Proposed NAS build
- Thread starter Stereodude
- Start date
Stereodude
Not really a
Really? This is the first time I've used a SAS expander and backplane. For something in my basement in a home use the thought never crossed my mind to test that.
Clearly buying cheap'ish enterprise hardware and playing mishmash with it has it's limitations.
Well...when I build an array that has a large quantity of spindles, I tend to test the very-likely scenario of a drive replacement before actually putting data on it. I may also be biased and conditioned to things like this being in the enterprise storage industry.
Stereodude
Not really a
Live and learn... Next time :cooler:
Stereodude
Not really a
Nope. No difference in the lights from the working drives with the hardware combination that I have.
Newtun
Storage is nice, especially if it doesn't rotate
On the surface, that would seem to be difficult to implement.
ddrueding
Fixture
One of the advantages of a pre-built NAS is the finer points like this.
Stereodude
Not really a
FWIW, the rebuild completed successfully without incident. It took 14hours and 30 seconds to complete. I used the machine normally during that time so it's possible that my playing media files from it may have slowed it slightly.
Stereodude
Not really a
I'm finally putting together the older, smaller brother to this last "NAS" build. I'm using the same IBM raid card + HP SAS expander in a Norco 20 bay chassis. I'm using the 10+ year old motherboard from my old "NAS" with a better CPU (Q9550S). I'm using the eight 2TB Samsung 5400 RPM drives from the old server to start with. My slightly longer term plan is to get eight 14TB drives for the "new" server that I built in early 2016 and move the ten 6TB drives to the older smaller one. I had to cut out the back of a PCIe x1 slot to put the HP SAS expander in it since the two x16 slots will have larger PCIe cards in that that need the lanes. Cutting the slot wasn't exactly fun. I damaged two of the pins a bit cutting the slot, but the SAS expander doesn't use them, so it's basically irrelevant.
This has been brought about by the nightly backup server I've been using for close to 10 years having hardware problems seemingly with the drives in it's RAID array. It seems like it's going bad pretty quickly.
Edit: Now it's complaining about NTLDR being corrupted on the boot SSD.
I have a backup of it from ~5 years ago when I swapped in the SSD as the boot drive. I'll have to see if I can restore it. The OS drive backup is still current AFAIK since I basically haven't touched it since. Projects for tomorrow.
All the more urgency to get this other one running in the Norco chassis. It's doing a background initialization on the drives since it saw them as a foreign array. I was a little surprised that it was able to import the existing array from the PERC 6i the drives had been on. I plan to trash the old existing array, but this is basically a way to do a bit of a stress test on the drives since they've been sitting for almost 4 years and were only powered for the first time today.
This has been brought about by the nightly backup server I've been using for close to 10 years having hardware problems seemingly with the drives in it's RAID array. It seems like it's going bad pretty quickly.
Edit: Now it's complaining about NTLDR being corrupted on the boot SSD.
I have a backup of it from ~5 years ago when I swapped in the SSD as the boot drive. I'll have to see if I can restore it. The OS drive backup is still current AFAIK since I basically haven't touched it since. Projects for tomorrow. All the more urgency to get this other one running in the Norco chassis. It's doing a background initialization on the drives since it saw them as a foreign array. I was a little surprised that it was able to import the existing array from the PERC 6i the drives had been on. I plan to trash the old existing array, but this is basically a way to do a bit of a stress test on the drives since they've been sitting for almost 4 years and were only powered for the first time today.
Last edited:
Stereodude
Not really a
Yeah, it's a bit of a dog... With a 10gB NIC the performance is not thrilling. CPU usage is high and seems to be holding copy speeds back. Time to scavenge something else. An i7-2600k is the next candidate.
Stereodude
Not really a
The i7-2600k performs about 50% better. I get a little more than 600MB/sec during the copy. The verify is slower. I'm using SyncBack SE V8. It seems like their verify routine is not well optimized. It rereads the file from both the source and destination drive and does something to compare them (seems maybe like a hash / crc), but it is slower than the copy. It reads from the network drive at ~180MB/sec during the verify. The verify from the local RAID array happens at ~240MB/sec. The read/write during the copy is ~600MB/sec, but it's using the internal copy routines of Windows.
The Thermalright True Spirit 140 heatsink on the i7-2600k is too tall for the Norco RPC-4220 so I've got to get something shorter. I looked at the stock Intel unit, had a good laugh and put it back in the box. I'm also going to have to open up a PCIe x1 slot on this motherboard for the HP SAS expander also. I'm thinking to try melting it with a soldering iron this time.
The Thermalright True Spirit 140 heatsink on the i7-2600k is too tall for the Norco RPC-4220 so I've got to get something shorter. I looked at the stock Intel unit, had a good laugh and put it back in the box. I'm also going to have to open up a PCIe x1 slot on this motherboard for the HP SAS expander also. I'm thinking to try melting it with a soldering iron this time.
Stereodude
Not really a
And the SAS raid card generated multiple correctable single bit ECC errors on the cache module in the process of performing a backup of several TB of data. I tried reseating the module, but the errors continued. I swapped the card for a spare. It insists on doing a background initialization because it's an imported foreign array so I can't test it yet. Hopefully this one is good.
The good news is pulled IBM M5016 cards are ~$45 now on eBay so getting more to have a spare or two will be pretty cheap.
The good news is pulled IBM M5016 cards are ~$45 now on eBay so getting more to have a spare or two will be pretty cheap.
Stereodude
Not really a
The spare M5016 card worked as expected with no ECC errors. I ordered two working pull cards for $75 (not each) on eBay via a best offer.
Stereodude
Not really a
So after buying the correct IBM/Lenobo Supercap (FRU 81Y4579) I was able to update the firmware for the M5016 to the latest LSI/Avago/Broadcom firmware (5.14PR) and not get any errors about the "BBU". The seller in China I had bought the supercaps from years ago isn't selling the correct supercap for the M5016 and starting with a particular firmware revision the card started doing a white list check for the "BBU" and would refuse to use the BBU/supercap if you had the wrong one connected even though they're they're made from the exact same Tecate supercapacitors.
The good news is the 81Y4579 is pretty prevalent on eBay now and I got 4 of them for $46 shipped using best offer from a seller.
The good news is the 81Y4579 is pretty prevalent on eBay now and I got 4 of them for $46 shipped using best offer from a seller.