Hi all, I am just a beginner home user and have been using the CentOS version of Rockstor for about 3 years. I have a Dell H330 as my SAS controller which I run in HBA mode and works perfectly fine with CentOS Rockstor.
I recently got some new drives for it and thought I would upgrade to the openSUSE version, but I get a bunch of errors on the install with it not being able to detect the SAS controller.
I reinstalled CentOS Rockstor and have it all up and running again no issues. How would I go about getting the driver from CentOS working in openSUSE so I can use the latest version of Rockstor?
@Lats welcome to the Rockstor community. A couple of questions for you:
I assume, you used the latest version available on the downloads page to attempt the installation?
Is the H330 just for your storage pool and you’re using something else for your OS installation, or is it all under the SAS controller?
If the OS disk is on a separate, e.g. SATA connection, when you performed the installation of a USB stick (I assume), did you disconnect the SAS drives beforehand? If not, I’d attempt the installation first without SAS in the mix. That way, it could help isolate other underlying issues that need to be addressed first.
Before you reverted back to CentOS, did you note down what errors you were encountering?
Fundamentally, I think a SAS controller in HBA mode should be recognized by OpenSUSE and all that, but I am not using one myself, so I can’t prove that.
There was some recent discussion here:
between @simon-77 and @dadozts that involved SAS & HBA, maybe they can some additional hints there? Of course, not to forget @Tex1954 who’s a hardware fiend.
Finally, I am not recommending sticking with the CentOS version, mostly because the btrfs tools + kernel have come a long way since then with all kinds of improvements from a speed and especially resiliency and recovery perspective, which is obviously important for your data (well, at least I think it probably is).
If, as I asked above, the OS disk can be isolated and you attempt the installation again (take out the CentOS one and put a new small SSD in, if you can, as well as leave the SAS drive cables disconnected) you can post more of the problems you run into without compromising your data pool. that way you can switch back quickly to your CentOS version when you don’t have time to play with the OpenSUSE install and don’t have to run any reinstalls.
Also as a sidenote, you can run a backup of your Rockstor configuration on the CentOS version, which you should then be able to use, once the OpenSUSE install is successful, your SAS disks are connected and the pool is imported (check the documentation).
Good morning, I don’t think I can be helpful. I simply inserted my SAS HBA card, an LSi 9220-8i, on the ASUS motherboard with B450 chipset, connected my HDDs and SSDs that were previously connected to the SATA controller and when I restarted the HBA card configuration option appeared, I didn’t touch anything, Rockstor started up and everything was there as before, perfectly working as if I hadn’t touched anything.
The only thing I can add is that the small SSD with the operating system remained connected to the motherboard controller, for a question of order in the cables in the case not a technical choice.
@dadozts and I where both using a SAS controller, but connected to SATA drives.
SAS and SATA are only compatible in this way, a SAS drive can NOT be connected via SATA.
And also as already mentioned, this was plug and play for both of us.
Was it the exact same hardware configuration as before trying to install Rockstor on openSUSE?
Or do you have e.g. unplugged your new drives when reverting to Rockstor on CentOS
Hey All, thanks for getting involved. I might have lead everyone astray here now. Last night I tried to install the openSUSE version again and ran into errors on install. It has been a while that I went back to using the CentOS version and I tried a bunch of other OS installs in the process looking to find something that worked but all failed, so I went back to the latest version of CentOS which works perfect.
To answer a few of the questions:
I am installing from a USB flashed with BalenaEtcher on Windows 10.
I downloaded the latest version of Rockstor Leap from the website.
I do have a separate drive for the OS running from the motherboard SATA controller set to AHCI.
I am not using the old and new drives simultaneously - under CentOS using perccli, all the drives show as good and all can been seen and pooled from the web interface.
No hardware changes between the installs.
I attempted a new install last night of Leap, but it stalls - image attached. I thought it was something to do with the SAS controller because the line it halts on says SCSI, but it looks like it is referring to the USB?
This text appears after the Loading Rockstor screen, so I never make it to any of the setup.
I tested the install with 3 different USBs one being the USB I installed the CentOS version from. I have also tested removing the HBA completely as well so it is nothing to do with that, sorry all.
Probably should have checked over all this before making this post.
Interesting. Thanks for the additional detail.
How big is the target disk you’re trying to install Rockstor on? Also, is the target OS disk completely empty (i.e. fully wiped as recommended in the documentation)?
I did a full wipe this morning and tried again but got the same result. I also tried installing Tumbleweed but got the same result as well.
The drive I tried to install on was 80GB but now I have the CentOS version successfully installed on there and I am trying to install Leap onto a 500GB drive.
After around 5 minutes or more it kicks out to emergency mode - image attached.
I saw a bunch of people have this error on the forum but I couldn’t find someone with a solution yet.
Just another note, I am trying to install at the moment without the PERC card installed.
And finally, when looking around a bit more, one of the solutions that seemed to work for multiple people (not on Rockstor but other Linux installations) is to force a regeneration of the initramfs. So, once/if you noted the info from the two commands above, you could try run this command chain at the emergency prompt:
Worst case, it won’t work, nothing is lost, since you’ve not been able to get it to the installation screen anyway, and it could possibly give some further insights what else might be wrong.
Thanks for the options Dan. I first tried the commands you suggested but neither dracut or grub2-mkconfig are commands in the emergency prompt, unless there is something else I need to do to get them to work.
I ran both and the output is also displayed in this image
I then put the hard drive into another computer and the install all went fine, I logged in as root and it gave me an IP address.
I then took the hard disk out and put it back in the server. The boot sequence looked different, but it still ended up in the emergency prompt. And still seems to be a problem with the drive labelling. Again, tested dracut and grub2-mkconfig but they don’t work at this stage of the boot.
That’s what I get for recommending something that can’t try out myself . Not sure what’s missing. It might require some mounting of volumes first, not sure.
Interesting, that you’re able to install it on the drive (so, the small chance of a second drive being bad or having some problems essentially eliminated.
I’m wondering whether a BIOS setting could prevent the successful boot. I’ve seen some instances (again, not physically verified) where the “secure boot” setting in the BIOS could prevent something. Also, did you specify the boot sequence in the BIOS to hit the hard drive first (or first after, say USB boot)?
Finally, when you installed it on another machine, did that one have UEFI booting or was it in legacy Boot mode, and the system you try to get it working in, does it run UEFI? Actually, what system are you running for Rockstor, is it a PowerEdge Server (which the older versions I think, by default don’t have the BIOS set to UEFI, which I would recommend to change to), too?
Thanks Dan. The system I have been running for the last 3 years is a bit of a hack together. It has a GA-H77M-D3H motherboard. The PERC card is something I have added along the way and the system all works with the CentOS version.
I put the drive back in the other computer I have, but it actually didn’t install correctly, I just thought it did because I went through the install process and I got to log in. But the second boot it goes to the emergency prompt. This system has an ASRock Z77e-itx.
Both run UEFI bios. I enabled fastboot on the GA board and it let me go to the install process now. In the BIOS I have UEFI boot only enabled.
It gives me the login screen after and I can login as root. Reboot give the recovery shell. In grub I can see 4 partitions, partition 2 is the EFI and partition 4 is BTRFS and seems to have a bunch of things install. The other two partitions are unknown file systems.
There is also no fstab.
It could be something in BIOS for sure, but CentOS installs no worries. Could it be fixed from the emergency shell. The error is still the same it can’t find the drive by id. I see grub.cfg is using the search command to find the disk and assign it to the label ROOT.
I am away for the next few days, so won’t be able to give anything a try over the weekend.
. Not sure what’s missing. It might require some mounting of volumes first, not sure.