Home brew NAS

well, i cant for the love of god figure out where the redundancy check logs too but it didnt email me any errors like it was supposed to :D
Obviously bit rot has caused the machine to lose them :D
 
Ahh yes.... :D
 
PS. on the ECC issue... I've done a little light reading and yes, ECC is going to be more resilient as it will detect and correct errors. However, ECC can also fail and if you believe your data is at risk with non-ECC, it is also at risk with ECC, it just happens less often. ECC is not a guarantee things will always work.
 
That paper made for interesting reading. Thanks for sharing. I've (hopefully) procured a Xeon E3-1220L and priced up a suitable mATX board to go with it. That processor has a TDP of 17W whilst returning reasonably good benchmarks.

Next thoughts are on cooling and a suitable PSU. I'm thinking along the lines of one of these:
http://www.scan.co.uk/products/nofa...-copper-cpu-cooler-for-all-intel-and-amd-cpus

Coupled with a few of these (1x 140mm and 2x 120mm) to get a nice and steady airflow through the case:
http://www.quietpc.com/nf-p14

The Mobo I'm looking at has 5 headers for running fans and the case has room for a couple more but I think three should be ample. Is it worth buying a separate fan controller?

I'm incredibly ignorant with what's feasible and what's not when it come to low noise cooling options.
 
I'm incredibly ignorant with what's feasible and what's not when it come to low noise cooling options.
Right...

If you connect to an onboard header, it is the BIOS that controls what you are able to do with the fans. Some just switch off/on, some have no control, some allow you to set autonomous limits and others interact with a program running on the main PC to control the fans. Unless you do a lot of careful choosing on boards, you are likely to end up with something that doesn't offer enough control for the an@l (i.e. me and possibly you ;)).

In my last PC overhaul, I added an AquaComputer which is around £60-£150 depending on how you want to spec it. See a partial build log: http://www.talkphotography.co.uk/threads/aiming-for-a-quieter-pc.481748/

It simply works (once you have your head around how to set it up). I'm sitting in a very warm room at the moment (the temp sensor is reading 30 deg C). The CPU is sitting ~ 50 deg C. The disk I use to monitor and control front airflow is 41 deg C. I have 3 fans and a fan on the heatsink (a Noctua DH14). The Noctua fans are idling @ 300rpm, the front fan is going at 1000rpm and the bottom and back fans are off...

I can hear the front fan slightly (the box is right next to my left leg), but this room is unusually warm (it has 3 computers in it and is 10ft x 10ft). The system is normally silent unless I'm recoding or editing images.
 
Thanks for that. Lots of food for thought there.

Looks like I've just snagged Asus P8B-M Motherboard on eBay for £25. Fingers crossed that it works. I'll have to fit it to the case before I get a clearer idea of whether the Nofan CR-80 will fit.
 
I finally amassed all the parts, shy of buying the disks and spent an hour or so this evening throwing it all together. And it works - a bit of a surprise considering the only reviews I can find of the MoBo are reports of it arriving DoA.
I booted it from USB and ran a few tests - not really proving anything conclusive aside from the fact it seems stable and stays cool (CPU 30 deg C) and it really is silent.

So far the build consists of:

Case: Coolermaster N400
Power Supply: BeQuiet 350W Gold Rated
Mobo: Asus P8B-M
CPU: Xeon 1220L v2 (17w TDP)
Memory: Kingston DDR3@1600 mhz, low voltage, ECC
CPU Cooler: Noctua NH-L9i
Cooling Fans: 4x Noctua NF-F12

Happy to post further details (and an update once I have disks installed) for anyone considering their own build.
 
Holy post resurrection batman.
I built a backup server without ECC RAM as a target for ZFS snapshots. And suspiciously I was seeing checksum errors in both my zpools. After every single scrub.
I've swapped out the ram, mobo and CPU - so they either the same/similar to my original build.

Still early days, but no checksum errors!

Also, with ECC under recent Linux kernels, it should be possible to determine if/when memory error correction is occurring. So there's another benefit of using ECC memory.
 
But how do you know? There could be a bit incorrect somewhere on the disk that you just haven't read. My ZFS arrays are checked weekly for data errors (this can be done whilst the arrays are online) and as of Saturday, there were none :p

But how do YOU know?

If they're only checked once a week then logically there are SIX other days when they could have errors!!!! :LOL::LOL::LOL:

Oh and BTW Happy New Year to all who read this :wave:
.
 
Back
Top