Pre-Build Critique request.

[Knowltey]

Here is my current plan for my NAS upgrade.

Motherboard: SuperMicro MBD-X10SAE-O ATX LGA1150 C226 (219.99)
CPU: Intel Core i3-4150 Haswell Dual-Core 3.5GHz LGA1150 BX80646I34150 (129.99)
RAM: Samsung DDR3-1600 8GB/1Gx72 ECC M391B1G73BH0-CK0 (101.99)
(RAM taken from the supported RAM list for the motherboard on SuperMicro's site)
Storage: +4x Western Digital Red NAS Hard Drives WD10EFRX 1TB (279.96)
Storage: 2x 1TB HDDs already owned (1 Samsung Spinpoint F1 and 1 Western Digital Caviar Black 1TB)
Storage Configuration: 6 disk RAIDZ2
Full zpool encryption: Possibly, haven't decided for certain.

PSU: Already owned Antec Earthwatts 350W (May need to upgrade) - Looking into eventual upgrade to a redundant PSU, so if need to upgrade is there suggestions for those would be great.

UPS: Already owned, Tripp-Lite SMART1500

Case: Already owned, Antec 300.

Usage scenario: Light CIFS fileserver duty primarily (Only being accessed by one maybe two clients at any given moment). Weekly overnight backup storage from a desktop PC and a laptop.

Any known incompatibilites that anyone can see there that I have perhaps overlooked? Any known issues with FreeNAS running on the hardware in question? Any performance issues you can see me running into with the RAID configuration or anything like that?

Also, was possibly looking at the Intel Pentium G3220 Haswell Dual-Core as an alternative to the i3 for a cheaper cost, but I noticed that it didn't seem to have AES-NI support. If I do decide to go with an encrypted pool will the performance degradation be too large for that processor, or would the G3220 still be powerful enough to overcome it for CIFS file transfer speeds to be not effected or negligibly effected?

 

[Yatti420]

I don't run encryption so I'm not sure regarding the CPUs.. Other then that looks fine.. You may want to pick up a UPS aswell..

 

[Knowltey]

I don't run encryption so I'm not sure regarding the CPUs.. Other then that looks fine.. You may want to pick up a UPS aswell..

Ah, forgot to mention that. I already have a Tripp-Lite SMART1500 on my current NAS, desktop, and networking equipment.

 

[Yatti420]

Just make sure it supports your PSUs.. Active PFC PSUs can cause issues depending on the UPS..

 

[Knowltey]

Yep, I blew out a power supply once with that on my desktop PC. Degraded to an unreliable state a few days after a power outage. That said, it was because I ran it on the UPS for two hours during an outage.

Part of the problem is that I can't seem to find a Passive PFC redundant power supply, which would be optimal with a PWM UPS like I have, however it would be a bit less of a concern if it was a redundant one since I'd have one lead in the wall (through a surge protector though) and one in the UPS, so worst case scenario I burn out one side of the PSU and replace that one. (Although from my experience with various companies I've worked for, including a quite well known networking hardware company that uses Active PFC PSUs in their servers with PWN sine UPS devices, the damage to the PSUs and UPSes really only comes into play if you're running on the PWM wave for a decent length of time. In the minute it takes for the server to power down with my settings {30 seconds to poweroff command} the risk is near as you can get to negligible.)

(Also a large hospital network that did the same in a setup that required "medical grade" rating, which means **extensive** testing.)

(Running a PWM sine into an Active PFC PSU is like redlining a car. (PSU has to work really hard to correct to continually convert the pulses to an actual wave) You can do it, just try not to have it sit there for more than a few minutes while doing so or else you'll start breaking shit.)

If you know of any Passive PFC (or no PFC) redundant PSUs that would be great though (may even be cheaper than the Active one for that matter.

 

[Knowltey]

Ok, did end up finding one called Sure Star TC-500R8A 500w + 500w PFC Mini Server Redundant Power Supply http://www.circotech.com/i-star-500-500-auto-selecting-mini-redundant-power-supply.html

Does anyone here have any experience with this brand?

 

[Ericloewe]

Yep, I blew out a power supply once with that on my desktop PC. Degraded to an unreliable state a few days after a power outage. That said, it was because I ran it on the UPS for two hours during an outage.

Part of the problem is that I can't seem to find a Passive PFC redundant power supply, which would be optimal with a PWM UPS like I have, however it would be a bit less of a concern if it was a redundant one since I'd have one lead in the wall (through a surge protector though) and one in the UPS, so worst case scenario I burn out one side of the PSU and replace that one. (Although from my experience with various companies I've worked for, including a quite well known networking hardware company that uses Active PFC PSUs in their servers with PWN sine UPS devices, the damage to the PSUs and UPSes really only comes into play if you're running on the PWM wave for a decent length of time. In the minute it takes for the server to power down with my settings {30 seconds to poweroff command} the risk is near as you can get to negligible.)

(Also a large hospital network that did the same in a setup that required "medical grade" rating, which means **extensive** testing.)

(Running a PWM sine into an Active PFC PSU is like redlining a car. (PSU has to work really hard to correct to continually convert the pulses to an actual wave) You can do it, just try not to have it sit there for more than a few minutes while doing so or else you'll start breaking shit.)

If you know of any Passive PFC (or no PFC) redundant PSUs that would be great though (may even be cheaper than the Active one for that matter.

Honestly, I would never trust a PSU that's crappy enough not to have Active PFC in any computer, much less a server.

Also, the X10SAE is more of a workstation motherboard than a server one - it doesn't even have IPMI. An X10SLL-F might be a better option.

 

[cyberjock]

Honestly, I would never trust a PSU that's crappy enough not to have Active PFC in any computer, much less a server.

Agreed. I can't remember the last time I saw a "good quality PSU" that didn't have active PFC.

 

[Knowltey]

I was reading a thread where a bunch of people were having various issues with that board which is why I switched over to the SAE. That and the SAE has more of the faster SATA ports, and less of the other stuff I don't want or need like IPMI.

Also Passive PFC is fine. PFC only effects the stuff going on between the PSU and the source. To the inside of the computer there is literally no difference. (Active will lower power pull to do it's job, Passive just impedes it down to where needed within the PSU, so more power usage, more heat, more sound)

Passive PFC will use more power in situations where it comes into play than Active, plus heat. PSU will make more noise than a Active one as well.

Passive PFC is less "green" than Active which is why you see it less, since many markets such as Europe for example don't allow it, so manufacturers prefer active if they are going to do PFC because more market for it.

The only thing that will ever notice the difference is whatever is supplying you power, be it the power company or a UPS. Active PFC is easier on them than Passive PFC. If you're using a UPS that pushes out a simulated sine wave it can fuck with Active. PFC PSUs because they have to work really hard to correct the wave. Passive will just heat up a lot. Inside the computer though? Same result.

No PFC is what you wouldn't want to trust because that means no voltage correction, Passive and Active are both the same end result though, one is just more green than the other.

Either way PFC damage won't occur unless you are running on UPS power for a long while since the damage in either scenario is caused by overheating in the PSU, and running for long time on UPS isn't what you're supposed to be doing anyways. Only thing I let stay up during outages is my router switch and modem.

 

[danb35]

less of the other stuff I don't want or need like IPMI.

It's your machine, so your call what goes into it, but unless you're going to keep it hooked up to a keyboard and monitor all the time, you may find IPMI to be useful after all. Since most servers are run in a "headless" mode, IPMI greatly simplifies administration in that mode. Sure, most of the admin can be done with the web GUI or SSH, but you can't get into the BIOS that way, nor can you monitor temperatures, voltages, or fan speeds.

 

[Knowltey]

It's your machine, so your call what goes into it, but unless you're going to keep it hooked up to a keyboard and monitor all the time, you may find IPMI to be useful after all. Since most servers are run in a "headless" mode, IPMI greatly simplifies administration in that mode. Sure, most of the admin can be done with the web GUI or SSH, but you can't get into the BIOS that way, nor can you monitor temperatures, voltages, or fan speeds.

The server actually sits on my desk behind the monitors in a nook that was made for it and it and has a VGA lead to one of my monitors if needed as well as a wireless keyboard that I am able to connect if need to. So IPMI is on a "if it has it yay, but not going to pay extra to get it." status.

 

[Ericloewe]

I was reading a thread where a bunch of people were having various issues with that board which is why I switched over to the SAE. That and the SAE has more of the faster SATA ports, and less of the other stuff I don't want or need like IPMI.

Also Passive PFC is fine. PFC only effects the stuff going on between the PSU and the source. To the inside of the computer there is literally no difference. (Active will lower power pull to do it's job, Passive just impedes it down to where needed within the PSU, so more power usage, more heat, more sound)

Passive PFC will use more power in situations where it comes into play than Active, plus heat. PSU will make more noise than a Active one as well.

Passive PFC is less "green" than Active which is why you see it less, since many markets such as Europe for example don't allow it, so manufacturers prefer active if they are going to do PFC because more market for it.

The only thing that will ever notice the difference is whatever is supplying you power, be it the power company or a UPS. Active PFC is easier on them than Passive PFC. If you're using a UPS that pushes out a simulated sine wave it can fuck with Active. PFC PSUs because they have to work really hard to correct the wave. Passive will just heat up a lot. Inside the computer though? Same result.

No PFC is what you wouldn't want to trust because that means no voltage correction, Passive and Active are both the same end result though, one is just more green than the other.

Either way PFC damage won't occur unless you are running on UPS power for a long while since the damage in either scenario is caused by overheating in the PSU, and running for long time on UPS isn't what you're supposed to be doing anyways. Only thing I let stay up during outages is my router switch and modem.

Yes, theoretically I can make an excellent PSU without active PFC.

In practice, such a thing does not exist, as the only reason active PFC is omitted these days is extreme cost-cutting.

I'd rather trust something that is of good quality (and thus always ends up having active PFC) with a stepped "sine" wave UPS than something dubious enough not to have PFC, even if it does cause a somewhat unusual scenario for the PSU.

 

[Knowltey]

Yes, theoretically I can make an excellent PSU without active PFC.

In practice, such a thing does not exist, as the only reason active PFC is omitted these days is extreme cost-cutting.

I'd rather trust something that is of good quality (and thus always ends up having active PFC) with a stepped "sine" wave UPS than something dubious enough not to have PFC, even if it does cause a somewhat unusual scenario for the PSU.

Well I wouldn't trust something without PFC either, but Passive PFC =/= no PFC. The end result to the inside of a computer is the same between Active and Passive PFC, they just do it in different ways.

To the internal parts of the computer the PSU either does or does not have PFC. It's only the connection from "the wall" to the PSU that Active vs Passive.

Either way, this is a moot point, usage scenario here would mean that PFC is only ever going to be used during battery power periods, which the server will be set to power off after 15 seconds of outage, so a minute at most using the PFC, which is negligible risk. Worst case scenario is one side of the PSU fails and I replace it and I know already that Active PFC PSUs of the voltage I'm looking at do not overload my UPS or get overloaded as my desktop uses an even higher voltage Active PFC one without issues and my current NAS build uses one of equivalent voltage as well without issue.

Do you have any recommendations for a good ATX form factor redundant PSU, be it Active or Passive?

 

[DJ9]

I just ran across this today on Tom's Hardware. It's a Who's Who of power supplies updated for 2014.
http://www.tomshardware.com/reviews/power-supply-psu-brands,3762.html

 

[Knowltey]

Ah cool, that is helpful, I had ran across some Enermax ones in my searches but was not familiar with the brand at the time so wasn't sure about them. Looks like they are one of the major OEMs though, so that's good to know.

View: http://www.amazon.com/Enermax-Mini-Redundant-Server-Supply-EMA2V4550/dp/B00FYKNFA8/ref=pd_sim_sbs_pc_1?ie=UTF8&refRID=18PPV8PSGTJD3A1K9ZAP

Honestly, I would never trust a PSU that's crappy enough not to have Active PFC in any computer, much less a server.

Also, the X10SAE is more of a workstation motherboard than a server one - it doesn't even have IPMI. An X10SLL-F might be a better option.

Another thing I noticed about the X10SLL-F is it doesn't really offer me much any room for expansion since it only has 6 SATA ports, and only supports a maximum of 16GB of RAM.

Although on the same hand I did notice the X10SLH-F-O just now that while doesn't have 8 SATA ports, the 6 it does have are at least all the same (SATA3), costs a few dollars less and does have IPMI. I figure if I really need to in the future upgrade to even more disks than 6 I can get an expansion card or something like that?

Although the link I was looking at on SuperMicro's site was for an X10SLH-F and the one I found on Newegg is an X10SLH-F-O but I can't seem to find what the difference is? What does the -O notate with SuperMicro boards? The SuperMicro model number guide doesn't seem to have information about the O?

Would there be any reasons to not pick the X10SLH-F-O over the X10SAE?

 

[Ericloewe]

Ah cool, that is helpful, I had ran across some Enermax ones in my searches but was not familiar with the brand at the time so wasn't sure about them. Looks like they are one of the major OEMs though, so that's good to know.

View: http://www.amazon.com/Enermax-Mini-Redundant-Server-Supply-EMA2V4550/dp/B00FYKNFA8/ref=pd_sim_sbs_pc_1?ie=UTF8&refRID=18PPV8PSGTJD3A1K9ZAP



Another thing I noticed about the X10SLL-F is it doesn't really offer me much any room for expansion since it only has 6 SATA ports, and only supports a maximum of 16GB of RAM.

Although on the same hand I did notice the X10SLH-F-O just now that while doesn't have 8 SATA ports, the 6 it does have are at least all the same (SATA3), costs a few dollars less and does have IPMI. I figure if I really need to in the future upgrade to even more disks than 6 I can get an expansion card or something like that?

Although the link I was looking at on SuperMicro's site was for an X10SLH-F and the one I found on Newegg is an X10SLH-F-O but I can't seem to find what the difference is? What does the -O notate with SuperMicro boards? The SuperMicro model number guide doesn't seem to have information about the O?

Would there be any reasons to not pick the X10SLH-F-O over the X10SAE?

The O denotes retail packaging, B is for bulk.
Speaking from memory, the SLH-F is perfectly fine and has no disadvantages (in a server) compared to the SAE. Fit what it's worth, I'm using the SLM+-F, which adds minor upgrades to the various connectivity options (more USB 3.0, more SATA 6Gb/s and less SATA 3Gb/s, Intel chipset GbE is upgraded to a second i210 to match the first controller) and I can't say I'm missing anything.

Future expansion is easily taken care of with an HBA like an IBM M1015.
However, if you foresee a short-term need for expansion, the X10SL7-F is cheaper than separate motherboard and HBA.

 

[Knowltey]

The O denotes retail packaging, B is for bulk.
Speaking from memory, the SLH-F is perfectly fine and has no disadvantages (in a server) compared to the SAE. Fit what it's worth, I'm using the SLM+-F, which adds minor upgrades to the various connectivity options (more USB 3.0, more SATA 6Gb/s and less SATA 3Gb/s, Intel chipset GbE is upgraded to a second i210 to match the first controller) and I can't say I'm missing anything.

Future expansion is easily taken care of with an HBA like an IBM M1015.
However, if you foresee a short-term need for expansion, the X10SL7-F is cheaper than separate motherboard and HBA.

It looks like that SLM+-F has the SATA ports mixed between 4 SATA3 and 2 SATA2 ports, would it cause any issues to run a RAIDZ2 where some of the discs are on SATA3 and some are on SATA2?

 

[Ericloewe]

It looks like that SLM+-F has the SATA ports mixed between 4 SATA3 and 2 SATA2 ports, would it cause any issues to run a RAIDZ2 where some of the discs are on SATA3 and some are on SATA2?

Not at all, mechanical HDDs don't even come close to saturating SATA 3Gb/s. Even if they did, it wouldn't cause any issues beyond a slight bottleneck.