Correct.
As for boot drives, any cheap and small M.2 NVMe will do.
2 Likes
My optimism comes from the viewpoint of: The WD SSD is not a SATA drive and, therefore, it doesn’t “block” the sata_0 line.
1 Like
That would be right. I see it accepts SATA or pcie/nvme.
1 Like
Data from the SATA ports is going to a controller, and the controller itself has PCIe lanes back to the CPU. So it makes sense for the SATA port to be disabled if an M.2 slot is populated, especially if, as it appears in this case, that slot is a second slot (the primary one likely has its own dedicated lanes).
Note also that depending on how you populate the PCIe slots, the x16 slot will get downgraded to an electrical x8 slot. Nothing wrong with this, just something to be aware of when you’re planning your layout.
@etorix , any thoughts on whether the onboard SATA has any issues to be aware of for TrueNAS use? It should be, since this is a server motherboard, but I’m so used to having to add an HBA…
@Ricky1979 , for people who find this later, would you mind updating your original post to indicate that you went with a different motherboard? I just came back here to reference this build and almost forgot you’d switched it.
SATA controllers in Intel/AMD chipsets are fine.
I think, for being ONLY a NAS, this config is more than capable.
Does not TrueNAS on bare metal spin up disks one-by-one?
Depends on the details. A long time ago my Adaptec SCSI 2940AU adapter did this by default with the HDDs I had.
On the other hand the Supermicro X9SRi-F server board I have now does not do it.
No, TrueNAS as an OS does not stagger spin-up.
Your controller may do it, check the settings. TrueNAS will not interfere with that functionality.
So, those of us using mainboards with on-board AsMedia SATA controllers probably aren’t getting staggered spin-up. 
Just a thought, why don’t we create a new category of tested combination of builds, where people can list the hardware combinations that are tested and working and any known issues. It might be helpful for such consultations right?
Anyone else think this is useful?
1 Like
I like solutions, but I think people would still post to ask.
There’s millions of combinations that will work. Maintaining a list would be an insane effort.
All the information that’s in these threads is information that already exists on the internet. If people aren’t willing to research, then I don’t expect them to look through a list of parts either 
You need a CPU, RAM, motherboard, and storage. Avoid SATA multipliers.
1 Like
This would be great–TrueNAS-specific PC Part Picker build section.
It’s very easy to burn a lot of time and money trying to teach yourself from scratch with TrueNAS–both the storage part and the virtualization part. Entirely self-directed research when starting from zero is not as useful as we’d like it to be–especially since a lot of the most often repeated information on ZFS hardware requirements surfaced on Google in wikis and and elsewhere tends to be years if not a decade out of date (see, e.g., older posts on how easily ZFS kills anything but the most expensive SSDs).
I’m always going to encourage people to ask for help. I almost always learn something when I try to help them, and we’re all here to learn.
Having a searchable collection of known working systems–particularly those put together by people in home/small office environments without huge budgets–would be an amazing, constantly updating teaching tool.
Maybe start a thread in General Discussion - TrueNAS Community Forums ? I don’t see a category specific to suggesting improvements to the forums themselves. @HoneyBadger or @kris might have a better suggestion for where to submit … suggestions.
…That sentence was not so great. More caffeine.
There is a document made by a community member that aimed to do just that, from a few years back.
Even though it may be somewhat dated by now, the specific hardware picks are still relevant when buying second hand. So is the reasoning behind the recommendations.
There’s also this, almost too comprehensive resource that appears to try to create one guide to rule them all:
How much testing and validation would be required?
Ten minutes by one guy is not enough. Millions of man-hour is out of reach.
If a description of how the build was tested is included, the reader can decide if they’re confident in the build.
To put some more structure to it, we could devise an easy to do standard suite of tests for people who want to submit builds, and have them post the results from that.
Datapoints people might be interested in: power draw (idle/heavy use/start up); thermals (CPU, drives, optionally 10 GbE NICs); noise (pick a noise measuring app that exists for iOS and Android).
1 Like