion below and get it replaced, or better, a refund. –  Tonny Apr 6 '13 at 13:10

3 Answers 3

up vote 4 down vote accepted

It's dead, Jim.

Replace the offending hardware.

Of course, since you're using a low-end desktop motherboard (DEAR GOD WHY?) you might spend a moment testing whether the drive controller or cable has failed.

share|improve this answer

On the one hand, I'd probably try to dig into the drive's documentation first. On the other hand, I'd probably consider an error message that said "Unrecovered read error - auto reallocate failed" as indicating the drive should be considered guilty until proven innocent. And I'd get the data off it immediately.

Now, at only a month old, it should still be under warranty. The quickest way forward might be to just try to return it for either replacement or refund using these unrecovered read errors as the justification.

The manufacturer might replace the drive regardless of whether the errors are significant--keep the customers happy. But if it's a common error, and it's not significant, they'll probably tell you. Otherwise, they risk just cycling inventory to every buyer on the market. (Every buyer who's a sysadmin, anyway.)

It could be that the problem is related to using a consumer-grade drive in a server. (According to the Adata web site, your drive isn't a server-grade drive.) I'd say you have at least three small projects in the immediate future.

  1. Get the data off that drive.
  2. Get a refund or replacement, preferably a refund.
  3. Develop a policy that makes it easier to make better decisions about buying server hardware.
share|improve this answer
    
Go for a refund in stead of replacement. Adata is cheap, but you get what you pay for. Quality is not the best in the world... –  Tonny Apr 6 '13 at 13:12
    
@Tonny: Good point. Expanded my answer to include that. –  Mike Sherrill 'Cat Recall' Apr 6 '13 at 15:16

The drive lost your data, you could ofcourse replace it but if you just want to continue using the drive (at risk of future data loss again) you can try to write zeros in that location. This is what a RAID setup would do for you, it would find it can't read, zero out the place and recover the data from the RAID structure and then rewrite it back into the bad place.

If you use the device on its own, the above recommendations to replace it with or without warranty are very sensible though. If it lost you data once, it will most likely do it again.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.