USB is not reliable enough for critical storage.
This is not true for USB4, which is based on the much more rigorous Thunderbolt 3 specification.
USB is not reliable enough for critical storage.
This is not true for USB4, which is based on the much more rigorous Thunderbolt 3 specification.
There is a serious difference between packaging software, distributing the software in straight forward installable images, and easy upgrades than what Open Source licenses require. Meaning iX could very well make the source code available per the relevant license.
But, that does not make it suitable for quick install. Take RedHat Enterprise Linux as an example. In the past they made the source code publicly available per the GPL license requirement. The Centos distribution then built it more or less the same as RHEL.
Yet, if iX supplied only the source, it would be incumbent on someone building the software and making it available. Most people would not bother building it themselves. Further, whoever built it, could favor Intel or AMD processors, making it less performant on the other. Or not run at all. iX more or less has to make it for the least common denominator of x64.
I am not sure what your point is.
Correlation⌠causation⌠iX selected components with licenses compatible with their value proposition. This approach does not mandate them to build additional middleware nor to maintain/support out-of-scope functionality for those of us on the fringes.
Iâd personally like to see more server-grade equipment support Thunderbolt attachments, and to see iX support more server-grade equipment regardless of architecture. Iâd also enjoy seeing them support more consumer-grade features wherever appropriate and feasible.
In service of those interests: I hope weâll keep this consumer-realm request light, good-natured and constructive. It would be a shame to tank a good idea through bad evangelism.
None of which are a suitable server platform for TrueNAS.
I have yet to see a Supermicro server board with Thunderbolt.
If we did that then I probably would not be running it at home.
So do I.[1] And Iâd accept to broaden the scope of search to current server boards from other manufacturers: AsRock Rack, Tyan/Mitac, the server arm of Gigabyte, etc.
Genuine server boards, no workstation.
On second thought, there have been âserversâ with Thunderbolt ports: The 2011, 2012, and 2014 Mac mini servers. Thunderbolt 1 and 2. Good luck with that! âŠď¸
This.
I think many failed to realize that the fact that FreeNAS and now TrueNAS can be installed on every day systems, does not mean it is a good idea for all, pending on the reliability you want or expect.
Now imagine if TrueNAS put in some code to block it from being installed on non-server based chipsets or something, everyone would lose their minds!
For use who use TrueNAS for free, we should be happy such a great product exists and does not lock us into an ecosystem like QNAP/Synology and gives us flexibility.
I should have said this explicitly: You are attempting to rewrite history by making it sound like we should feel privileged that iXsystems is deigning to allow us to use their product for free. The truth is that when iXsystems took over FreeNAS they committed to continuing to provide a usable free version of FreeNAS / TrueNAS.
Google Gemini:
When iXsystems took over the FreeNAS project in September 2009, they made a clear commitment to continue and support its open-source, community-driven nature. This commitment was crucial because at the time, there was a possibility that FreeNAS development might shift away from FreeBSD to Linux. iXsystems stepped in to sponsor and continue its development on FreeBSD, ensuring its future as a free, community-supported project.
Did they say anything about continuing to provide an installable / usable product as opposed to just making the source available?
Absolutely, iXsystems made it a very central point of their commitment to continue providing an easily installable and usable product, not just making the source code available.
The ongoing gate-keeping by MVPâs is sad to see. Please, where do I find the formal specification of what makes a suitable server platform?
How about this?
Apart from the official documentation there are a ton of resources on this forum covering a wide range of hardware related topics from networking to why you just cannot use USB drives or RAID controllers.
Thatâs not gatekeeping. If you install TrueNAS with USB drives for your data pool you have a very high probability of losing the entire pool.
We really try to keep people from making fatal mistakes.
HTH,
Patrick
Thereâs no gate-keeping. You can install TrueNAS on an old laptop and create a non-redundant pool on a single externally attached USB drive. So long as you understand the risks and caveats.
Look under the âUSB Hard Disksâ subheader if it doesnât expand automatically.
Youâre welcome.
@neofusion @pmh @winnielinnie
I was referring to specifically @pmhâs hand wavy use of âplatformâ, something that several MVPâs love to do when they want to poo poo something without providing supporting non-anecdotal evidence, not this discussionâs main point of the problems with USB. (Though millions of people use USB drives under heavy loads without issue.) (Or potentially with Thunderboltâwhich is clearly a support issue as opposed to a data loss one.)
The Hardware Guide is far from perfect or even complete:
It makes no mention of the problems that some have seen with using Realtek NICs or inexpensive SATA controllers, though it would be hard to quantify the level of actual riskâespecially given that fact that thousand upon thousands of NAS and server âplatformsâ ship with them.
It says nothing about motherboards or their chipsets. (Or the hardware bugs that they have just like any other hardware. E.g. Intelâs Atom Server CPUs.)
It confounds SAS expanders with the port multipliers seen on inexpensive DAS boxes.
It recommends, with no supporting evidence, against using direct attached copper (DAC) despite its widespread use in businesses and enterprises.
Its discussion of PSUs is almost useless.
It claims that âAMD CPUs are becoming more popular thanks to the Ryzen and EPYC (Naples/Rome) lines. Support for these platforms is limited on FreeBSD and, by extension, TrueNAS 13.â is misleading at best.
Excellent research. I had not bothered to do such.
To be clear, their are many companies that make statements in the past, that they later revise. Or outright completely deprecate.
My point about USB still stands, FOR NOW. That is the point of this thread. Not that you canât use USB attached disks, just suggesting you âavoid USB attached drives for data pool disksâ.
Their is a LOT of people writing forum posts, both old forum, (available via a link at the top of this page), and this new forum, that have had problems with USB attached data pool disks. That was WHY I wrote this thread, years ago. Perhaps less relevant today. Of course, people have had problems with SATA controllers and even SAS controllers. That comes with the turf of generic hardware, verses a well tested and well understood hardware & software.
For example, the Sun Microsystems V440 server was at one point Sunâs first, very high disk speed server, with 4 disk slots. It was quite un-reliable when it first came out. Turns out their was more than 1 problem. It needed updated: disk firmware; disk controller firmware; disk controller driver software; and disk driver software. Whence all those were updated, the 4 disks worked very well. But, use a third party disk, and find out itâs firmware was not designed to run at that high speed, and was flaky.
This is an example of what constitutes a server, well tested hardware and software. Throwing together random bits and bobs, like USB attached disks or USB multi-disk enclosures, and expecting them to work perfect, is path to potential problems.
For the last time, just because something works for 1 person, does not mean it is going to be reliable for a multitude of others. Just saying USB does not mean people will be buying quality products. The exact opposite seems to be the case.
Yup, I was at Sun during the days that we had to fix/modify drive firmware, even those used in workstations, to make them work reliably.
Add to that that with USB itâs not only a matter of âtried and trueâ or âworking out the kinksâ but many external enclosures for example cannot ever work reliably with TrueNAS if you want to use more than one disk.
Many of these present the same serial number for each disk connected, be it a single disk case and it presents a generic one for all units sold, or a multi-disk case that also presents the same serial for all drives.
TrueNAS will by design never work with these products, because it relies on the disk serial to manage the drives.
Iâm currently building my first truenas system (just waiting on the new drives). but iâve had running a raid array (mdadm) through a usb hdd dock via a mini pc running 24/7 for over 5 years now. itâs quite janky and i am slightly in awe that it hasnât caught on fire yet (but appreciative)
looking forward to upgrading to a truenas system in the next few days!