TrueNAS refuses to install onto 8 GB SATA DOM - Says drive too small?

[aaronouthier]

For some reason, I thought the minimum requirement for a boot drive was 8 GB?

I am trying to install TrueNAS core on a SuperMicro brand SATA DOM of exactly 8 GB. MoBo is a SuperMicro x8dt3-ln4f with 64 GB ECC Registered RAM and 3x 1 TB drives, and old 360 GB drive + the 1 SATA DOM. I seem to recall having the same issue since FreeNAS 10 or so. It installed fine in FreeNAS 9, and worked wonderfully when upgraded from 9 to 10. When I tried to reinstall 10 from scratch, however, I had a similar message. That was years ago. Was frustrated enough that I moved away from FreeNAS entirely. I would so hate to waste 360 GB on a FreeNAS boot partition. I would much rather use it for storage.

Now, I would like to try it out again, but it still won't install! Am I truly not meeting the hardware requirements, or is there a Bug in the installer?

I am including a screenshot from IPMIview showing the drive selection screen in FreeNAS. I am noticing the installer thinks it is only 7.4 GB in size. It is definitely exactly 8 GB however!

 

[jgreco]

For some reason, I thought the minimum requirement for a boot drive was 8 GB?

That's what the docs say.

I am trying to install TrueNAS core on a SuperMicro brand SATA DOM of exactly 8 GB. MoBo is a SuperMicro x8dt3-ln4f with 64 GB EEC Registered RAM and 3x 1 TB drives, and old 360 GB drive + the 1 SATA DOM. I seem to recall having the same issue since FreeNAS 10 or so. It installed fine in FreeNAS 9, and worked wonderfully when upgraded from 9 to 10. When I tried to reinstall 10 from scratch, however, I had a similar message. That was years ago. Was frustrated enough that I moved away from FreeNAS entirely. I would so hate to waste 360 GB on a FreeNAS boot partition. I would much rather use it for storage.

Now, I would like to try it out again, but it still won't install! Am I truly not meeting the hardware requirements, or is there a Bug in the installer?

I am including a screenshot from IPMIview showing the drive selection screen in FreeNAS. I am noticing the installer thinks it is only 7.4 GB in size. It is definitely exactly 8 GB however!

No, the installer thinks it is 7.4GiB (gibibytes) in size, and the installer is certainly correct. "8GB" means 8000000000 bytes, using drivemaker's preferred definition for gigabytes being 1000^3 bytes, whereas a gibibyte is 1024^3. This certainly sucks.

It is possible that the installer is enforcing 8GiB instead of 8GB. If so, I would file a Jira ticket, see the "Report a Bug" at the top of the page. Either the documentation or the code should be fixed, and it will likely be the documentation.

FreeNAS started out requiring 1GB many years ago, and it was quite annoying every time it's been bumped. I will note that 8GB is *really* tight, so you might just want to go out and get a larger device. Dell's 64GB modules are going for about $25-$40 on eBay, depending on used or new.

 

[aaronouthier]

Um, what? I find your response quite puzzling.

Any school kid knows that a Gigabyte is 1024 ^ 3. Computers use binary, so they count in exponents of 2. 1 2 4 8 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 ... Thats as far as I have it memorized. I'd need to look it up beyond that.
You say that a gigabyte is 1000^3 bytes, but then in the next sentence you say that it's 1024^3. What kind of doubletalk is that? Or did the definition change since the 1980s when I learned this!!?? People have already redefined the definition of sick from disgusting to a good thing. I can handle that (even if I don't like it). People are even confusing forward and backslashes. You don't put backslashes in a URL people, unless you're doing windows file sharing from a windows box (ie. \\host\share\file.ext).

Sorry, I'm ranting again. It's just that I hate learning something, and years later, find out someone changed it on me. I hope that's not the case here. I guess I don't get out much. Perhaps I'm too much like my hermit uncle. Time will tell.

 

[jgreco]

Um, what? I find your response quite puzzling.

Any school kid knows that a Gigabyte is 1024 ^ 3. Computers use binary, so they count in exponents of 2. 1 2 4 8 16 32 64 128 256 512 1024 2048 4096 8192 16384 32768 65536 ... Thats as far as I have it memorized. I'd need to look it up beyond that.
You say that a gigabyte is 1000^3 bytes, but then in the next sentence you say that it's 1024^3. What kind of doubletalk is that? Or did the definition change since the 1980s when I learned this!!?? People have already redefined the definition of sick from disgusting to a good thing. I can handle that (even if I don't like it). People are even confusing forward and backslashes. You don't put backslashes in a URL people, unless you're doing windows file sharing from a windows box (ie. \\host\share\file.ext).

Sorry, I'm ranting again. It's just that I hate learning something, and years later, find out someone changed it on me. I hope that's not the case here. I guess I don't get out much. Perhaps I'm too much like my hermit uncle. Time will tell.

Please re-read my response, and, this time, please pay attention to the little details.

Any school kid knows that a giga-anything in SI is 1000^3, just like kilo-anything is 1000, and mega-anything is 1000^2. This is the normal exposure school kids have to sizing prefixes.

I did not misspell "gibibyte", and meant what I said.

For better or for worse, compsci has misused "KB" 4E4 (remember when that number was big enough and computers were slow enough that we used 4e4 in loops to mean run them "forever"?) If a kilobyte is 1024 bytes but a kilometer is 1000 meters, that's confusing.

Now I'm old enough that my first computer didn't have a kilobyte of memory; I still have a Synertek SYM-1 with 1KB of RAM here in my office. At 1KB, the small difference between KB (1000) and KiB (1024) didn't make a big difference, and, yes, we all called it a KB, "incorrectly". It's only a 2.34% error. The problem is that this "rounding error" magnifies as the powers increase, so when you go to MB/MiB it's a 4.63% error, and at at GB/GiB it's 6.87%, and at TB/TiB it's 9.05%, give or take a bit.

It was really the hard drive manufacturers who started to "correct" the terminology, because they wanted a way to sell a larger-sounding drive, so they decided to define "GB" as the SI 1000^3 unit when contemporary compsci usage was GB = CS 1024^3. They were able to do this because there isn't a natural powers-of-two thing for hard drives; you have cylinders, heads, and sectors transformed into LBA.

"GiB" is an unambiguous term as it absolutely means 1024^3, which at least nails that down.

I don't think there's any winner here. CS still largely uses GB = 1024^3 for lots of things, including memory sizing, but storage has been sold using GB = 1000^3 for many years. Interestingly, this plays to strengths in SSD, where a "500GB" SSD is 500,000,000,000 bytes but the backing flash is probably 512GiB, so while you might think that this leaves "12GB/12GiB"-ish for the free page pool, but 512GiB works out to 549,755,813,888, or 49,755,813,888 bytes underprovisioned, which is in the range of 48GB/48GiB-ish. That's the best positive spin I can put on this. ;-)

I am not discussing the way I'd *like* the world to be. I'm discussing the way it *is*.

 

[aaronouthier]

My apologies. I wasn't trying to be rude. It's just that this is a big deal for me and a game changer. For years now, I've seen references to GiB, and always scratched my head. I seem to have missed the difference when you wrote "Gigibyte" (you may have figured that out already ;) - this is the first time I am aware of coming across that term spelled out. Then again I didn't see it a few hours ago when you used it, so anything is possible at this point.

This is all so much for me to process right now. It is 4 AM here in my native California. Finding out that something I thought I knew for almost 40 years doesn't mean what I thought it means, well that is just inconceivable. (Seems I fell for one of the classic blunders?)

I was somewhat of a child prodigy. Learned to program on the Apple II series (Apple ][e, specifically), by the time I was 10. Never had to try too hard to learn if it was interesting to me, despite a diagnosis of "learning disabled". I was too stubborn to let that stop me...

All-in-all, it was not my intention to challenge you or insult your understanding. It's just hard for me to have my world turned upside down and shaken, and not walk away feeling like I've been mugged. It was nothing personal against you. I was always the "smart kit", and the one people would go to to have computer questions answered. This will certainly be a day to reflect.

Before, I had always seen drives show up smaller than the advertised size, but was told early on this was due to the "formatted" vs. "unformatted" capacity. Ie, the advertised size, minus the size of the MFT, FAT, or superblock, etc. comes out to the size reported by the linux 'fdisk -l' command. I'm just realizing this is probably not correct either...

I'm going to try to get some more sleep now. If it doesn't come, heck, it might just be a good day to stay in bed...

 

[jgreco]

No need to apologize. I find GB/GiB annoying too.

It is certainly just one example of getting shortchanged in the grand scheme of things. As you note, we've seen things such as formatted vs unformatted, which wasn't limited to metadata -- recall MFM vs RLL? Same drive, same electronics, different encoding, resulted in much greater density. Or CMR vs SMR.

We are, however, in a fantastic era, think of how stuff has changed. 8 bit computers with 16KB of RAM and a 1MHz clock speed, storing things on 140KB HDD's for $2500... best not to get too uptight about the little stuff.