March 5, 201016 yr Hi, I've been using unraid with no problems at all since October. The upgrade to 4.5 final was a smooth transition. Wednesday I decided to increase my array by adding a 1.5T drive. I admit I didn't use Joe's preclear script, instead I let unraid do its work. The drive has been formatted and I decided to perform a parity check before adding any data to it. After the parity check completed it corrected 47 errors (that is probably due to a couple of times the server wasn't gracefully shut down) and the new drive was showing 13.900 errors. I attached the syslog of the parity check (it's huge). I rebooted the server and captured the syslog (attached) as well. I performed a short smart test on the drive and it won't even finish after 7 runs. it either was aborted by the host, results in a read failure or was interrupted by a host failure). Smart test attached as well. Next thing I did was to make a parity verify (not check) and that resulted in like 8.000 errors on the new disk halfway the check. My unraid configuration is as follows: P4 2.4Ghz Asus P4P800 - Intel 865PE chipset Promise TX4 PCI Serial Ata card (4 ports) 1.5T Samsung ecogreen drive as parity 3 x 1.0T Samsung ecogreen drive as data Newly added 1.5T Samsung ecogreen as data The 4 original drives are attached to the Promise card (never had any problems). The new drive is attached to the motherboard header. There is an option in the bios to run the serial ata ports in a compatible or enhanced mode. At the moment it's running in enhanced mode. I will investigate if changing the mode has any effect on the drive. My other options are to change the data and power cable but I wanted to have your opinion on the syslogs and smart test. Maybe you can tell me right away if the drive is not worth keeping and I should return it. Thanks for the advice. syslog-2010-03-04_Part_1.zip syslog-2010-03-04_Part_2.zip
March 5, 201016 yr Author Attached the syslog after reboot and the short smart test syslog-2010-03-04_after_reboot.txt Short_Smart_Test_04032010.txt
March 5, 201016 yr The disk has 382 re-allocated sectors...with another 18 pending re-allocation. RMA it. You need to disable the spin-down timer when running long test, (and during short ones too if the drive is not current spinning) Otherwise you get the "Aborted by host messages" you see when you spin down the disk and (in effect) ask it to terminate the test..
March 5, 201016 yr Author Yeah, I thought I'd have to RMA it. Just for information, what could be the meaning messages underneath? Obviously nothing good, but could you explain me? Mar 4 08:53:24 Tower kernel: ata5.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x0 Mar 4 08:53:24 Tower kernel: ata5.00: BMDMA stat 0x24 Mar 4 08:53:24 Tower kernel: ata5.00: cmd 25/00:00:6f:c2:08/00:04:1b:00:00/e0 tag 0 dma 524288 in Mar 4 08:53:24 Tower kernel: res 51/40:00:09:c4:08/40:00:1b:00:00/e0 Emask 0x9 (media error) Mar 4 08:53:24 Tower kernel: ata5.00: status: { DRDY ERR } Mar 4 08:53:24 Tower kernel: ata5.00: error: { UNC } Mar 4 08:53:24 Tower kernel: ata5.00: configured for UDMA/133 Mar 4 08:53:24 Tower kernel: ata5: EH complete Mar 4 10:39:48 Tower kernel: md: disk4 read error Mar 4 10:39:48 Tower kernel: handle_stripe read error: 472492480/4, count: 1 I have a bios option to enable/disable 32bit transfer. What's your suggestion? Does my syslog after reboot look normal? Thanks for the advice about spindown, hadn't thought of that.
March 5, 201016 yr That UNC error is the disk reporting it could not read a sector. It is probably one of the sectors pending re-allocation when you next write to it. The read-stripe message is just another message further in the process describing the fact the sector could not be read.
Archived
This topic is now archived and is closed to further replies.