Jump to content
  • 0

Building new server from scratch!


hansolo77

Question

Hey all!  I've been working on learning how to setup and install WSE 2012 R2 and I think I'm finally there.  However, I'm feeling very limited in my capacities as my case is overflowing with drives.  So I'm going to be building a new server from the ground up and would like some advice and suggestions as to what I should get.  I'm only a part-time worker and don't really make a lot of money.  So the purchasing time line is going to be really long.  So far, I know for certain that I'm going to but the Norco 4224 case.  I've read reviews and it has been recommended time and time again.  The only issue with this case appears to be the fans, and the potential for the backplanes to be DOA.  But at the price, it's a steal compared to other similar cases.  Plus, I'm going to get it through NewEgg, as they're the cheapest place around, and they offer really quick RMA's for exchanges if there's something wrong with.
 
Now I'm at the point of internals.  The first order of business is the motherboard.  Form factor isn't really an issue as the Norco case fully supports a whole range.  Individual features are where I'm struggling.  I know that I want to have room for expansion.  So I'm kinda staying away from those Mini ITX boards, since they, for the most part, all seem to have only 1 expansion slot.  I already have 1 SAS controller, and plan on getting an expander.  So that would be 2 slots.  As for the processor, I'm not sure what I need.  My usage scenario is a simple home file server for client backups and media streaming.  So I suppose I don't need anything major.  The same goes for RAM. 
 
As it is right now, I'm thinking about getting one of these:

SUPERMICRO MBD-X9SCL-F-O LGA 1155 Intel C202 Micro ATX Intel Xeon E3 Server Motherboard
or
SUPERMICRO MBD-A1SAM-2550F-O uATX Server Motherboard FCBGA 1283 DDR3 1600/1333
 
The first board is nice, in that it has plenty of expansion slots, Support for Xeon processors, and is cheap.  However, it only supports ECC memory of max 32gb, SATA 3.0GB/s, and is cheap.  The second board is nice, in that it has just enough expansion slots for my needs (controller and expander), already comes with a processor, supports (but not require) ECC memory of max 64gb, and has 2xSATA 6.0Gb/s, but is a little more expensive.
 
So which board should I get?  Integrated CPU and more RAM, or more slots, less RAM, and mandatory ECC?  Or should I look at something else?  What are your suggestions?

EDIT ->  I just looked at and am now also adding this contender:
SUPERMICRO MBD-A1SAM-2550F-O uATX Server Motherboard FCBGA 1283 DDR3 1600/1333

 

It's got more everything.  More SATA 6.0Gb/s, TONS more RAM, support for faster CPU, and more expensive.  To put it in perspective, this new board would probably take a month to save up for.

Link to comment
Share on other sites

220 answers to this question

Recommended Posts

  • 0

As I said earlier in this thread.. I had trouble with the BETA.  I think it was because I was having trouble with it saying something about insufficient space (swap, etc) when I tried to transfer files ONTO the DrivePool from another computer.  I'm pretty sure that's what the issue was.  I couldn't figure out a solution that worked other than downgrading to the latest stable build.

Link to comment
Share on other sites

  • 0

I don't remember.  It might not have been just that issue either.  Is there a chance if I install a new version that it would screw up the current pool?  I'd hate to have it "working" the way it is now, then go and install the beta just to find out I have problems, or the pool somehow gets corrupted and I'll lose my (now) 27tb of data.  If there's no real risk, I suppose I can try the latest beta next time I have a free day off from work.

Link to comment
Share on other sites

  • 0

Ok, so I upgraded to the latest beta.

I still have a bunch (nearly 500-600mb on each drive) of "other" data, totaling about 8gb.  I wonder though if this is because of CuteFTP downloading from my seedbox.  If I sit and watch, that 8gb will fluctuate.  CuteFTP is running multiple downloads in parallel then combining them afterwards to increase download speeds.  So it will in effect download 8 copies of the same file up to say 1gb each, then combine them down to a single 8gb file.  Maybe DrivePool is just picking up those files.  Something I plan to do is get a separate drive for downloading, then do a manual move into the drivepool once everything is completed.  Do you think that would help?

 

Anyway, here's what my current pool looks like, and I purposely hovered over 1 drive to show it's data breakdown.

 

UWeyaWK.jpg

Link to comment
Share on other sites

  • 0

If CuteFTP is downloading to the Pool directly, and downloading 8x 1GB file parts in parallel, then yes, that would account for the ~8GB of "other" data, definitely. 

 

And that it's fluctuating seems to indicate that this is exactly what is happening. As the file is being written to, it will remain as "other" data until it's no longer locked. At that point, it should show up as normal (duplicated or unduplicated, depending on your pool settings). 

 

If that is what it does look like, then this is definitely normal.  

 

 

As for downloading, I would recommend a dedicated drive for this, personally. You can then schedule (or script if CuteFTP supports that) batch moves to the pool (which should minimize this). 

Link to comment
Share on other sites

  • 0

UHG.. I just have no luck with servers.  Ordered another 4tb NAS drive (Seagate) and ran it through it's tests and slow (not quick) format.  Connected it to the server, everything's good.  Wake up today, and one of my 3 year old no-name drives had completely died, and one of my other new 4tb drives (the first one I ever bought for this server.. 208 days ago) is now reporting over 1000 unreadable sectors in Scanner.  I looked, and I just can't justify the $100+ price difference between Seagate's 4tb NAS drives, and WD's 4tb RED drives.  Still, it might be time to reconsider, with this drive "dying".  I haven't even made it to the 1-year anniversary mark yet (let alone the 5-year for warranty) and it's already reporting a HUGE chunk of bad sectors that didn't exist originally.  I know, because I ran a Scanner sector-scan prior to filling it up, and have had it 90% full since.  Maybe I should just quit trying to make this work.

Link to comment
Share on other sites

  • 0

Well, double check the drives.  If you're getting a lot of unreadable sectors and seeing no SMART warnings, try running a burst test on the drive. If that comes back with errors as well, then this may be a communication issue with the drive. That's easier to fix than an actually damaged drive.

 

I've seen this happen with some controllers (HighPoint) with specific hardware combinations. It just doesn't like them and errors out. 

 

 

 

Also, keep in mind, any drive under warranty can be RMAed. Meaning you can get a new drive without having to pay for a new disk. If you can, the advanced RMA option as they'll ship the drive first, and then you ship the drive back.  They say that they put a hold on your credit/debit card, but the couple of times I've don't it, it's $1 as long as you return it before the "time out" period for the drive. 

Link to comment
Share on other sites

  • 0

I've done the advanced replacement before... it actually costs $12 I think, because they charge for the expedited replacement.  I had to go that route because I didn't have any saved HDD boxes with the plastic suspension holders on the sides and no ESD bag.  I've been saving them ever since, but I still like having the new drive arrive quicker than waiting for them to confirm the delivery before shipping out the replacement.

 

It's definitely not a controller issue.  Well, maybe I shouldn't say "definitely".  It's that LSI controller you sold me.  It's been working great even before I rebuilt the new server.  When I look in Scanner, it had alerted me that the drive was failing.  So I did a manual sector check and once it got up to 97% clean, it hit the snag with the +1000 unreadables.  It's strange too that it would have such a LARGE area all at once, within like 12 hours.  The SMART details shows specifically that 72 sectors have been reallocated, and the 1327 "unreadable" are listed as Pending and Uncorrectable.

 

I'm going to take this drive out and connect it to my regular computer to run the Seatools scan on it.  It doesn't run on Windows Server for some reason.  It's been sitting here for about an hour "Removing Drive" from DrivePool, saying it's checking for open files, which there are none.  I'll probably have to leave it running overnight and come back to it tomorrow.

Link to comment
Share on other sites

  • 0

Well, the LSI controller was well tested and in good condition, so yeah, that shouldn't be an issue. 

 

Though, cables can always go bad (I've burned out SATA cables! Seriously!) and it could be a single bad connection (or even loose).  but based on what you've also said: 

 

The SMART details shows specifically that 72 sectors have been reallocated, and the 1327 "unreadable" are listed as Pending and Uncorrectable.

 

Ah, yeah, that would do it. :(

 

You'll probably see a matching (or very close) number of unreadable sectors as you will the pending and uncorrectable SMART value.  

 

Especially if this is a Seagate ST3000DM001.  

 

I've seen this on the list drive and others. And generally, it goes down very fast. Clear all the data off of that drive, if possible. 

Link to comment
Share on other sites

  • 0

This is actually one of their 4tb drives.  I think that model number you listed indicates a 3tb one.  I was looking at Seagate's website, rather than just NewEgg or Amazon.  They offer different levels of NAS drives (not just capacity).  The drives I've been buying are just regular "run of the mill" NAS drives, but they also have Enterprise and even an "Enterprise Capacity" line.  Looking at the differences, these Enterprise drives seem to be advertising a longer shelf life (2M MTBF vs 1M) and higher workload (550TB/yr vs 180TB/yr).  They also mention better ideal environment of SAS enclosures over 16 drives (and my Norco RPC-4224 has 24 drives) whereas the regular NAS say 1-8 bays.  But I'm not in a business scenario.  I don't pump out nearly that much data workload streaming videos and music do I?  The "MTBF" (whatever that stands for) being higher looks nice on paper, but even 1M hours is 114 years.. so that's not really a major selling point either.  So I'm trying to decide if continuing to buy Seagate (and maybe get Enterprise drives instead) or if I should just change gears completely and start buying Western Digital drives.  A bunch of my friends swear by them, but I can't just drop another $100 bucks for the same amount of storage.  With Seagate having a quick and easy RMA process, and their advertising to last 114 years, surely that's gotta be good enough.  I mean, I bought this most recent 4tb NAS drive from Amazon for only like $140 (was on special sale with a discount coupon).  WD RED's for the same size are $220+.  Are they really that much better?

 

Do you think the problem might be with the SAS cable?  I'm no longer using those SATA>SAS cables now that I have that HP SAS Expander.  I'm just using 6 direct SAS cables to the backplane.  One thing I still need to do is get a replacement backplane for "Rail 3" as that 4th drive bay still fails to power and detect drives connected to it.  But this drive I'm RMA'ing was in Rail 1 Bay 1 (Top Left, first bay).  All my other NAS drives are in that whole top row (Rail 1) and are working fine.  *KNOCK ON WOOD*  I just installed this latest one yesterday, so I'll have to wait and see how it performs.  The next oldest NAS drive is in Bay 2, and it's clocked at 102 days.  Still another 3 months to go before it reaches the plateau that this one hit.

 

One thing I still love about these drives, and this case, is that I only have 3 drive bays left to fill, and none of my drives reach temps higher than 40c.  And it's in a bedroom closet with the door shut but for an inch.  :)  Speaking of which, I definitely don't think the fail of this drive was from heat, as it's recorded lifetime temp range was 29c-40c.

Link to comment
Share on other sites

  • 0

ST4000DM000 then? 

If so, it's basically the same line. Not as bad, but yeah, can still happens. 

 

 

 

MTBF is "Mean time between failure" and basically, it's the average lifespan of the drive, calculated in hours.  Longer is better, but this also depends on workload.  

 

As for brand, it doesn't really matter. Aside from the ST3000DM001 and similar drives (which I'd recommend against), either brand should be fine.  A lot of people have different preferences, but whatever you are comfortable with buying. 

 

 

As for the NAS vs Enterprise drives, the limit is a recommendation. You can definitely run more than that, but it's what the manufacturer recommends as their limit. Due to things like heat, power consumption, vibration (and vibrational harmonics).  

Enterprise drives have a lot more features that make them ideal for rackmount use, and NAS drives only have some of them.  However, there are plenty of people using Greens in production without any issues. 

 

 

As for the cable?  No, not likely. As I said, given what you've said, .... that's basically the death throws for the Seagate drive. That's why I said to remove it from usage as soon as possible.  I've had drives exhibiting the exact same behavior as what you're describing, and most of them failed shortly after removing them (fortunately, AFTER data was removed, for the most part).  And this was for more than a few of these drives... and yes, this is why I duplicate everything, because I did lose a chunk of data. 

 

 

As for the temp, that's not horrible, but it's hotter than I'd personally run them. But anything above 40C bothers me.  

But this definitely wasn't due to the temperature, as that really isn't high (40C == 104F).  It's when it hits 50+ that you should definitely start getting concerned. 

Link to comment
Share on other sites

  • 0

Hi

 

Bad power supply's can also cause this it has happened to me in the past check the 12+ Rail output if it's around 20 to 30 amp it's no good most new ones put out between 60 to 75 amp and get a decent make.

its also good practice to connect all the power connections on the back of the back planes you should have 2 per backplain try mix them to spread the load so instead of using a splitter and going 1 & 2 go 1 @ 6 that way the draw on spinup is spread per output and try not to connect more than 8 drives per output from the power supply.

 

My power supply came with 12 molex and 12 sata over 6 output leads from the psu try look for one that does the same.

Link to comment
Share on other sites

  • 0

Hi lee.  Thanks for the information.  I checked my peripheral +12 line, and it's pumping 78 amps.  So I think I'm good there.  :)  As for the multiple connections to the backplanes.. I didn't really do this because those secondary connections are used when you have multiple power supplies for redundancy.  I don't think they have any benefits if you connect all of them.  However, the staggering of connections using separate cables is a good idea I think.  I believe right now I'm using 2 lines, with all 4 of one connected and then the last 2 from the other.  I'll see if I can find some time and maybe try splitting them so cable 1 is connected to 1, 3, 5... and cable 2 is 2, 4, 6.  It's worth a shot anyway, and it might improve the PSU's longevity. 

 

Chris... the model is actually ST4000VN000.

Link to comment
Share on other sites

  • 0

I'm going through (just initiated) all my drives doing a manual scan.  Something kinda strange has me concerned..  A lot (if not most) of my drives are reporting they haven't been checked since back on 12/17/2015.  I have Scanner set to check every 30 days, and to re-check sectors marked in 15 days to make sure they were re-mapped and working ok.  So I'm kinda concerned why Scanner hasn't been checking my drives.  I am running v 2.5.2.3100 beta.. might that version be buggy?  I do see there is a later beta out.  Just curious.

 

Still waiting to hear back from Seagate.  When I started the RMA, they said they were out of stock of this particular model and asked if I wanted to still proceed. Uh.. yeah?  I need a replacement drive, or I wouldn't have started the process of the RMA.  LOL stupid.  Anyway, I haven't heard from them since on when they'll ship me a replacement.  Hopefully it's soon, because they already charged my card (not a hold, though it might get refunded) and their website says once the RMA is started, they'll charge me a restocking fee if the old drive isn't received within 30 days.  I'd hate to get hit with a fee because they're taking to long to ship me a replacement.

 

I did run the drive through SeaTools on my other computer last night, and although it passed it's SMART check, it failed the short drive self test almost immediately, and gave me an error code.  So at least I know officially it doesn't pass it's own tests.

Link to comment
Share on other sites

  • 0

If you manually started a scan, that could be why.

 

Make sure that the "Start / Stop automatically" button at the top is enabled/highlighted/etc.  If it's not, click it to enable it. 

 

 

 

Hopefully it's soon, because they already charged my card (not a hold, though it might get refunded) and their website says once the RMA is started, they'll charge me a restocking fee if the old drive isn't received within 30 days.  I'd hate to get hit with a fee because they're taking to long to ship me a replacement.

 

Odd, it's been a hold in the past, But they may have changed things. 

 

And check the status tomorrow (monday). If it hasn't shipped yet or hasn't updated in status, then contact/call them to get it expedited.   Though, I don't think it will be an issue. 

 

 

 

And yeah, I'm not really surprised that the SMART tests failed on the drive, all things considered. 

Link to comment
Share on other sites

  • 0

Got the new drive, all is well. 

 

Brings up something else I've been meaning to ask... is there a way to force a re-balance of the pool?  Since adding this new drive, I now have it sitting completely empty for 2 days or so, while my other drives are exceeding their 90% storage limit I though I had configured in the DrivePool Balancer settings.  I tried to "recheck" the pool but it doesn't do anything but verify everything is properly duplicated.

Link to comment
Share on other sites

  • 0

Help Help Help!!!

(ok, mostly to Chris)

 

Running Windows Server 2012 R2... doing a daily backup of my server's OS drive twice.  Log on today and see a warning message that my backup drive has less than 10% free space.  I'm using a single drive dedicated for server backups.  So I look, and find NO means to remove old backups.  I have a drive FULL of double backups dating back MONTHS.  In the WHS days, it was a simple, configurable lifespan (I think for client backups too).  Just tell it to keep x backups for y number of days/weeks/months.  I don't see ANYTHING like that for the server backup in this OS (with Server Essentials).  Any ideas on how to fix this?  I clearly don't need 2 backups every day for the last 8 months lol.

Link to comment
Share on other sites

  • 0

I believe that it should start pruning the backups as it runs out of space.  Unless you start seeing errors about the backups (like it not running because it doesn't have enough free space), then you should be okay.

 

Worst case, hook up a second drive for backups, and let reconfigure it to use both. Then put the full drive away for the time being. 

Link to comment
Share on other sites

  • 0

Sorry I never got back to this... it looks like it cleared itself up on its own like you said.

 

So I got another question for you...  I finally got around to buying another SSD drive specifically to use for recorded TV shows.  I've been having some issues with playback quality because it seems like the recording is taking place and DrivePool is writing the data to multiple drives simultaneously, causing lag and video corruptions.  I decided I don't really need to keep a double copy of my TV shows I'm recording on the server since they're typically watched and erased within 24 hours.  So I want to add a new drive to my system and dedicate it for use of TV recording.  HOWEVER, I also know there is a way to use the SSD drive with the DrivePool so that it will act as a "loading dock" for new files and then write those files off to their more permanent drive destinations later.  I have no idea how to do this.  Can you give me some hints?

 

So here's what I think I need to do..  Install the SDD.  Give the SSD a specific Drive Letter so PVR software can write to it.  Then in DrivePool, add this drive to the pool and some how configure it to be the NEW FILE drive.  Am I correct in this progression?  Or can I not utilize a single drive for both functions?

Link to comment
Share on other sites

  • 0

Update... I went ahead and just bought me another SSD.  So now I have 3.. the OS drive, a PVR drive for recorded TV, and a Download drive.  Is there any documentation on how to setup the SSD Optimizer?  What I'm going to try to do is just use this 3rd drive as location where all my stuff I download via sFTP from my Seedbox goes, then have it move the data out to the pool.  This is a new feature for me, so I'm open to any help.

 

EDIT - Doesn't seem to be working...  I added the drive to the DrivePool, then configured SSD Optimizer to use that drive as an SSD and the rest as Archive.  I then tried to download a file from my seedbox like always, and it is using the DrivePool as it's location rather than the SSD.  It's a large file, so I've been able to watch it for an extended period.  The new SSD hasn't been touched since adding it to the pool (other than adding a hidden PoolPart folder).  What am I doing wrong?

 

EDIT 2 - I looked at the File Placement/Rules tab.  I've never been in here before, and didn't think I'd need to be.  However, using the path structure in the left window, I went to my "downloads" folder I have in my DrivePool, and configured it to ONLY USE the SSD, and unchecked all the rest of the drives.  I also told it to never use any other drives via the option checkbox.  That did the trick as far as getting the SSD to be used.  However, now I feel like these settings will cause the SSD Optimizer plugin to no longer work, as it won't move the data OFF that drive once it's full because I've told DrivePool to only store files in that folder on the SSD.  So I'm completely lost now.  Using just straight up defaults, and having SSD-O configured with the SSD as the SSD and pool drives are Archive, it seems like the plugin is being ignored.  All new files are put in the Pool straight up.  If I go in and configure specific file placement rules, the SSD is then used, but the automatic pool placements probably won't happen when it balances.

 

PLEASE HELP!

Link to comment
Share on other sites

  • 0

You'd want to use the "SSD Optimizer" balancer to do this:

https://stablebit.com/DrivePool/Plugins

 

 

As for specific setup, it depends heavily on what you want to do. 

 

Also, if you're using the SSD Optimizer and the File Placement rules, you will want to disable the "Unless the drive is being emptied" option in the main balancing settings. 

 

Also, if you want to balance data out periodically, change the balancing options, to "balance every day at", or the "balance immediate", but check the "not more often than every X hours" option and set this to 24+ hours. 

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Answer this question...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
×
×
  • Create New...