I am running 5.05 + UnMenu and have been suffering problems which seem to be related to hardware, I have recently updated my CPU and RAM

 

I recently had a disk redball, however after runnng SMART and Hitachi diagnostics, I reintroduced the disk and reconstrcuted data OK.

 

I have had a issue where the arrary has crashed after suffering a kernal panic, I had been advised that this is likley to be memory related. I have run 8 cycles of memtest taking 18 hours with no fault found, however after another panic I put the orginal RAM back, which had been used for 3-4 years with no problem.

 

However on 3 occassions now I have had the arrary crash on me, the screen attached to the array has a blank image, the motherboard is still live with CPU fan spinning. Connectivity to user shares is lost and I can't telnet in.

 

I have already put the original memory back in to discount a memory problem, does this sound like it could be CPU related? I have a mind to put the orginal CPU back in which had been used for 3-4 years with no problem, but would welcome any comment before I take the array apart

 

I am running 15 hard disks + 1 SSD from a Corsair TX650 PSU, and don't see that it could be power related

 

The array is just started a parity check, and it has found 70+ sync errors in 20 minutes of running. I atatched a syslog but I can't telnet in and UI( can't access UnMenu, however I can access the UnRaid GUI and mapped shares?

 

PS I did buy a second license to allow me to use another USB stick, to discount a USB problem, but this looks like a hardware problem

 

syslog.txt

Run: REISERFSCK --check on your Toshiba Drives...see this:

http://lime-technology.com/wiki/index.php?title=Check_Disk_Filesystems

 

 

I woke up this morning and found the screen blank, and telnet & GUI access and shares!

 

I think that I have now found my problem, I had not updated the BIOS since changing the CPU. On checking I have found that the existing BIOS was so old that it shouldn't have worked with my old CPU!

 

Array is currently performing a parity check again, I guess that I should waint and let it complete before running REISFER?

 

 

Yes... let it complete. And from what you say about the history of faults, I think your old parity is unlikely to be correct, so let it continue to run and rebuild.

 

Your syslog shows that one of your Toshiba drives was being fixed by unRAID during startup. ('fixed' as in: it replayed the journal of past  file changes to fix the file structure.  )

The Reiserfsck --check step is to determine if that replay of the journal was a one time thing or some larger problem with the file system.

 

From what you say about the BIOS/CPU, I'd probably run Reiserfsck on EVERY disk (except Parity) just to be sure.

 

 

Thanks, my problems started when I changed CPU and memory

 

I did check the ASUS CPU support list for my motherboard for compatibity, it said that my old CPU (E1500) was supported from BIOS version 2024. It also stated that my new CPU (Q9300) was also supported since BIOS version 2024

 

As my old CPU had worked OK for 5+ years, I had assumed that version 2024 or later was already been installed, and as such I did not need to update the BIOS

 

I had a shock this morning to find that BIOS version was in the 1600's, and can't understand why my old CPU had ever worked. Parity has been running now for 7+ hours with no problem

 

Parity Check completed with no sync errors, Now running REISERFSCK on each disk

I am half halfway through running REISERFSCK on all disks and all is OK so far

With 14 disks to check this is going to take a while

Using screen you should be able to run these is parallel.

Thanks for the reply, nearly there now, only 2 more disks to go

Phew I have now run REISERFSCK -- CHECK on all disks, no corruption was found, and parity is valid

 

I am going to leave it just with a basic 5.05 install for a day or two, before installing UnMenu, and new SFF-8087 cables (which are locking at both ends)

 

I have run memtest on the new RAM which I was going to install to my server for 24 hours with no fault found. I am worried out installing it again in case it kicks off another load of problems, but I assume that the root cause of my issues was not updating my motherboard's BIOS after installing a new CPU.

 

The whole reason of updating my server, from E1500 & 4GB RAM to Q9300 & 8GB RAM, is so that I can dabble with virtualisation which V6 brings

 

I'd bet on the BIOS problem. That crops often.

If your RAM passed Memtest before, swap it in when you're ready.

(I'd do it when you open the box to insert the new locking cables.)

Just be sure that you don't tie your SATA cables together.

And be sure that power connectors are properly seated.

---

In your future syslog, cable faults are usually easy to spot.

Your new memory passed memtest, so it should be fine. If its not seated properly, you'll know on boot.

 

 

 

Thanks for the help

Anytime you open the case you risk knocking something loose. If you ran a clean parity check and all disks check out you're good. I would just do what you're going to do in terms of cabling and swapping parts and then test it all out. Only when the patient is back together will you be able to truly test that everything is stable - hopefully for a year or more until another hardware upgrade. I like to run weekly parity checks after "major surgery" to make sure nothing is wiggling loose and creating an issue. After a month of this I go back to monthly.

 

I am also in the midst of an upgrade to 6.0. Finally got my MB working (initially I bought non-ECC memory not realizing it required it). And updated the BIOS. Good luck to both of us. I am trying to chronicle my upgrade in a thread in the motherboard section, so feel free to follow if you're interested. Going to try to convert VMware images to Xen and may try to install ESX to compare with Xen.

 

Cheers!

Thanks, I remember using Adaptec 4 port cards (4130?) which woudn't accept locking SATA connectors and having to pray to the UnRaid gods everytime I rebooted after I had opened the case, especially as my hard disks were not in 5x3 enclosures as they are today, but each was seperately mounted instead

 

I also remember replacing them with Supermicro 8 port cards and at the time (3+ years ago) finding it difficult to find SFF-8087 to SATA cables which locked at both ends. At the time I settled on unlocked SATA ends, however I have now finalised on fully locking ones

 

I also thought it clever to get a USB to motherboard header converter, so that the USB stick is kept inside the case. Mostly version updates can be be done live via a flash share, however on every occassion that I neeed to get to the USB stick, it always mean't reaching past SATA cables to get to it, with the risk of disturbing them. As such my recommendation to anyone would be to use a micro USB stick plugged to an external USB socket. I used to plug a full size USB stick in externally until I moved house and forgot to unplug it before moving the server, with the net result that I broke the stick and had to ask Tom to transfer my license to another stick

As such my recommendation to anyone would be to use a micro USB stick plugged to an external USB socket. I used to plug a full size USB stick in externally until I moved house and forgot to unplug it before moving the server, with the net result that I broke the stick and had to ask Tom to transfer my license to another stick

What brand Micro USB did you find that has a unique S/N so that it can be licensed? I've got several micro-usb's that I'd LOVE to use as unRAID keys, but they all have non-unique serial numbers.  

I am not using a micro USB, it was just my my recommendation to others based on my experience. I am currently using Sandisk Cruzer Blade, which are smaller than a full size USB and have unique GUID, if fact I have 2 as I took up Tom's offer of another license for $30

 

I do like the look of Sandisk Cruzer Fit http://www.sandisk.co.uk/products/usb/drives/cruzer-fit/ and can't see any reason why they should not have a unique GUID too

I do like the look of Sandisk Cruzer Fit http://www.sandisk.co.uk/products/usb/drives/cruzer-fit/ and can't see any reason why they should not have a unique GUID too

That's the one I tried first...Tom rejected the serial number as being non-unique.

But that's EXACTLY the solution I'd like to figure out, and for all the same reasons.

It wouldn't stick out the back and get broken off, but is removable without opening the case.

 

I'm glad that I didn't buy one as a second USB

You would think that the "guts" are the same accross different models with just the  case being different

 

 

I do like the look of Sandisk Cruzer Fit http://www.sandisk.co.uk/products/usb/drives/cruzer-fit/ and can't see any reason why they should not have a unique GUID too

That's the one I tried first...Tom rejected the serial number as being non-unique.

But that's EXACTLY the solution I'd like to figure out, and for all the same reasons.

It wouldn't stick out the back and get broken off, but is removable without opening the case.

Isn't that the same one that Limetech pictures being used here http://lime-technology.com/preconfigured-usb-flash/

I do like the look of Sandisk Cruzer Fit http://www.sandisk.co.uk/products/usb/drives/cruzer-fit/ and can't see any reason why they should not have a unique GUID too

That's the one I tried first...Tom rejected the serial number as being non-unique.

But that's EXACTLY the solution I'd like to figure out, and for all the same reasons.

It wouldn't stick out the back and get broken off, but is removable without opening the case.

Isn't that the same one that Limetech pictures being used here http://lime-technology.com/preconfigured-usb-flash/

Huh.

It IS, at least from the looks of it.

I've got a second license (unused)...I'll try again and see what happens.