Had a circuit breaker kick and my older unRaid server had my very first drive show up with a red ball so I thought no big deal. Did a preclear on a another 2tb drive and stuck it in the system and started a data rebuild. Got a few hours into it and it stopped with 288 errors on the new precleared drive. Still have the drive that I took out of the system with that data so does someone have suggestions on how to approach my data recovery dilemma?

 

Thanks

Had a circuit breaker kick and my older unRaid server had my very first drive show up with a red ball so I thought no big deal. Did a preclear on a another 2tb drive and stuck it in the system and started a data rebuild. Got a few hours into it and it stopped with 288 errors on the new precleared drive. Still have the drive that I took out of the system with that data so does someone have suggestions on how to approach my data recovery dilemma?

 

Thanks

post a system log.

 

Were the errors "read" errors? (shown on the unRAID web-console?)  Were the errors on the drive being replaced? or a different drive in the array?

 

It could as easily be a power supply issue, a cable issue, a memory issue, or a disk controller issue. 

Did you get a SMART report on the failed drive?  Does it respond?

Did the circuit-breaker take out the power supply?  (or the opposite)

 

What version of unRAID are you running?  How many disks?  Which disk showed "red"?  (parity or data?)

 

Whatever you do, do NOT set a new initial disk configuration and DO NOT press the button labeled as "restore" on older releases.

 

Joe L.

Joe L.

This system is running 4.5 in a Norco 4220 with 18 data drives and 1 parity drive. I have always seen it do a parity check when it loses power like it did this weekend and when it finished there were no sync errors but 1 data drive had a red ball next to it. I did a preclear in my newer system and replaced the affected drive. I started a data rebuild and checked later and another red ball was there. The system is running fine and I can access all files. Attached is my syslog.

syslog-2010-12-20_Tower.zip

syslog-2010-12-20_Tower.zip

What power supply are you using?

 

I see from the syslog that one of the disks starts out as /dev/sdo, then gets reset as the disk controller attempts to re-establish comminucations with the disk.

 

Here is the disk initially detected.

Dec 20 13:13:17 Tower kernel: ata16: SATA link up 3.0 Gbps (SStatus 123 SControl 300)

Dec 20 13:13:17 Tower kernel: ata16.00: ATA-8: SAMSUNG HD203WI, 1AN10002, max UDMA/133

Dec 20 13:13:17 Tower kernel: ata16.00: 3907029168 sectors, multi 0: LBA48 NCQ (depth 31/32)

Dec 20 13:13:17 Tower kernel: ata16.00: configured for UDMA/133

Dec 20 13:13:17 Tower kernel: ata16: exception Emask 0x10 SAct 0x0 SErr 0x0 action 0x9 t4

Dec 20 13:13:17 Tower kernel: ata16: edma_err_cause=00000020 pp_flags=00000000, SError=00190002

Dec 20 13:13:17 Tower kernel: ata16.00: configured for UDMA/133

Dec 20 13:13:17 Tower kernel: sd 16:0:0:0: [sdo] Unhandled sense code

Dec 20 13:13:17 Tower kernel: sd 16:0:0:0: [sdo] Result: hostbyte=0x00 driverbyte=0x08

Dec 20 13:13:17 Tower kernel: sd 16:0:0:0: [sdo] Sense Key : 0x4 [current] [descriptor]

Dec 20 13:13:17 Tower kernel: Descriptor sense data with sense descriptors (in hex):

Dec 20 13:13:17 Tower kernel:        72 04 00 00 00 00 00 0c 00 0a 80 00 00 00 00 00

Dec 20 13:13:17 Tower kernel:        0c 8a c9 87

Dec 20 13:13:17 Tower kernel: sd 16:0:0:0: [sdo] ASC=0x0 ASCQ=0x0

Dec 20 13:13:17 Tower kernel: end_request: I/O error, dev sdo, sector 210422151

Dec 20 13:13:17 Tower kernel: ata16: EH complete

Dec 20 13:13:17 Tower kernel: ata16.00: detaching (SCSI 16:0:0:0)

Dec 20 13:13:17 Tower kernel: sd 16:0:0:0: [sdo] Synchronizing SCSI cache

Dec 20 13:13:17 Tower kernel: sd 16:0:0:0: [sdo] Stopping disk

 

Several disks get reset.

Dec 20 13:13:12 Tower kernel: ata13: SATA link down (SStatus 0 SControl 300)

Dec 20 13:13:12 Tower kernel: ata13.00: link offline, clearing class 1 to NONE

Dec 20 13:13:12 Tower kernel: ata13: limiting SATA link speed to 1.5 Gbps

Dec 20 13:13:12 Tower kernel: ata14: SATA link down (SStatus 0 SControl 300)

Dec 20 13:13:12 Tower kernel: ata14.00: link offline, clearing class 1 to NONE

Dec 20 13:13:12 Tower kernel: ata14: limiting SATA link speed to 1.5 Gbps

Dec 20 13:13:12 Tower kernel: ata16: SATA link down (SStatus 0 SControl 310)

Dec 20 13:13:12 Tower kernel: ata16.00: link offline, clearing class 1 to NONE

Dec 20 13:13:12 Tower kernel: ata16.00: disabled

Dec 20 13:13:12 Tower kernel: ata15: hard resetting link

Dec 20 13:13:16 Tower kernel: ata15: SATA link up 3.0 Gbps (SStatus 123 SControl 300)

Dec 20 13:13:16 Tower kernel: ata15.00: configured for UDMA/133

Dec 20 13:13:16 Tower kernel: ata15: exception Emask 0x10 SAct 0x0 SErr 0x0 action 0x1 t4

Dec 20 13:13:16 Tower kernel: ata15: edma_err_cause=00000020 pp_flags=00000000, SError=00190002

Dec 20 13:13:16 Tower kernel: ata15.00: configured for UDMA/133

Dec 20 13:13:16 Tower kernel: ata15: EH complete

Dec 20 13:13:17 Tower kernel: ata14: hard resetting link

Dec 20 13:13:17 Tower kernel: ata13: hard resetting link

Dec 20 13:13:17 Tower kernel: ata16: hard resetting link

 

Here it is after it gets reset... but as /dev/sdu.

Dec 20 13:13:18 Tower kernel: scsi 16:0:0:0: Direct-Access    ATA      SAMSUNG HD203WI  1AN1 PQ: 0 ANSI: 5

Dec 20 13:13:18 Tower kernel: sd 16:0:0:0: [sdu] 3907029168 512-byte logical blocks: (2.00 TB/1.81 TiB)

Dec 20 13:13:18 Tower kernel: sd 16:0:0:0: [sdu] Write Protect is off

Dec 20 13:13:18 Tower kernel: sd 16:0:0:0: [sdu] Mode Sense: 00 3a 00 00

Dec 20 13:13:18 Tower kernel: sd 16:0:0:0: [sdu] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA

Dec 20 13:13:19 Tower kernel:  sdu:

 

I'd be more suspicious of the power supply and/or disk controller than the disks.

I am using 2 Supermicro AOC-SAT2-MV8 cards on a ASUS P5E-VM DO motherboard with a OCZ 750 watt power supply. I would say it isn't the power supply due to it being on a backplane with 3 other drives on the Norco. I have 1 sata port open on the motherboard and one on the the Supermicro. Should I move that drive to one of the open sata ports and restart the system and see if it will do a complete data rebuild? Do I need to clear the drive off before I do that?

 

Thanks

Hate to bump a thread but I could really use some help as I don't want to screw up a drive.

 

Thanks

I only see one 750W power supply on OCZ's site: the Fatal1ty 750. It's a quad-rail design with four 18 amp rails on +12V. Without knowing how those circuits are designed, you could be loading all your drives into only a couple of those rails, which means you're only pushing 36 amps of current, which would be JUST adequate for 18 green drives. A lot of these multi-rail PSUs are designed for the needs of the gaming community, and anything with "Fatal1ty" on it is bound to be such, since he's a world champion gamer. So one or two rails might be dedicated to the PCI-E slots, with the rest split between the ATX, molex and SATA connectors.

 

In short, it's a pretty bad design for use in an unRAID box, where it's best to get a single rail PSU. No mysteries there. What you see is what you get.

 

Also, it seems like you're implying that you've had multiple power failures, and seen the ugliness that ensues in unRAID, and you still don't have a UPS.

 

I'm not clear about your situation: did you get a red ball next to the same drive, or a different drive?

I only see one 750W power supply on OCZ's site: the Fatal1ty 750. It's a quad-rail design with four 18 amp rails on +12V. Without knowing how those circuits are designed, you could be loading all your drives into only a couple of those rails, which means you're only pushing 36 amps of current, which would be JUST adequate for 18 green drives. A lot of these multi-rail PSUs are designed for the needs of the gaming community, and anything with "Fatal1ty" on it is bound to be such, since he's a world champion gamer. So one or two rails might be dedicated to the PCI-E slots, with the rest split between the ATX, molex and SATA connectors.

 

In short, it's a pretty bad design for use in an unRAID box, where it's best to get a single rail PSU. No mysteries there. What you see is what you get.

 

Also, it seems like you're implying that you've had multiple power failures, and seen the ugliness that ensues in unRAID, and you still don't have a UPS.

 

I'm not clear about your situation: did you get a red ball next to the same drive, or a different drive?

According to one hardware review site, the 4 rails are allocated as follows:

 

    * +12V1 (solid yellow wires): Main motherboard cable.

    * +12V2 (yellow/red wires): Half of the ATX12V/EPS12V connector.

    * +12V3 (yellow/green wires): Half of the ATX12V/EPS12V connector, half of the modular connectors (two for video cards and two for SATA/peripheral).

    * +12V4 (yellow/black wires): EPS12V connector, half of the modular connectors (two for video cards and two for SATA/peripheral).

 

So, the 12v3 and 12v4 each have two SATA connectors, but the 12v3 also powers one half of the ATX12V CPU power connector.

 

Depending on your cabling, it could easily be overloading the power supply, and easily explaining the hardware resets we are seeing in the syslog.  You probably have, at best, 30 amps of capacity for the disks... if you connected everything optimally.

This server has been bulletproof for almost 2 years now with no issues at all. This is my first issue and now I am just trying to figure out how to approach this and get my data back.

 

I can throw a second power supply in the mix and make sure that it will not be short of power if I have some idea as how to get it rebuilt. So, if it is not the power supply do I go with a clean drive on a different sata port or leave everything the same and see if it will rebuild? Do I preclear another drive and start with that? I can solve the power issue but I need some idea as the best approach as how to rebuild the data.

 

Thanks

The power supply could be failing. Heavily loading the 12V rails could be stressing the power supply sections and causing the capacitors to degrade. The basic result is a reduction in the power supply capability as they degrade.

 

You could move that drive to the motherboard SATA port and try another rebuild and see what happens.

 

Peter

 

 

Thanks, preclearing another drive now and as soon as it is finished I will start another rebuild. Drives will be split between two power supplies and I have a CORSAIR CMPSU-650TX 650W on its way to replace this one.