I have not been able to run any Segate tests on the drive. My combination of hardware doesn’t seem to want to work with their software, unfortunately.
I am going to reach out to the people who sold me the equipment to see about a replacement, as it seems likely that the drive itself is at fault here.
That said, until I actually send it away, I’m more than willing to keep testing.
Hi,
thanks for your attitude. Can you please confirm that when you say you have downloaded the image again, was it the 2.38.3+rev4 version as Florin suggested and the issue came up again?
Ah, no, I have not tested the older version yet. I got lost in trying to test the hdd. But I will turn to that now, and report back when I’ve tried it.
OK. That was fairly quick, compared to the previous flash attempts. It is up and running on 2.38.3+rev4, as downloaded from a starter application created on my dashboard.
Update: just to verify, the journal shows no problems with sda6, which does appear to be resin-data, as would be expected.
Hi. Can you share the exact hw specs of the NUC you have? Maybe a link to where you bought it from?
Hey @Caligari,
thanks for working with us to resolve this issue. There does seem to be an issue with the partitioning script on 2.50.1+rev1. We would be gratful if you could help us debug this. Here is how you can help:
Thanks,
Rahul
I’ve tried a 2.50.1+rev1 build from my dashboard (so, not the one that works only locally - I can try that again if needed).
Here is the contents of initramfs.debug:
initramfs.debug.log (2.2 KB)
I hope this helps. I’m ready to do whatever else you need.
(Edit: to be clear, this came up as most of the others attempts - without the server.)
Is there any other additional information I can provide?
That should be enough for now, thank you. I’ll forward this to the OS team.
Hey there Liam,
Thanks for your continued support in debugging this issue.
This seems related to resize2fs having trouble with large disks, that leaves the filesystem unmountable.
The OS team is looking into it and we will inform you once we have more details on this. There is also a ticket you can track if you wish https://github.com/balena-os/meta-balena/issues/1977
Thanks
Hi Liam, just a final question. Could you please give us some more details about the disk? Specific brand/model would be great, but if not just interface type (SATA/NVME), capacity etc.
Great to see this progress, and great to have been abel to help!
The disk I have is a Seagate 2TB BarraCuda 2.5" hdd model ST2000LM015 (SATA III). Amusingly, I kinda wanted a smaller one. 
So, ideally I’d like to spend some time working with the non-dashboard version of balena OS. What are my options? Do I have to use the earlier version? Is there a way to download that (as it isn’t listed on that downloads page)? Am I able to upgrade from an earlier version to a later version, in situ? Or is that really just a reinsall, which we know won’t work at the moment? Or should I bite the bullet and get a different sized media?
Thanks for all the help!
Take care,
Caligari
Hi Liam, thanks for that information.
You should be able to download older versions from an application in the dashboard, and then use the balena cli to run standalone (balena leave <device>).
Once the data partition has been correctly expanded you should be able to update to a newer version and don’t hit this problem as I presume the failing script will not run for an already expanded partition. To perform a hostOS update for a standalone device you will need to copy the new image to the device and do a manual update withhostapp-update -f <filename>.
Regards,
Thank you so much for your help!
Just to briefly followup.
I grabbed a cloud build of 2.38.3+rev4, and used this to boot my NUC. (I had also tried the next version available, 2.41.1_rev1, but that didn’t come up, as with the later releases.)
Everything seemed to be functioning fine. I went into the dashboard and to the Actions for the Device, and asked for the BalenaOS update, which decided I would go all the way to 2.50.1+rev1. That process was relatively speedy, and it seems to have come up just fine.
So the path through those two versions should, I assume, work for others with similar issues with large (>=2TB) drives, until there is a longterm fix.
And I understand I could enable Local Mode if I wanted an install that isn’t tethered to the dashboard.
Hey Liam, glad to hear this is working for you, once again thanks for all your feedback and effort.
And I understand I could enable Local Mode if I wanted an install that isn’t tethered to the dashboard
Yes, Local Mode - while still being activated and deactivated through the dashboard - will make it so that all the communication with the device goes through your local network (which has some caveats, e.g. loosing the service logs through the web terminal). The idea is that it can be used while developing your code for a very fast feedback cycle. Have a look at Develop locally - Balena Documentation for more information, or feel free to come back with any questions you may have about it.
So, I’ve purchased a 500G SSD for my NUC, and have been trying to flash the device with my application, but I’m not having success. I’ve grabbed the most recent, 2.50.1+rev1 image from the dashboard.
The flashing process is not getting to the point where I remove the USB, but instead is sitting there in what the dashboard calls “Flashing BalenaOS on internal media”.
I now have two physical drives in the device - the SSD and the previous HDD. I am hoping to use the SSD as the main drive, and the HDD as persistent storage for some larger data. Is this a setup that is going to cause me problems? Or is there something going on that I do not understand?
I can open this as a separate topic, but I wanted to be sure none of the information here was directly related, somehow.
Hi @Caligari, this is a different issue (no worries on not opening a new thread). Currently balenaOS utilizes a specific order for picking the drive as seen here:
# On Intel genericx86-64, we flash on internal hard disk drive
# The first one found will be used
INTERNAL_DEVICE_KERNEL = "nvme0n1 sda sdb mmcblk0 mmcblk1"
We already have an open issue to look into how this could be made customizable: https://github.com/balena-os/meta-balena/issues/1835
I noted your interest in there and we’ll post an update as we make progress.
For now, you could utilize this tool as a workaround: https://github.com/balena-os/resin-image-flasher-unwrap
The tool extracts the disk image from the flasher so you can write it directly to the SSD. Once you make sure that your device appears on balenaCloud by booting from SSD, you could add HDD as persistent storage.
Let us know how it goes.
OK. I found that I could disable one drive in the BIOS, and then the process worked just fine, but if I had both enabled, it would hang before the “Flashing BalenaOS on internal media” step completed.
Once I disabled the HDD, I was able to quickly get an install on the SSD. Then I re-enabled the HDD in the BIOS, and repartitioned it as I needed. It is now recognized, which is all fine.
I now have a separate question over in the Cloud section, asking about how I am supposed to go about mounting (and unmounting) this other drive safely. But other than that, things are running very smoothly. I’m loving the workflow that Balena offers me!
Thanks!