Hi all,

 

I had a drive redball on me.  During the rebuild process I hooked up my second 5-in-3 Supermicro bay in order to expand the array when the rebuild completed.  When I plug in the power to that 5-in-3 bay it drops my rebuild speed from about 47 MB/sec to 2 MB/sec.  If I unlug the power, it goes right back.  Any thoughts?  I am using a 1 x SATA to 3 X molex power converter to hook up the bay.

 

I have 5 drives in the system currently (in the first bay), 4 of them green.  The second bay I was trying to hook up contains 1 green drive.

 

My system build is:

 

Motherboard: ECS A885GM-A2 (V1.1) AM3

CPU: AMD Sempron 145 Sargas 2.8GHz Socket AM3 45W Single-Core Desktop Processor SDX145HBGMBOX

PSU: CORSAIR Builder Series CX430 V2 430W

RAM: Patriot 4GB (2 x 2GB) 240-Pin DDR3 1333 (PC3 10666)

Controller Card: Rosewill RC-218

Hard Drive Array: 2 X Supermicro CSE-M35T-1B 5x3

Case: AZZA Helios 910

 

Thanks!!

How does the system work after unplugging the power? My system stops working without power.

He is obviously talking about the power towards the bay, not the mb power.

 

Are there allready disks connected in that bay or does this behaviour happen when only connecting the bay itself with no drives in it ?

It had 1 green drive in it.  The drive would have connected to my controller card.

 

I tried to "recreate" the rebuild by initiating a parity check and plugging in / unplugging the power to the bay.

 

Without the second bay plugged in I run about 95 MB/sec.

When I plug in the power to the bay containing the single green drive it drops to around 58 MB/sec.

When I plug in the power to the bay containing NO drives it drops to about 40 MB/sec.

When I plug in the power to the bay with NO connectivity to it whatsoever except for the power it drops to about 40 MB/sec.

 

Any thoughts?

Well... If you are really sure that you get a drastic effect like this when only connecting power, and if you can reproduce this, then it can only be an issue with your power suply, you have taken the rest out of the mix. As sherlock holmes said: if you remove everything that is not possible then what remains, however unlikely, is the truth.

It has to be a faulty PSU or a faulty bay.  Any idea how to figure out which one?  I don't know how to test the power pulled by an individual molex or sata power port.  I have a spare Rosewill RV2-700.  The problem is that it is 700W and can probably supply the second bay with what it wants, even if it is faulty.

 

If you have a spare psu I would try switching that... If the performance thing moves with it then its the cage..

 

Do double check though... It all sounds pretty unlikely...

You did not say what Unraid version you are using and I suspect the reason is that your new HD is connected to add-on controller that may have problems working well under your chosen Unraid version.

 

Connect all HDs to the motherboard SATA slots to reconstruct your data ASAP and then you will think why it does not work

 

The power supply (unless bad) should support your configuration without any problem.

My unraid version is 4.7.  All drives are connected to the motherboard, not the RC-218.  The drive that was in the second bay was connected to the RC-218, but the bay/PSU causes the problem even without a drive in it, and disconnected from everything except the power.

If switching out the power supply foers not work then call the ghostbusters...

I ran some additional tests, and have produced some interesting results.

 

I hooked up four drives to my windows system inside one of the supermicro backplanes.  I set up a striped disk across three of the drives, and leave the fourth as a normal simple partition.

 

Test 1 - I run a speed test on the fourth drive, with no activity to the other drives.  I get the expected R/W speeds.

Test 2 - I run a speed test on the fourth drive, while copying a large file to the other drives.  The drive performance drops significantly.

 

I remove the drives from the backplane, and hook them directly up to the system.

 

Test 3 - I run a speed test on the fourth drive, with no activity to the other drives.  I get the expected R/W speeds.

Test 4 - I run a speed test on the fourth drive, while copying a large file to the other drives.  I get the expected R/W speeds.

 

Here are the actual test results.

 

Expected - From Test 4, and consistent across Tests 1, 3 & 4

-----------------------------------------------------------------------

CrystalDiskMark 3.0.2 Shizuku Edition x64 © 2007-2012 hiyohiyo

                          Crystal Dew World : http://crystalmark.info/

-----------------------------------------------------------------------

* MB/s = 1,000,000 byte/s [sATA/300 = 300,000,000 byte/s]

 

          Sequential Read :  132.396 MB/s

          Sequential Write :    93.653 MB/s

        Random Read 512KB :    66.123 MB/s

        Random Write 512KB :  117.524 MB/s

    Random Read 4KB (QD=1) :    1.160 MB/s [  283.1 IOPS]

  Random Write 4KB (QD=1) :    3.037 MB/s [  741.4 IOPS]

  Random Read 4KB (QD=32) :    1.401 MB/s [  341.9 IOPS]

  Random Write 4KB (QD=32) :    2.991 MB/s [  730.3 IOPS]

 

  Test : 50 MB [E: 1.9% (17.3/931.5 GB)] (x1)

  Date : 2012/12/28 22:31:29

    OS : Windows 7 Ultimate Edition SP1 [6.1 Build 7601] (x64)

 

 

Poor - From Test 2

-----------------------------------------------------------------------

CrystalDiskMark 3.0.2 Shizuku Edition x64 © 2007-2012 hiyohiyo

                          Crystal Dew World : http://crystalmark.info/

-----------------------------------------------------------------------

* MB/s = 1,000,000 byte/s [sATA/300 = 300,000,000 byte/s]

 

          Sequential Read :    22.792 MB/s

          Sequential Write :    12.029 MB/s

        Random Read 512KB :    10.667 MB/s

        Random Write 512KB :    16.446 MB/s

    Random Read 4KB (QD=1) :    0.073 MB/s [    17.8 IOPS]

  Random Write 4KB (QD=1) :    0.184 MB/s [    44.9 IOPS]

  Random Read 4KB (QD=32) :    0.092 MB/s [    22.5 IOPS]

  Random Write 4KB (QD=32) :    0.195 MB/s [    47.6 IOPS]

 

  Test : 50 MB [E: 1.9% (17.3/931.5 GB)] (x1)

  Date : 2012/12/28 22:20:47

    OS : Windows 7 Ultimate Edition SP1 [6.1 Build 7601] (x64)

 

 

This happens with both of my supermicro backplanes, identically.

 

Anyone have any insight on the CSE-M35T-1?  Maybe they aren't all they were cracked up to be?  Maybe I got a bad batch?

 

Thanks in advance.