A great problem. Please help

[Alexandro]

OK, here is what happened today.

I was at the local supermicro dealer's premises to see the performance of my own cage. It was attached to a supermicro system, with a fancy (and very expensive) Raid controller. The cage was filled with 5 WD RE4 HDDs. The speed between the HDD's was around 750 MB/sec. and no problems in the log file during transfers. So I took the cage with the new cables which the guys provide me with and ran home to install it at my Unraid.

I put the cables in, doubled checked everything and started the system. I decided to sync parity again and push the Parity Sync button.

While the previous parity sync (with the drives directly attached to the motherboard) was performed with ~ 110 MB/sec. what i get now is something like 35-38 MB/sec.

I explored the syslog and saw some strange warnings:

 

Oct  9 21:16:05 UNRAID kernel: pci_root PNP0A08:00: address space collision: host bridge window [mem 0x000c8000-0x000dffff] conflicts with Video ROM [mem 0x000c0000-0x000cdfff] (Minor Issues)

The SASLP-MV8 controller is serving currently only the cache drive and the data and the parity HDD's are directly connected to the motherboard SATA ports.

I am attaching the syslog here.

Could you please put your thoughts about the situation here.

I can't understand what might be the problem.

 

syslog-2012-10-09.txt

[Alexandro]

This morning DISK1 was disabled (red) with a lot of I/O errors in the log file.

It confirms that the cage is not working reliable with my motherboard.

Probably the sata ports on the cage are not negotiating with the sata ports of the motherboard.

Is there something more i can try or sell the cage or change the motherboard with something better?

Currently I am doing Parity sync with 110 MB/sec with the disks attached directly to the MB.

Please help. I am lost here.

 

[dgaschk]

It looks like a conflict between the MB and the SASLP. Make sure that BIOS on both is current. Otherwise a new MB may the the least expensive option.

[Alexandro]

Thank you very much.

The SASLP-MV8 is with the most current firmware.

I will update the mainboard this evening.

After the experiment (drive cage) I have again a current pending sector on DISK1

It is time to stop experimenting and just to throw out the cage.

Here is a smart report of DISK1

 

SMART.TXT

[dgaschk]

Disk 1 has a pending sector. It needs to be rebuilt. Is parity correct?

[Alexandro]

I dont know. Currently I am doing parity sync after initconfig.

 

Total Size 2,930,266,532 KB

Current 124,272,404 (4.2%)

Speed 102,730 KB/sec

Finish 453 minutes

Sync Errors 105 (corrected)

 

 

[dgaschk]

This may be a problem. Parity cannot be properly build because disk 1 has unreadable sectors and disk 1 cannot be rebuilt because parity is not valid. This is akin to having 2 failed disks. Copy the data off of disk 1.

[Alexandro]

Thank you very much dgaschk being so helpful.

I am on 65% rebuild parity process now and have something like 3 hours to complete (speed decreased to 96 MB/sec).

I will let it finish and if failed i will buy another 3tb HDD to transfer the data from DISK1.

Something very weird is happening here. Is it possible the cage to be the cause of tis bad sectors?

[Alexandro]

Parity was updated  105  times to address sync errors but now I have a green parity and data drive.

I am a bit scared to start a Parity-check.

I ran a short self tesk on Disk1 which ended with a message that "Errors occurred".

Beside this 1 Pending sector i have also a LBA error now which was presented in the SMART.log

Are there any steps that I can take to clean it or I need to RMA this drive too?

SMART.txt