Failing to install comletely with new 8TB+ server

Virtualmin
1
SYSTEM INFORMATION
OS type and version Ubuntu Community 24.04 (26.04) LTS
Webmin version Latest
Virtualmin version Latest
Webserver version Proliant DL360p Gen8 & DLP380 Gen9
Related packages Nothing special

I’ve DuckDuckGo’ed around the Internets, and this forum, looking for a way to deal with this without a new post, but it seems I’m at another #EdgeCase again. Some Internet results seem like bad AI hallucinations. Others talk about the difficulty of most operating systems having trouble loading on systems over 1.9TB.

Maybe this isn’t important, but the little 360p has one backplane for the 8 drives, so it’s hardware RAID-6 producing about 8.97 TB. The OS thinks it’s one disk. The big 380 has 3 cages that hold 8 drives in each. That’s giving nearly 21.6TB of aggregate space with RAID-6, but the Gen9 has more RAID options than the Gen8 does.

The main problem I’m having is getting all of my RAID space to be available for the server. It might be a flaw with Ubuntu, so a better pick for the base OS could be my answer.

After my first OS+Virtuamin installations on both my newer servers, I shockingly discovered that I was only using 100GB of each server; the smaller one has 8.97 TB available.

My first move was to go into the Logical drive management and tell it to use 100% of the free space on the LVM partition. After doing that on both server installations, neither would reboot. They would start the normal reboot process, but get stuck after BIOS tossed-off to the OS.

Because of paranoia, I set out to make a new image on my USB installer stick and discovered that Ubuntu released 26.04 LTS, so I went ahead and installed that on the DL360p to see how it would go. I was running into the same problem with the size of the install, and I saw that message about Grade-B OS’es. :roll_eyes: I looked for all kinds of ways to try making a separate partition(s) that I could mount with the directories that would be the data hogs, but my ignorance “skirt” started showing.

It turns out that when using Ubuntu 26.04 that, either telling Virtualmin’s module to use 100% of the free space, or the Ubuntu installer’s obscure near-final-stages thing to use 8.970T instead of 100.000G would work, allowing all the space to be available for web server data.

I’d rather not have a Sword of Damocles dangling over my Virtualmin system as a Grade-B OS, but I don’t know what to do with this whole kerfuffle.

2

You’re talking about multiple unrelated issues.

Virtualmin has nothing to do with disk size. Virtualmin absolutely does not care about your disks. Whatever problem you have with disks is between your OS and your hardware.

So, removing that, it sounds like your question is “how do I use big disks in a hardware RAID with Ubuntu”, and my answer is, “I don’t know, probably more of an Ubuntu community question.”

That said, no Linux in the past decade or two has had a 2TB limit. That’s a very old limit, long ago surpassed by current filesystems. Current filesystems limits are measured in exbibytes, which is so large I don’t even know what it means. Filesystem Limits: Maximum File and Partition Sizes - Linux Bash

So, this seems like maybe it is a compatibility issue with the hardware. I haven’t dealt with server hardware in a couple of decades, so I dunno. I know in the past, it used to occasionally be challenging to use some RIAD hardware with any Linux version other than RHEL, because Dell or HP or Adaptec or whoever only shipped drivers for RHEL kernels. These days, the kernel has a mechanism for distributing drivers without them being built specifically for each kernel version, just the bindings get rebuilt and that’s handled transparently by tooling. But, you may have to install the drivers from your vendor.

I would recommend you leave Virtualmin and Webmin out of it until you have a system that boots and recognizes all of the storage. Remove variables until nothing is broken before making things more complicated. Partitioning can/should be done during OS installation, rather than later. Ubuntu has a graphical partition tool. It defaults to only using 100GB, but you can do a custom partitition layout to use all of the storage.

3

I haven’t worked with one of these in over 10 years probably. I seem to ‘kinda think’ there were different modules you could download that weren’t maybe standard in the kernel. Been a long while though. How old are these? The place I worked for had one that was in a couple of partitions because when they bought it there was no support for that large a volume. One of the techs did manage to convert it to LVM and that worked as I remember.

PS: I think the modules were actually for the management side of the server? Lights-Out?

https://www.hpe.com/us/en/collaterals/collateral.a50010868enw.html#toc-block4-l1

4

Looks like they were produced in and after 2012. Wikipedia on Proliant’s.

That’s why I’m wondering if there is a favorite OS to put under Virtualmin. I didn’t have a good reason to pick Ubuntu, other than I was already used to it. Virtualmin (I assume) would look and feel the same regardless of the OS. Maybe all forks of Linux are basically the same, but I feel like one might be different and better in this particular respect.
(oops) I didn’t know what RHEL meant. I thought it was some hardware driver. :flushed_face:

Another related topic I didn’t add details to is that the recommended sizes of partitions are seemingly impossible to find. I’d blame LVM. When I was first using Ubuntu, one of the hardest decisions to make was what proportions to point some of the directory tree to. I used to have detailed notes on this but that was at least decades ago. I vaguely recall /home /etc /var & /data were some of the things to point at other partitions before the installer would run, [citation needed]
I would adore a link to a resource outlining guidance on that topic from anybody.

5

Rocky and Alma a RHEL clones so I’d go with that. We used RedHat on them at the place I was at for awhile. HP is more likely to have modules for that.

Do you have the latest firmware?

6

TBH I don’t know. There were no complaints while I wandered all around in the BIOS.

P.S. your handle makes me want to change mine to PEBCAK, but mine would NOT be ironic.

7

Oh! That’s old!

BIOS often had problems booting off of big partitions back then. I’m pretty sure you need a smallish /boot partition for GRUB on a system that old. Beyond that, I would think the OS would take over and it wouldn’t matter, but hardware RAID complicates things. You need drivers for hardware RAID to work, and if they’re proprietary and never made it into the mainline kernel, I’m sure they’re no longer maintained…and, you may be SoL on using the hardware RAID controller. If that’s so, you might be able to use it as a regular controller, for all disks and use LVM to setup the RAID type configuration. (Which will be faster, anyway. Linux software RAID options surpassed all but the fastest most exotic and expensive hardware RAID controllers in performance two decades ago.)

But, once again, it’s been two decades since I was involved in IT and big hardware (big for the time, you can get a lot more than 8TB on a single disk these days, I’ve got 18TB of storage in my desktop), so I don’t know anything.

You still need to be wondering what OS you can put on top of that hardware to make it work. I wouldn’t involve Virtualmin in your decision if you’re using cranky old hardware. Virtualmin doesn’t care all that much. I like EL distros better than Ubuntu, but I usually recommend people use what they know best. In your case, you should use what you can make work. (But, EL is more likely to work on old enterprise hardware.)

8

I would not trust raid hardware that is 12+ years old. Old bios, especially pre UEFI always have weird bugs and lots of limitations.

9

I think the latest BIOS update for this that I saw was 2019.

10

You were definitely onto something. That Gen8 server has the E5-2603 processors in it. In the RHEL line of software I had to go back to Version 8 to have the instruction set for those processors supported.
Deprecation of x86-64-v1 and x86-64-v2 x86_64 microarchitecture CPUs in RHEL9 / x86-64-v3 is required by RHEL10

Since the Red Hat was $$$ to use it, I’m going to go with AlmaLinux v8 for a while to see how far I get.

Just noticed this about the difference between x86_64 & x86_64 v2 for the AlmaLinux 10 installer choices.

Official AlmaLinux 10 is x86_64-v3 like upstream. Besides that, AlmaLinux also offers a separate version with support for x86_64-v2, which is nice.

I don’t know who I’m quoting, but “Complexity lies at the edge of chaos.”

11

Would that matter if I’m only using it to host a private MySQL server with a project that only I have access to? I’m in the not ready for prime time with a clandestine operation on it.

12

The processor is not a problem, current Linux supports all the Xeon processors (they only removed i486 support this year!). You should use a current version of Alma (10).

I was talking about RAID drivers, which might be an issue that requires an older OS. But, if you have to choose between using the RAID controller and a new OS, I would recommend the new OS.

13

Neither the Red Hat or Alma version 10 installers would run on that machine. They would not run the installation script to the GUI, but instead the machine would just reboot and drop to the first dummy screen that gives the option to install or test the installation media and then install.

14

Ah, that sucks. Do you have an option to make your RAID controller not a RAID controller? Most can act as a regular multi-port SCSI or IDE controller. That would probably work with newer Linux.

15

Let me start with I would ignore every post above, your over thinking this way too much. we have some old servers back as far as g6 and g7 running hardware raid with more then 10 TB of disk space and do NOT have the issue you are speaking of. G8s also work with no issues.

Whilst I’m not going to comment about redhat and the derivatives as we don’t use them I will say debian in all current and previous versions run just fine, and hence I would doubt very much that redhat wouldn’t work either.

Take a breath, and try again.

16

Yeah, I doubt there’s any reason any major Linux won’t run on that hardware.

17

You think getting the latest firmware is wrong?

18

No that’s not wrong, but has nothing to do with his problem.