I just ordered two ST2000DL003.  I will replace the bad drive with this one and I need to preclear and have a drive in waiting.  The pre-clear process takes so long I hate driving with my pants down for so long in the event of a failure...

See this for replacement info: http://lime-technology.com/wiki/index.php?title=Un-Official_UnRAID_Manual#Replace_a_failed_disk

 

You could pre-clear one disk and use the second as a temporary replacement. But then unRAID will have to clear the disk which will take about 1/4 as long as a pre-clear and the array will be offline for the entire time. You will then want to replace the temporary with the pre-clear one and then pre-clear the temporary.

 

I suggest you skip the temporary and just wait for the pre-cleared one to be ready. I use my cache drive as a ready spare in case of a failure. My cache drive is green. I don't think the performance benefit of a faster drive is worth it because I mostly read from the array. Alternatively, just keep a spare pre-cleared 2TB drive at hand.

I Can't have a precleared disk sitting on the shelf and ready to go?  I thought preclearing data was written to the drive?

 

Anyway you know how they say when it rains it pours?  Well my other unraid box was unresponsive this morning.  When you went to /mnt/user it would lock up.  I was able to do some stuff on the console but I ended up running a shutdown and now parity is rebuilding.  I think a drive there is bad as well.  Attached is my syslog.  This machine is 4.7 and the newer unmenu.  Man it is nicer   I need to update my other box.

 

Let me know if you see anything wrong.

 

Thanks!

 

Neil

syslog-2011-04-07.txt.zip

sdg is failing. The Reallocated_Sector_Ct has gone from 31 to 67 in 3 days. I suggest you replace it ASAP. Then you can preclear it 4 or 5 times and see if the count stops increasing or it dies. Hopefully you can RMA it.

 

EDIT: based on the power on hours I'm guessing you can RMA. They may give you an issue saying that there are thousands of spare sectors. That is why you need to pre-clear it to death.

 

This HD also has been running at high temperature of 57 degC and has tripped the Seagate (which limits the HD to 55 degC) flag for airflow temperature:

 

190 Airflow_Temperature_Cel 0x0022  072  043  045    Old_age  Always  In_the_past 28 (10 76 35 24)

 

Worst value of 43 (100 minus 57 degC) is lower than 45 (100 minus 55 degC limit)

 

Check the cooling on your system

Yeah the room is warm.  I have a lot of crap in there.  It sucks and I wish it were cooler but Its a bit of a problem.  I don't think it is a fan issue.  This is a norco case with like the 16 bays...

 

Neil

 

See this for replacement info: http://lime-technology.com/wiki/index.php?title=Un-Official_UnRAID_Manual#Replace_a_failed_disk

 

You could pre-clear one disk and use the second as a temporary replacement. But then unRAID will have to clear the disk which will take about 1/4 as long as a pre-clear and the array will be offline for the entire time. You will then want to replace the temporary with the pre-clear one and then pre-clear the temporary.

 

I suggest you skip the temporary and just wait for the pre-cleared one to be ready. I use my cache drive as a ready spare in case of a failure. My cache drive is green. I don't think the performance benefit of a faster drive is worth it because I mostly read from the array. Alternatively, just keep a spare pre-cleared 2TB drive at hand.

 

I am preclearing a bestbuy drive currently.  When it is finished I need to upgrade the server to 4.7 (currently at 4.6) and then I can use the pre-cleared disk.  I assume that is ok at this point.  Should the old drive come out and the new one put in its slot?  I didn't think that would matter.  I am just really worried about F-ing it up

 

Thanks,

 

Neil

 

Yes. Select the new drive in the correct slot on the devices page. The physical connection should not matter.

 

EDIT: Is the drive AF? The safe way to do this is to first replace the drive and then upgrade. If it is a WD EARS AF, you could use a jumper and then upgrade to 4.7 without changing it and no loss of performance. If it is another AF drive, you would have to rebuild the drive after the upgrade to optimize performance.

 

pre-clear -C 63 will quickly change the drive alignment to be compatible with 4.6.

I did preclear.sh -A because it is an AF drive seagate so no jumpers.

 

Thanks,

 

Neil

 

Yeah at this point the array isn't broken.  Does upgrading cause the parity to be recalculated?  If not then I think I am good.  I will copy over the new files for 4.7.  reboot for good measure and at that time add the new drive to the array after taking out the suspect one.  After I do all this I will ask you guys what to do with the old drive.  I mean preclear it a few hundred times or just try to RMA it to seagate?

 

Neil

 

You can run the Seagate RMA program and see if it give you an error code. You can call them and see what they say about it. Worst case, pre-clear it to death.

Windows program =  blah

 

Neil

 

So I was about 24 hours in to pre-clearing the replacement drive and the machine shut down again.  AC company coming Wednesday to look at adding an AC unit to the room.  It needs it.

 

Looking at more commercial options for storage....now.

 

Neil

 

So I was about 24 hours in to pre-clearing the replacement drive and the machine shut down again.  AC company coming Wednesday to look at adding an AC unit to the room.  It needs it.

 

Looking at more commercial options for storage....now.

 

Neil

Get a portable a/c unit on wheels for the room. They are easy to find.

 

Sent from my HTC Incredible using Tapatalk

In regards to the rolling AC:

Totally agree but you need some sort of an exhaust and I don't currently have that really. 

 

replacement drive:

 

Ok I precleared my drive and upgraded to 4.7.  I think I Have a longer term plan I will ask about shortly but I want to know the best way to replace this drive.  Should I just pull the old one while the server is down.  Reboot it and tell it to rebuild on this drive that is now in its slot?

 

Thanks!

 

Neil

 

Yes. Give it a try. But is sounds like it could be a PSU problem.

I agree at this point.  The machine just turns off.  I have to imagine that is the case.  Could be a bad UPS? I don't know really unless I do some more homework. 

 

What size PSU's are people using for 10 drive machines?

 

Also I am thinking of replacing this machine with one I have on the way.  My question is how can I do so.  Can I move my drives and just boot back up?  I would imagine no?

 

Thanks for all the help!

 

Neil

 

Move the drives. Configure BIOS and boot up. UnRaid does not care much about hardware.

 

See this thread regarding PSU: http://lime-technology.com/forum/index.php?topic=12219.0

The power supply is 750 W.  Certainly not too small by those standard.  I mean 200 Watts for the drives leave plenty of excess!

 

In terms of pulling the drives how does unraid know which drive is which or does it just not matter?  I.e. if I moved SDG to SDC unraid doesn't care?

 

Thanks!

 

Neil

 

Is it a single rail?

 

The drive letters do not matter. They may change from reboot to reboot.

This is the PSU used:

 

http://www.newegg.com/Product/Product.aspx?Item=N82E16817371026

 

Thanks,

 

Neil

 

There are 4 12-volt rails, each rated at 25 amps or 300 watts. All of the hard drives and the motherboard share a single rail. For calculation purposes, this is a 25 amp 300 watt PSU.

So it's a hunk of junk huh?

 

Neil

It's intended to power PCIe video cards; usually used in a gaming or graphics rig. It should work with 8-9 green drives or 5-6 non-green. See the PSU thread for recommendations on a replacement.

This is a server board.  Asus running a i5.  Has 7 green drives and one SSD cache drive soon to be replaced with a non-green real platter disk.

 

I will go do some research on PSU now but I have a honking server coming soon with triple power supplies..........