Been running CORE for a long while, but it’s time to start looking at finally moving to SCALE. I believe that the recommended migration procedure is to update to the latest version of CORE prior to migrating. However, I’ve noticed a problem that I kind of slept on for a while now that’s preventing me from upgrading.
It seems that after running the updater, after reboots I end up staying in the same old version, despite the new environment being selected as the reboot environment in the web GUI. I’ve been stuck on 13.0-U4 for over a year as a result.
In fact, when I go into the boot options menu on startup, the new boot environment doesn’t even show up when cycling through the available environments.
Anyone have any ideas where to look to get clues into troubleshooting this?
System Details
Old Dell R520 booting from mirrored USB drives (I know, I know…)
I didn’t notice anything obvious on the release notes. Enough space on the boot-pool? Tried doing the manual download of the update and applying instead of automatic upgrade? One option may be to make sure you have your saved system config download and then download the ISO for 13.0u5, do a fresh install and upload the system config back.
As you have not posted any information about your system or extra applications that you are using, it is not possible to offer specific advice. However, another way to upgrade would be to do a clean install of Scale and then restore a saved configuration from your Core system. If you review the documentation about migrating from Core to Scale, you will find advice about this approach.
Your data disks can be unmounted from the Core system and then re-imported to the Scale system. Data will not be at risk, although as always you should have a reliable backup before doing anything.
50G free on the 64GB mirrored USB boot drives. Pretty sure that I have enough free space there.
I had considered the possibility of a fresh install, and it remains a possibility. The main reason for the post is seeing if there’s anything simpler that I might have overlooked as to why the environment in the boot menu doesn’t line up with that in the web GUI.
What does the server currently do. If you are just serving files and have a smb share, those can be recreated under Scale easily. If you have Jails, plugins and a complex setup, we might make different suggestions.
Have you gone through the Core to Scale migration article or guide and done the checks? Geli encryption. No USB sticks for boot (recommendation) If you are stuck with USB, suggest a nvme or sata ssd in a USB adapter. USB sticks were wearing out quick. SMB sharing is done a bit differently and you don’t share a ROOT dataset, You create a Child dataset and share that.
The more details on your system, the more likely we might catch something.
Using an old NVIDIA graphics card? Latest drivers dropped support. Things like this.
I probably muddied the waters by mentioning the eventual upgrade to Scale and the best procedure for doing so. I’m not looking at actually making that upgrade yet. I’m considering that migration itself out of scope for this topic. When I do start the migration process, if I run into any trickiness I’ll be sure to post those details in a separate topic.
For the purposes of this topic, I’m just concerned with the upgrade of Core. Specifically if there’s a file or other kind of setting that might be misconfigured for the boot environment discrepancy. Though if there’s no insight or clues that I come across here or elsewhere, I’ll just follow your suggestion to do a clean install and import the config.
That all said, It’s an old Dell R520 with dual E5-2450v1 and 32GB of ram (but I want more) with an HBA in IT mode, 8 bays filled with drives ranging from 8TB to 20TB mirrored as one single main pool, PCI NVMe adapter to fit a 32GB Optane stick to use as a SLOG. Don’t have a lot of room inside to put in proper SSDs for boot, though if I find a cheap solution for the USB adapter, some kind of PCI adapter, or something else I’ll look into it. I know USB boot isn’t ideal, I’ve had to replace a couple drives over the years, but otherwise it hasn’t been too bad so far. No graphics card is installed, or needed.
No Jails anymore, though for a while I used the NextCloud plugin until that made me lose too much sanity. Network is gigabit ethernet using the built-in NICs. Looking into finding a decent SFP+ (or 10G-BaseT) NIC to upgrade to 10G ethernet eventually
It’s used exclusively to share files over SMB in the home, mostly serving media for my plex server which is running on an M4 mac mini, but also for general filesharing to macs and PCs, and as a TimeMachine destination. All of which are their own datasets, not at the root of the pool. No encryption is in place at the moment. Replication of the pool is done by SSH+Netcat over a direct ethernet line to a pool sitting in a nearby proxmox host.
I am aware that SMB auxiliary parameters (in my case used to try to improve compatibility with MacOS’s sorry SMB implementation) are a sticking point for the migration to Scale, which is the main reason why I am not ready to actually begin that process, yet.
This is an odd one. Normally a bad USB boot environment would just mean no boot, not “refuses to upgrade”
A config backup and fresh install of CORE 13.0-U6.8 should work - although I wouldn’t suggest USB boot for SCALE/CE.
We do have an option to enable additional protocol extensions for macOS in CE - you’ll want to match any existing settings for case sensitivity and name-mangling/character compatibility in your shares as well.
