Areca HW controller: Copying from one RAID to another cause RAID array degradation

[Revil]

Hi,
I noticed very strange behavior of my Areca ARC-1880IX-24 RAID controller. Currently I have RAID 6 array with 5 pcs of 4 TB HDDs. I added another array of 6 pcs 1 TB HDDs. After foreground initialization of the new array the array works just fine. I can copy a large amount of data from internal storage (SSD connected to MB) to this array without problem. But when I start copying from the RAID 6 array then the new array became degraded. Most of HDDs or all HDDs in the new array failed in the same time. I can see such events in the log:

Enc#2 PHY#3 Time Out Error
Enc#2 PHY#3 Device Failed
Volume Failed
RaidSet Degraded
Enc#2 PHY#3 Device Removed

Before I noticed that array fails only when I start copying from the current array I tried various RAID configuration (50, 5, 6, 10) with full foreground initialization but nothing change. The new array is going down when I start copying. During last incident the old RAID 6 went down too but after restart it was OK.

Please can you tell me what is wrong?

Thank you very much.
Revil

----------------------
My configuration
RAID Controller
Controller Name ARC-1880IX-24
Firmware Version V1.52 2014-02-07 (newest)
BOOT ROM Version V1.52 2014-02-07
PL Firmware Version 18.0.0.0
Main Processor 800MHz PPC440
CPU ICache Size 32KBytes
CPU DCache Size 32KBytes/Write Back
System Memory 1024MB/800MHz/ECC
PCI-E Link Status 8X/5G

RAID
Current – RAID 6:
- 5x Western Digital WD Red 4TB HDD Review (WD40EFRX)
New – RAID 50 / 6 / 5 / 10:
- 6x Samsung SpinPoint F3 1TB (HD103SJ)

OS
Microsoft Windows Server 2012 R2 – All updates
Hyper-V host
Drivers: 6.20.0.28 and later I upgraded to the newest 6.20.0.29

HW
Intel Core i7 875K (Lynnfield) – No overclocking
Gigabyte GA-P55-UD6
16 GB RAM

 

[mwroobel]

Please post a complete log from the card. What is the temp of your controller/cpu when it starts to fail? Are all the drive connected to the internal ports or are some on external? What kind of power supply do you have (brand/wattage), what other PCIe cards are in the machine and how are the 11 drives attached to the power supply (all on 1 rail, multiple rails, etc?)

 

[houkouonchi]

What chassis/enclosure do you have these new disks in? Going by twhat you pasted (and it being Enc #2) suggests that you must be using some sort of SAS expander. Not all of them have compatability. For example the expanders in supermicro chassis (The LSI 3 gig one) is not compatible with areca controllers and you will get random drives failing and timeouts where as the LSI 6 gig one works just fine.

 

[Revil]

Please post a complete log from the card. What is the temp of your controller/cpu when it starts to fail? Are all the drive connected to the internal ports or are some on external? What kind of power supply do you have (brand/wattage), what other PCIe cards are in the machine and how are the 11 drives attached to the power supply (all on 1 rail, multiple rails, etc?)

Hi mwroobel,
I am sorry for the late response. I did another several tests and full foreground initialization takes 20 hours.

RAID Controller temperature: ~50 C (I have a very good air cooling)
Ports: All internal; Original Areca cables; I tried different ports / cables but no change…
PSU: Corsair Professional Gold AX1200 1200W (I am sure that PSU is OK because the PC is stable during whole initialization)
RAILs: PSU have only one strong 12V and there are 4 HDDs on every cable
PCIe:
- 8x: Sapphire Radeon HD 5670 1GB DDR5 Ultimate
- 8x: Areca
- 1x: Intel PRO/1000 MT Dual Port Server Adapter

Log from the last incident with RAID 50. I tried differen cables and different Areca ports. I got timeout in 30 seconds when I started copying.

---------------------
2014-04-07 15:27:23 Enc#2 PHY#31 Device Failed
2014-04-07 15:27:22 test-raid-1 RaidSet Degraded
2014-04-07 15:27:22 v-0 R50Vol2-2 Volume Failed
2014-04-07 15:27:08 Enc#2 PHY#31 Time Out Error
2014-04-07 15:26:59 Enc#2 PHY#29 Device Failed
2014-04-07 15:26:59 test-raid-1 RaidSet Degraded
2014-04-07 15:26:59 v-0 R50Vol2-2 Volume Degraded
2014-04-07 15:26:45 Enc#2 PHY#29 Time Out Error
2014-04-07 15:12:57 H/W Monitor Raid Powered On
2014-04-07 11:57:09 v-0 R50Vol2-1 Complete Init 020:12:06
2014-04-07 11:57:01 v-0 R50Vol2-2 Complete Init 020:11:58
2014-04-16 15:43:42 v-0 R50Vol2-2 Start Initialize
2014-04-16 15:43:42 v-0 R50Vol2-1 Start Initialize
---------------------

What chassis/enclosure do you have these new disks in? Going by twhat you pasted (and it being Enc #2) suggests that you must be using some sort of SAS expander. Not all of them have compatability. For example the expanders in supermicro chassis (The LSI 3 gig one) is not compatible with areca controllers and you will get random drives failing and timeouts where as the LSI 6 gig one works just fine.

Hi houkouonchi,
no external enclosure. But in administration in “SAS Chip Information” I can see (I suppose there is some enclosure directly attached to the hardware of the controller):

---------------------
Controller:Areca ARC-1880IX-24 1.52
SAS Address (deleted number)
Enclosure ENC#1
Number Of Phys 8
Attached Expander Expander#1[(deleted number) ][8x6G]

Expander#1:Areca ARC-8018-.01.07.0107
SAS Address (deleted number)
Component Vendor LSI
Component ID 0223
Enclosure ENC#2
Number Of Phys 38
Attached Expander Controller[(deleted number) ][8x6G]
---------------------

Thank you very much. Btw. please consider that the other RAID array works just fine. I also tried a small RAID 1 with the problematic Samsung drives. I did not got timeout error from the start but I got it in 10 minutes.
Thank you.

 

[mwroobel]

There have been issues in the past with some Samsung 1TB drives and various Areca controllers. Some of the problems were solved by disabling NCQ on the Areca and some others by forcing the drive mode to SATA150. Unfortunately, Areca themselves have mentioned issues here and here. In the end, this is an enterprise HBA and Areca doesn't guarantee any drive not on their QVL and yours is not.

 

[Revil]

Thank you very much mwroobel. I am currently initializing another RAID 50 with disabled NCQ. Hopefully it will solve the issue.

I know that I am doing crazy thing because I am using server component (Areca) with desktop components (Gigabyte MB) with desktop disks that are not recommended for RAID (no TLER). But I need some temporary storage to test virtual machines and I want to use hardware (Samsung HDDs) that I have at home.