Taking bets...

[Constantin]

Do you think that a Flex ATX board like the X10SDV-7TP4F can fit inside the Lian Li Q26 case?

A preliminary review with a measuring tape suggests it may just fit without shorting pin headers on the power supply? If FedEx is being truthful, I'll find out tomorrow...

At minimum I'll have to add two new standoffs for the two additional mounting holes (the case only features 4 mounting posts which are configured for Mini ITX).

 

[PhiloEpisteme]

I couldn't find super-specific specs on that case; you linked to youtube which did not contain specs and which contained broken links. Do you have a better link to that specific case?

 

[Constantin]

Lian Li took the specifications off their web site unfortunately. They only verified use up to Mini ITX and Extended Mini ITX such as the fairly similar, but slightly smaller ASRock E3C224D4I-14S (relative to the X10SDV-7TP4F). The length difference is 21.3 cm for the ASRock board vs. 22.86 cm for the Supermicro board. A tape measure suggests the SuperMicro motherboard might just fit, assuming that none of the pin headers are too close to the offending board edge or stick out too far.

 

[PhiloEpisteme]

If FedEx is being truthful, I'll find out tomorrow...

Are you basing your bet on FedEx's external measurements? I'm quite curious to see how this comes out. It seems like you've purchased the case already; do you have the board as well?

 

[Constantin]

I’ve been using the case for a while (see signature) as intended with a mini itx board. Unfortunately, the current board has just failed again (intel clock bug once more is suspect). IXSystems has been great regarding the extended warranty and will send me a another replacement in two weeks or so, when the current backlog has been cleared.

I’ve used the extended downtime as an excuse to upgrade the motherboard to the Supermicro board referenced above. FedEx claims that the Supermicro board should be in tonight, which will give me the opportunity to test fit before modifying the case (if it fits).

Two additional motherboard mounting studs would be necessary and any kind of drilling work like that basically requires a complete disassembly, vacuuming, etc to ensure no metal filings remain. I may also modify the back to remove the ribs on the rear exhaust, improve air flow, and fit that fan externally.

Good news is that I also have a much larger A76 case ready to go in case the q26 is too small. I just like the Q26 for its comparatively much smaller overall size, it takes up way less space on the shelf. On the other hand, the A76 allows the use of a 5.25” drive bay external fan controller, etc.

 

[jgreco]

Do you think that a Flex ATX board like the X10SDV-7TP4F can fit inside the Lian Li Q26 case?

A preliminary review with a measuring tape suggests it may just fit without shorting pin headers on the power supply? If FedEx is being truthful, I'll find out tomorrow...

At minimum I'll have to add two new standoffs for the two additional mounting holes (the case only features 4 mounting posts which are configured for Mini ITX).

Well, keep us informed.

You'll probably love the X10SDV-7TP4F for the most part. Its only real downside is the lowish clock speed. I've put these into some 2U SC216A. It turns out that if you are looking for a low power converged hypervisor/storage solution, this is pretty awesome:

24 bays: 16 bays attach to the 7TP4F for FreeNAS use. Drop in an LSI 9271CV-8i attached to the remaining 8 bays for fast RAID1 datastores.

2 PCIe slots: One for the 9271CV. The other for a dual 10GbE, because our formula here for hosts requires 4x 10GbE.

Onboard NVMe: SSD for cache (FreeNAS L2ARC, ESXi Flash Read Cache, etc).

 

[Chris Moore]

16 bays attach to the 7TP4F for FreeNAS use.

The onboard SAS controller is listed as a Broadcom 2116. Which appears to be a hardware RAID controller. Is that working properly with FreeNAS?

 

[jgreco]

The onboard SAS controller is listed as a Broadcom 2116. Which appears to be a hardware RAID controller. Is that working properly with FreeNAS?

Why do you say that?

off-the-cuff taxonomy for 2116 => XYZZ

X = {1 = SAS Gen 1 3Gbps, 2 = SAS Gen 2 6Gbps, 3 = SAS Gen 3 12Gbps}
Y = (for SAS Gen 2, breakdown is 0 = PCIe2 HBA, 1 = PCIe3 HBA, 2 = RAID-on-chip)
ZZ = lane count, 4 = 4 lanes, 8 = 8 lanes, 16 = 16 lanes

 

[Chris Moore]

Why do you say that?

I blame Google, and lack of coffee. I searched for that model but Google returned this page:
https://www.broadcom.com/products/storage/raid-on-chip/sas-2208
Which is clearly not the page I was looking for.

 

[jgreco]

I blame Google, and lack of coffee. I searched for that model but Google returned this page:
https://www.broadcom.com/products/storage/raid-on-chip/sas-2208
Which is clearly not the page I was looking for.

We'll forgive you. This one time.

 

[Constantin]

Well, looks like the package missed its connection in CT and barring any further mishaps, it will be delivered tomorrow. Not sure what happened between 8:33 in the morning on Sunday and 2PM Monday but my best guess is that the package sprouted legs and ran around in circles until it got tired and then the brave men and women of FedEx used brooms, fire extinguishers, and air horns to corral the package into a Boston-bound trailer, about 18 hours later.

FedEx is funny. Even when they miss their intended delivery window (today, 8PM), they still are on time (at least per the chat agent). Well, it's not like I'm going to get my free shipping refunded, so there's that. :)

Good news is that the other stuff has arrived (Optane, RAM, extra SATADOM, SAS breakout cables) so I'll probably work on extracting the current motherboard tonight so I can send it back to IXSystems in two weeks or whenever. I may also try to get a fresh copy of FreeNAS installed on one of the new 64GB SATADOMs and then transfer the config files.

 

[Constantin]

When it rains, it pours...

IXSystems just informed me that the replacement board for my failed C2750D4I has been shipped... via FedEx.

Another big thank you to IXSystems for their great customer support and faster turnaround time than initially quoted.

The other FedEx delivery should make it today, since the package status was “out for delivery” as of this morning. But you never know... that pesky package may go wild turkey on the driver, no doubt inspired by the wild turkeys in our back yard or the turkey waiting for its arrival.

 

[jgreco]

Well you know the reason for all the FedEx drama is because I redirected the package here.

 

[Constantin]

You may very well have succeeded. The iXsystems package with my replacement motherboard is here already, the other package has had its delivery time moved from "10:45-4:30" to "end of day". It's interesting how consistently this freight package has missed its delivery estimates and the only way you'd know / be able to document it is if you took screenshots because FedEx doesn't cop up to past estimates even though:

Code:

FedEx does not determine money-back guarantee or delay claim requests based on the scheduled delivery. 

:)

By the way, my smilie icons are all messed up. Ditto for you?

 

[Constantin]

Horray! Second package arrived. The board looks a bit shook up as the antistatic bag rubbed through at the SFP+ connector. I guess the gray foam didn't prevent the board from swishing around the inside of the retail package. The SFP+ connector still feels solid (no wiggling) yet the soldering associated with it and the SAS connectors strike me as odd.

I've done some PCB development in the past and typically, I'd solder metal elements like the cages on the SFP+ connector to the motherboard. That improves the mechanical strength and helps spread any mechanical load across a wider board area. Pins that penetrate through the PCB are soldered on the opposite side of the board. For example, with a connector mounted on the top, the pins are soldered on the bottom side of the board. Neither the SFP+ cage nor its pins, nor the pins of the SAS connectors appear to be soldered on the underside.

Here is the SFP+ Cage from the underside. The entry pokes out to the left.

And here are two SAS2 connectors from below.


It's possible (though highly unlikely?) that these connectors got somehow soldered on the other side? The vias are visible, the pins inside them too, but no solder on the underside.

Contrast the SAS2/SFP+ pin / via combinations with those of the the PCIe or RAM connectors, which all feature little circular solder tents on their pins, as expected.


Does this look right?

 

[jgreco]

Yeah, I'm with you. Take some pictures and mail them to Supermicro support and see what they have to say.

 

[Constantin]

Ok, will do. Thank you!

 

[jgreco]

Thought that comes to mind is that it's interesting that a set of conditional parts is unsoldered. That same board is, I'm pretty sure, used as the basis for a number of variants on that board. It's been a really long time since I've had more than glancing involvement with stuff like this, but I wonder if they might actually do some sort of two-phase assembly where they build the basic board, soldered etc., and then maybe later do a second pass for options.

 

[Constantin]

You are correct, that board has many different variants in terms of options:

• LSI SAS 2216 HBA
• Two stacked SFP+ cages
• Two additional 10GBe copper ethernet
• Two additional 1GBe ethernet ports
• A variety of D-15xx CPUs ranging from 2-16 cores.

The rear I/O cover reflects that approach also - while the holes are pre-punched for various connectors, the user has to remove the applicable port covers that are still attached to the shield.


The evidence suggests that Supermicro is customizing the boards in the US or Taiwan after shipping semi-completed units from China. That would help minimize inventory costs and help them stay very responsive to swings in demand.

 

[Constantin]

Just got in touch with WiredZone where I bought the board. They have been very helpful so far and they hope to get a response from SuperMicro soon. If this board really wasn't soldered correctly, what are the odds?