Greetings,

 

Setting up my first unRAID server.  Installed three hard drives, precleared them, and assigned them (no parity disk yet).  Added two more hard drives, and am trying to preclear those while copying some data to the first three disks.

 

Got a Kernel Panic.  Took a screen print.

 

Ran memtest for hours without any issues.

 

Any suggestions on what to do?

 

- Thanks,

    Marcos

I would start by running memtest and get that out of the way.

I would start by running memtest and get that out of the way.

 

Thanks for the reply.  I ran memtest for two six hour sessions already without error.  How long of a burn in do you recommend?

I would start by running memtest and get that out of the way.

 

Thanks for the reply.  I ran memtest for two six hour sessions already without error.  How long of a burn in do you recommend?

I think your memory is fine then.

Hmmm...  I just noticed that the two 1.5 TB Western Digital green drives I previously pre-cleared and brought online did *not* have pins 7-8 jumpered.  Could that cause my problems?

 

If it can, what's the remedy?  De-assign these drives, jumper the pins, and then re-run pre-clear?

 

 

- Marcos

FWIW, I'm still troubleshooting problems with this new server.  I'm trying to preclear all of my drives and the system would mysteriously lock up.  Only once did it crash with a "Kernel Panic" which prompted me to start this thread.

 

I haven't resolved anyting yet --- still can't get the drives to pre clear without crashing.  Attached is my latest syslog that I captured right after trying to pre-clear the three drives I currently have attached to the system (two 2TB and one 1TB).  At time Dec 28 14:07:08 in the log, I'm getting errors that I don't understand nor could find any good references to by doing some searches in the forum.  Any ideas?

 

My setup:

- AMD X2 245 CPU on an ASUS M4A785-M motherboard, 2GB DDR2

- Supermicro AOC-SASLP-MV8 SATA controller

- OCZ StealthStream II 500W PSU

- Drives: 1x Hitachi 7200 rpm 2TB; 1x Seagate 7200 rpm 1.5TB; 1x Western Digital 2TB (with pins 7-8 jumpered before installing); 2x Western Digital 1.5TB (with pins 7-8 *not* jumpered)

 

History:

1. Burn-in ram with the memory tester that comes with unRaid 4.6.  Did two six hour burn-in sessions without error.

2. Before installing the two 2TB drives, I successfully pre-cleared the three 1.5 TB drives with this setup. 

3. Added the two 2TB drives and got the system to crash while pre-clearing the two 2TB drives while simultaneously trying to copy data to the 1.5 TB drives (before assigning the parity drive).  First crash was the system became complete unresponsive.

4. Second crash was a kernel panic.

5. Swapped mini-SAS cable and tried again.  Ended in another crash.

5. Reformatting the USB thumb drive and reloaded softare, including umenu, smtp/mail, smartctrl, etc.

6. Tried pre-clearing the two new 2TB drives plus 1.5TB seagate, leaving the two 1.5TB WD drives connected. Resulted in another hang somewhere near the end of the preclear PRE-READ.

7. Disconnected the two 1.5TB WD drives and trying again (current situation).

 

Any ideas what to try next?

 

I'm at the point where I start to swap out the Supermicro SATA controller followed by motherboard and memory.  Swapping out PSU will have to wait a week until my new PC Power & Cooling Silencer 600W arrives.

 

- Thanks,

    Marcos

syslog-2010-12-28b.txt

Have you checked your memory settings in BIOS and compared them to either the motherboard manufacturers recommendations (QVL for memory) or against what the memory manufacturer suggests. 

On my testing server I had mysterious crashes several hours in and it wasn't until I asked on the ASUS forums that I got my systems running stable.

They would pass several hours of memtest only to crash when I was running something on them.

I never had a kernel panic with them but they would mysteriously lock up or if running windows either lockup completely or reboot without a 'bluescreen'.

 

Hope this helps,

Bobby

 

Thanks for the suggestion.  It didn't occur to me that this motherboard could have significant memory compatibility issues and/or not set the proper memory timing in BIOS automatically.  Doing a quick search on the ASUS site, I also so a lot of comments about memory issues. 

 

So, just as soon as the server crashes again, I'll take the system apart to see what brand/model memory I have installed and go from there.

 

- Marcos

Newegg identifies this power supply as having two 18 amp rails for the twelve volt line. Sometimes each individual rail is current limited sometimes they are just tied together.  If indeed it has two rails you may have an issue with your power supply.  I think it is good that you are planning to change out your power supply. At least you will be able to eliminate that as a possible cause.

 

Have you tried to move your drives around to different ports on your motherboard and/or SASLP? Is there any difference?

Greetings,

 

Following Professor101's suggestion, I looked up ASUS' Memory Qualified Vendors List for the M4A785-M.  Of the five pairs of DDR2 800 I have lying around, only one showed up on the list (Corsair CM2X1024-6400) and for this memory the recommendation was to use only one stick!  I could not quickly figure out this MB's BIOS settings to confirm or manually set the memory timings and voltage and so elected to try the one stick.

 

The good news is that the system didn't crash when I tried to preclear the three drives (sdb - Western Digital 2TB green; sdc - Seagate 1.5TB; sdd - Hitachi 2TB).  However, from the looks of the console messages, ran out of memory and two of the three preclear sessions died.  Currently tryig to preclear one disk at a time, starting with the Hitachi. So far, so good. Finally got through the Pre Read and am now roughly 50% through the zeroing disk. 

 

Don't know if I want to live with just one stick of RAM and so need to either find memory that works, try two sticks in different slots to get single channel/play with timings and voltages, etc. --- or swap out the motherboard.

 

Given ohlwiler observation on the two 18 amp rails, I will probably swap out the PSU sooner than later as well for the piece of mind.

 

I do have a question about the syslog file (attached) does this look like issues with my new Western Digital 2TB drive (with pins 7-8 jumpered)?

 

Thanks again for reading this / advice.

 

- Marcos

syslog-2010-12-29b-snippet.txt

Greetings,

 

Well, my first server has been running flawlessly for a few weeks now, so I think all my issues are resolved. 

 

Here's what turned out was going on:

 

1. My new 2TB Western Digital green drive was having issues.  Figured this out when I ran the Western Digital Diag program - smart test passed; short test passed; long test got stuck at 55 minutes remaining for over 24 hours.  Replaced the drive.

 

2. Got lots of errors when pre-clearing two 1.5 TB Western Digital drives for the 2nd time, because the first time I forgot to put the jumper on pins 7-8.  Did the low level zero write with the jumper on pins 7-8 under Windows, then pre-cleared the drives, ignoring the errors.

 

3. The Supermicro AOC-SASLP-MV8 card spits out lots of error messages when you're pre-clearing a drive and will continue to do so until the drive is added to your array.  If you don't like that, pre-clear the drive attached to a motherboard connector and then re-attach to the Supermicro board.

 

 

LESSON LEARNED:

 

1) Don't assume preclear alone is going to be sufficient for testing a brand new drive.  Too much potential for confusion, especially if you are aren't familiar with all the common errors you can get.  Better to test the drive with a drive test utility first.

2) don't assume that unRAID won't crash over a bad drive --- it can and did for me (I think it was because my log file got too big and I ran out of RAM, but, not sure.)  In any event, when it crashed, I got the 4.6 Kernel Panic which lead me down a completely different troubleshooting path.

3) The Supermicro AOC-SASLP-MV8 is a great card, but, it does have known features when it comes to error messages.

 

 

Thanks again to those who read this thread and provided some tips.  It helped me eliminate hardware unknowns that were very important to take care of.

 

 

- Marcos