blk_update_request: I/O error issue

[zenmak]

Hi All,

 

I recently got a nice supermicro server from a friend of mine with 8 bays.. I bought 8 SAS 2TB drives, stuck them in there and ran 3 cycles of preclear on the drives (they were refurbs), they all passed with no issues.  I also have a separate DAS case with a bunch of drives from the original server in it that I connected externally and they are all working fine.  I've had the "new" system up for a couple weeks now and no issues really until the other day.  

 

I got a notification on my phone that there was an issue with the array with two of the SAS drives.. The 8 drives are connected with their own seperate HBA LSI adaptor (latest FW).  There was nothing on the drives btw, but I had downloaded the SAS plugin to allow the drives to go to sleep.  I had spun down the drives manually, and they were sitting there for a few days asleep with no issues.  All of the sudden the other night they woke up with these errors on them.. I ran SMART against the drive and one appeared to have failed while the other said it was OK.  I took them out of the array and rebuilt the array without them since these drives were empty, I didn't lose anything. 

 

I ran another preclear on them and the preclear came out fine with no issues... I was doing some reading on here and seen that maybe it was the because the drives were spun down I had this issue?  I purposely haven't spun down anything since.. Normally, I wouldn't spin anything down, but since I wasn't using them for anything, I had.

 

Here are the error messages in question:

Aug 22 04:31:28 TowerOne kernel: blk_update_request: I/O error, dev sds, sector 1953531096 op 0x0:(READ) flags 0x0 phys_seg 1 prio class 0
Aug 22 04:31:28 TowerOne kernel: sd 11:0:5:0: [sds] tag#2453 UNKNOWN(0x2003) Result: hostbyte=0x00 driverbyte=0x08 cmd_age=0s
Aug 22 04:31:28 TowerOne kernel: sd 11:0:5:0: [sds] tag#2453 Sense Key : 0x2 [current]
Aug 22 04:31:28 TowerOne kernel: sd 11:0:5:0: [sds] tag#2453 ASC=0x4 ASCQ=0x2
Aug 22 04:31:28 TowerOne kernel: sd 11:0:5:0: [sds] tag#2453 CDB: opcode=0x2a 2a 00 74 70 84 d8 00 00 08 00
Aug 22 04:31:28 TowerOne kernel: blk_update_request: I/O error, dev sds, sector 1953531096 op 0x1:(WRITE) flags 0x0 phys_seg 1 prio class 0

 

I'm attaching my diagnostics as well.

 

On a side question about rebuilding the array/parity.. If when I removed these drives from the array with "new config", could I have selected "parity is valid" already if removing the drives? or does it have to rebuild the array?  I wasn't sure, so I rebuilt the parity on the array to be safe.

 

Thanks!

 

Jacktowerone-diagnostics-20210823-1923.zip

[itimpi]

6 hours ago, zenmak said:

On a side question about rebuilding the array/parity.. If when I removed these drives from the array with "new config", could I have selected "parity is valid" already if removing the drives? or does it have to rebuild the array?  I wasn't sure, so I rebuilt the parity on the array to be safe.

You can only keep parity valid if the drive you are removing is all zeroes.    This would not be true if you had ever formatted the drive after adding it to the array.