DrKK's New FreeNAS; Build Experiences

[DrKK]

Final Update:

Pictures are coming.

I took extra care to tie back those cables as far as I could without stressing them so that putting the side panels back on would not be as significant of a stress point. So far, everything is running great.
Everything up and running, we've seen some of the guys come through the Mumble server on daneel. The mumble server is open to all in the FreeNAS community by the way, if anyone wants to stop by. PM me in irc or here and I'll give you the digits...

Currently cleaning up the dining room.

 

[Stux]

Yes, yes, OH YES, yes, yes.

Indeed yes.

This.

Whew.

Not least because existing ZFS documentation would then be applicable to FreeNAS, potentially saving significant effort reinventing wheels that could be more profitably spent elsewhere.

And as someone who thinks in ZFS terms, it can be quite confusing when I have to translate into the strange FreeNAS terms... oh ... pool... i mean Volume... that's right.

 

[J. Bowman]

...they have a nice fan controller for 3 case fans with 3 speed settings. I decide to use this post and the hardware guide, instead of the mobo fan headers...

Is this an artistic choice, or are there other considerations at play here? I've been working through my first build (using this post and the hardware guide), and my Node 804 has the same controller. I wound up using the mobo headers, because I didn't have enough SATA power cables, but that is a remedied situation.

 

[DrKK]

Is this an artistic choice, or are there other considerations at play here? I've been working through my first build (using this post and the hardware guide), and my Node 804 has the same controller. I wound up using the mobo headers, because I didn't have enough SATA power cables, but that is a remedied situation.

Pro forum tip: If you want to quote a post, highlight the text in question, and a "REPLY" button will float up that will quote, and backlink, the referenced prose. Much better than whatever you just did, which seemed to inadvertently also quote a piece of what you were writing, which confused the hell out of me :)

So on this case, the built-in fan controller had 3-pin fan sockets already installed and dangling; I simply connected the already-installed case fans' 3-pin connectors. I think all I needed to do was provide a SATA power connector to the case's fan controller (don't quote me, I can't remember for sure now). My logic for not using the mobo headers was twofold:

1) Everything in here is well-ventilated, and I have a pimp cooler on the CPU. I'll never need to "spin up" any case fans, in fact, I probably barely need them at all. Since they're going to be more than sufficient on the lowest setting of the built-in controller, why not use it?

2) I have no understanding at all of how the Supermicro boards actually control these fans. I don't appear to have fine-grained, reliable, control over them. e.g., when you send the IPMI raw fan commands, it appears to ignore them some of the time. I just can't be bothered with that. Since I don't appear to gain anything from putting them on the mobo, why bother.

In all previous builds in my life, I found myself unable to NOT install fans in every single fan mountpoint. If I saw 120 or 140 mountpoints, I bought a fan for them. A bit OCD perhaps. But I cured myself of this in this build.

 

[Stux]

SuperMicro boards only support controlling fans via PWM, whereas some consumer level boards can also control fans via variable DC voltage, which allows fan control with 3 pin fans.

If you don't want to run wizzbang fan control scripts, then you can go ahead and use the Fractal fan controller and just set the fans to a fixed speed.

 

[J. Bowman]

You had me at "wizzbang fan control scripts". :)
Seriously, though, I'll probably switch my plugs over to the case controller once the extra cables get here, which should be tomorrow.

 

[Stux]

You had me at "wizzbang fan control scripts". :)
Seriously, though, I'll probably switch my plugs over to the case controller once the extra cables get here, which should be tomorrow.

There's a bunch of them

This one is mine:
https://forums.freenas.org/index.php?threads/script-hybrid-cpu-hd-fan-zone-controller.46159/

 

[DrKK]

There's a bunch of them

This one is mine:
https://forums.freenas.org/index.php?threads/script-hybrid-cpu-hd-fan-zone-controller.46159/

I never got satisfactory results on my X10SLM (my previous FreeNAS) with the various fan scripts. They would certainly work after a fresh cold power up of the IPMI, but then, after some period of time, I found they no longer controlled the fans as expected, until the new cold restart.

I don't know if my X10 was glitchy, but I didn't have enough patience to mess with it any further.

 

[Stux]

My script has a fair bit of logic to deal with that.

Ie it periodically checks the results and resets the BMC if necessary

 

[DrKK]

My script has a fair bit of logic to deal with that.

Ie it periodically checks the results and resets the BMC if becessary

Fancy.

But it offends my sensibilities. It's supposed to work, without needing soft/hard resets. Engaging in a workaround only justifies the incompetence!!!!!!! SuperMicro should be using BMC's that work! :)

 

[Stux]

Fancy.

But it offends my sensibilities. It's supposed to work, without needing soft/hard resets. Engaging in a workaround only justifies the incompetence!!!!!!! SuperMicro should be using BMC's that work! :)

At least you know the failure modes have been tested ;)