I have experienced a number of drive issues with my server lately. The most recent is I will get a bunch of I/O buffer errors on a dive and then when I stop the array unraid marks it as missing. Then I moved the drive to another SATA control card and the system is stable for a while and then the problem happens again. I also have recently had a brand new 2 TB drive fail within a month period as well as a whole bunch of reiserfsck errors with other drives that get corrected when Ido a reiserfsck check on them. After a rebuild is done, everything is good for a week and then something else crops up. I am running out of ideas here and feel like I am chasing my tail with this server. Any ideas guys? I am stumped.

 

Here is a small sample of the System Log now.

 

I am running unraid 5.0 rc12a with 12 drives and and cache drive. Most drives are 750 Gigs but I have a 2 TB parity and a couple of 2 TBs in the array.

 

Nov 19 04:40:02 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki last message repeated 265 times

Nov 19 04:40:05 Loki kernel: quiet_error: 408 callbacks suppressed

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki kernel: Buffer I/O error on device md1, logical block 153589646

Nov 19 04:40:05 Loki shfs/user: shfs_read: read: (5) Input/output error

Nov 19 04:40:10 Loki last message repeated 393 times

Post a full syslog so people know what your hardware is.

if you suspect the drive, disable spindown of the drive.

do a smart long test with smartctl.

Wait the appropriate time and then dump the smartctl log.

 

If you pass the long test chances are the drive is OK.

After that it's the cable, controller or driver.

 

I don't see any sector I/O errors, so I'm thinking there may be a driver issue with the controller.

 

I am running unraid 5.0 rc12a

I would suggest you upgrade to the most current stable version.

This could be a kernel/driver issue with that particular version.

Cool I will give the  smart long test a try when I get home and report  both logs at that time.

Ok So Here is the long smart test report for that drive. I think everything looks ok based on reading the wiki, but if someone could double check me I would fell better.

 

I plan on updating to the release of 5.0 over the weekend. So.....fingers crossed.

smart.txt

The drive looks OK according to the long test.

However there are some UNC errors, which may have caused a timeout during a read.

If you see more of these showing up, perhaps copy your data off the drive and do a preclear.

 

Another choice is to do a multi-pass badblocks readonly test.

This will exercise each sector in a read only fashion (as does a full parity check).

With badblocks you can have it do a read test multiple times to exercise the drive.

 

On one of my older servers, the constant jittering of the head would cause the SATA cable to loosen up thus causing drives to go offline that were known to be good. I think I replaced some cables and swapped some drive positions.

Thanks for looking at that. How would I go about running the "multi-pass badblocks readonly" test on that drive? Is this part a different SMART test?

Sorry I had not heard about that one before.

 

 

Yeah I am thinking I need to change out all my SATA cables too. Things in the case are kind of cramped and they are the cheap cables (No Locking). So with about 14 drives spinning there may be vibrations causing loose cable. Also they are about 7 years old.