January 7, 201412 yr I've built a new server with 5 3TB drives, and added 4 2TB drives later. But my parity drive is showing occasional Errors on the Main screen. I've precleared all the drives, but don't remember any before or after stats. I filled the 4 drives with data, then enabled the parity drive. I ran a parity check but it found 0 errors. But the syslog is filled with tons of "disk0 read error". The SMART doesn't show any reallocated sectors, but does show some possible "Raw_Read_Error_Rate" and "Multi_Zone_Error_Rate". In the log there's a lot of "ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x0 (Errors)" errors. So it looks like they're all coming from the disk0 parity drive. I've tried 3 different SATA cables, 2 different onboard SATA ports, and only the parity drive powered on/plugged in. After the parity check, there were 605 Errors displayed on the Main page, but yet 0 parity errors: Device Identification Temp. Size Used Free Reads Writes Errors View [spin Down] Parity WDC_WD30EFRX-68EUZN0_WD-**********60 (sdb) 2930266532 72 °F 3 TB - - 128,406,159 32,564 605 Last checked on Thu Jan 2 06:58:55 2014 PST (yesterday), finding 0 errors. > Duration: 7 hours, 55 minutes, 37 seconds. Average speed: 105.1 MB/sec No bad sectors, but could it be a bad controller on the drive, or bad drive memory chips? Below are my logs. syslog.zip smart_test_-_short.zip smart_test_-_long.zip HD_Sentinel.zip Edited January 22, 20224 yr by Pro-289
January 8, 201412 yr Author How is the parity drive connected? What are the complete system specs? System: ASRock - H55M/USB3. CPU: Intel® Core i3 CPU 530 @ 2.93GHz - 2.933 GHz Cache: 128 kB, 512 kB, 4096 kB Memory: 4096 Module (max. 8 GB) Network: eth0: 100Mb/s - Half Duplex (What's up with this? It's a Gigabit LAN in a Gigabit router.) 5 3TB HDs in onboard SATA 3.0Gbps, Parity one of them 4 2TB HDs in PCI-E IOCrest (SI-PEX40064) SATA expansion card I believe it's got a 500W power supply.
January 9, 201412 yr The pre-clear reports should be in /boot/preclear_reports. Running pre-clear again may be helpful.
January 11, 201412 yr Author The pre-clear reports should be in /boot/preclear_reports. Running pre-clear again may be helpful. Well the pre-clear logs show nothing, the values are just zero, because the drive was brand new. The raw_read value was 100 at start, then went back to 200 at finish. The worst was 253 then went to 200 at finish. Some of the same sector areas come up each time I do a parity check. But I end up having 0 parity errors. So it's hard to believe if these warnings are critical or not. UnRAID calls them errors, but the data is still okay. This is driving me nuts. I just wanted to build this thing and be done with it.
January 11, 201412 yr The SMART report shows that the drive failed the long SMART test with a read failure. Running pre-clear may correct the read problems. If the drive continues to fail SMART tests it needs to be RMA'd.
January 19, 201412 yr Author Yea, I've found some bad spots on the drive that just aren't being remapped through SMART. I ran Spinrite to check for bad sectors, and sure enough around 720400000, the same area in the syslog, brought Spinrite to a halt. I know my drive is not fully compatible with Spinrite, it only found the drive as a 2.2TB instead of a 3TB, but it was able to get to a bad spot at 16.77% in. I even tried a level 4 test to read and write each sector, but it just froze when it got to the bad spot. I also tried the jumpers on the hard drive to put it in 3Gbps mode and spread spectrum mode. When I had it in IDE mode I was getting the error "This sector contains data requiring recovery using a deeper testing level", and it never kicked in to Dynastat recovery mode. In IDE or AHCI mode it just eventually quit with a red error box. I've seen the same error in the past with other bad drives that Spinrite was able to recognize. With all these reads and writes using Spinrite, parity check, and normal parity use never kick the bad sectors into the spare area. So I doubt running another preclear would flag the bad sectors if everything else hasn't yet. I think it's off to RMA I go.
Archived
This topic is now archived and is closed to further replies.