New 32 4TB HDD Server

[jgreco]

For what it's worth, under Free BSD 8.0 beta, with the beta version of ZFS, the crappy cheap SATA port multipliers in my numerous massive drive chassis worked flawlessly.

Well anyways hi, nice to meetcha, I hear a lot of your frustrations and I even share some of them. I'm not going to try to address your post on a point by point basis as it is kind of a mix of fact, perspective, exasperation, and even some error. jkh can speak up if he wants to but I don't wish to engage on the topic, except to say that yes, those of us who are busy deploying FreeBSD have at times had to work around the bad parts. Most definitely including FreeBSD 5 (didn't get further here than the lab) and FreeBSD 6 (got a small number of systems in production before deciding it was a SMP train wreck).

As someone who's done some device driver work in the past, though, and who has extensive experience with PC hardware, I'm going to go out on a bit of a limb here and point out that Linux also doesn't have support for "all hardware," and that the PC market is plagued by its inherent nature. It is actually very difficult to cope with the stream of knockoff hardware that emits from Shenzhen and other tech manufacturing ghettos, particularly when they work so hard to manufacture stuff that looks authentic (including knockoff Intel ethernet controllers, LSI HBA's, etc). And that's on the server side of things, where there's substantial profit potential. I have no idea why someone would try to manufacture something like knockoff Prolific PL2303's but apparently they do. Stuff like SATA controllers and SATA port multipliers are even worse as they sit in that middle ground of stuff that is just profitable enough... and as long as someone can code a Windows driver that works plausibly well for at least an hour or two between crashes, it's deemed acceptable quality. It's often not even knockoff grade, it's just what I like to call "PC grade."

Given enough coders and enough vaguely broken hardware, yes, FreeBSD could be made to work on it all. Linux has more coders working at more edge cases these days. FreeBSD is a smaller, higher quality project, in my opinion - it does not strive to work on every possible bit of PC hardware, and that's probably a good thing in some ways. But it means that you build a machine to suit the OS, rather than picking random hardware and then expecting the OS to behave flawlessly.

We know certain things are well-supported and will work well. Twenty years ago that was an Adaptec 1542 and a SMC ethernet card. Today it's an LSI HBA and an Intel ethernet card.

Cyberjock is busily guiding as many people as he can to the things that he knows works. I respect that. It's quite an undertaking and he does a hell of a job. Please don't be too critical of him for providing his opinion to you, for free, as a member of the community. The people who are "official" respresentatives of the company are conspicuously labeled as "FreeNAS Core Team" or otherwise readily identifiable as iX employees. Please remember that the rest of us are here voluntarily and aren't paid by iX to provide assistance to others in the community.

 

[cyberjock]

Bugtexen,

I'm an engineer too. And you know what, that comment that cheap stuff often doesn't work. It's true like 99% of the time. And instead of trying to keep a list of what does work it's FAR simpler for a project of this magnitude that has this kind of growth year after year to have to cut a few corners to provide the most accurate answers. If you don't like the attitude because you think it's too political or religious, feel free to leave. I guarantee you I won't miss you and you seem to make it pretty clear you won't miss us.

I've seen tons and tons of people lose their data. I've gotten a general feel for what hardware works 99.9% of the time, and I've gotten a feel for what hardware doesn't work 99.9% of the time. Yeah, there's always "that one firmware" or "that one revision of the card" but I don't have time for that kind of thing. We see a ton of complaints about Realtek it's easier to just not recommend Realtek.

And trying to argue that "we don't have working drivers for card x" versus "card x has some problems" is you just crying because you can. Saying card X has some problems is correct. It has problems. Do I know what the problem is? Nope. Do I care? Not really. If 50% of Realtek users have problems then I'm gonna start not recommending them. And for Realtek they'll gladly make you a custom NIC for you with a unique model number that you'll never see anywhere else on the planet. So what do I do then? Keep an infinitely long list? Absolutely not. I push people in the right direction. If this doesn't make you happy because you want 100% accuracy on what works and what doesn't feel free to contact iX and buy a TrueNAS box. Those work well. So when I say card X has problems it's not blaming any side. I'm saying it has problems with FreeNAS. How you want to interpret it and spit it out because you want to be a linux shill here is totally up to you. But I can solve your problem for you.. the ignore list.

If you want the facts feel free to read the last 2.5 years of threads. I did, and those last 2.5 years is all the "fact" I have to provide, and it *will* back up my claims. I'll provide as much good advice as I can within the limits possible for "free". If you want to start paying me a 6-figure salary to maintain infinitely long lists of what works I'll gladly start providing a list that is more accurate so you can sleep better at night knowing the forums are 100% accurate. Until then realize that free is free.