Hmm, the point about getting and analyzing the SMART output was always shell command line, makes the difference between Enterprise and Community Edition clearer.
In the Enterprise, they would have limited HBAs and storage. Any new storage device, (larger HDDs, new NVMes, etc…), would of course be vetted by iX. Thus, no problem for various monitoring tools.
However, for the CE, Community Edition, we have seen such a wide and odd variety of hardware, both actual storage devices and connection methods, that having a procedure like SMART short & long testing would be un-testable. It literally is way beyond rational to expect USB, Thunderbolt, VM passed through storage, hardware RAID, fake BIOS RAID, and such to work flawlessly with SMART. Let alone SMART short & long testing.
So I am actually leaning towards shell script launching and cronjob for SMART short & long testing.
So, how about this?
Script that performs SMART short test, with data file on which devices are tested
Same script that performs SMART short test, but with long testing option, with data file that specifies long testing
Ability to schedule them as Cronjobs from GUI
A GUI launched script that checks all the devices in the data file, for self test status, and displays it in the GUI
Well documented procedure to both enable the 2 separate SMART tests, short and long, and how to edit the data file.
To be clear, this is both new work, and not bringing back the SMART test scheduling to the GUI.
Part of the issue I can see is that the SMART test scheduling has to account for device renaming. Thus, would have to specify names by UUIDs / GUIDs.
I agree that the strong control over hardware that iXsystems has obviates the need for SMART scheduling to be removed for paying enterprise customers. That group won’t ever have to worry about weird HBAs, RAID controllers, and whatever else is causing edge-case issues because they’ll never experience them. Same thing with PCIe tunneling under Thunderbolt, NVME via smoke signals, or SATA over rotary phone connections.
For the community edition, it would be fine to specify: “here is what we support, and if you’re outside that, tough nuggies”. I don’t think this needs to be more complicated than that. For non-edge cases, scheduling works just fine. When was the last time we heard from someone on the forum with a non-edge-case situation where the SMART scheduler went haywire?
In other words, rather than define supported and non-supported configurations or take the opportunity to support the smartmon / smartd / smartcli team by sponsoring them with $$$ or helping them out with code, the company is trying to shift responsibility for something as core as SMART configuration / execution to a 3rd party (who is looking for a new project maintainer and who hasn’t updated their code in 1.5 years).
That is hard position for me to reconcile with what truenas is supposed to be.
I cannot stress enough how happy and pleased I am with TrueNAS Core being a rock solid and stable NAS. It handles memory gracefully. The ARC behaves as expected, giving me very snappy performance with my NFS and SMB usage, even when browsing massively large folders with thousands of files.
That I get to run “jails” is a big bonus.
Even though I would love to see additional NAS features incorporated into its GUI and middleware, I would be fine to keep what I have, with the only “updates” being the underlying version of FreeBSD.
A “TrueNAS Steady” train, if you will…
I’m not comfortable seeing the pattern of issues regarding OOM, performance, and spontaneous reboots with SCALE/CE. If I don’t use the Apps subsystem, I don’t see the added value of “upgrading” from Core, other than, “At least you’re no longer on an EOL product.”
I just don’t have the energy to recreate the convenience of Core with a vanilla FreeBSD base. I know it can be done, but really, I just want to use a NAS appliance.
I encountered numerous bugs while using TrueNAS SCALE, including over 30 unexpected reboots, a kernel error after a system update that prevented me from reverting to an older version, requiring a system reinstall once, and dozens of inexplicable bugs in the WebUI, some of which still persist. These include requiring me to log in again after just a few minutes in the background, regardless of the timeout setting; requiring me to log in again after only 10 seconds in the background on my mobile phone’s WebUI; cloud sync speeds frequently displaying as 0 when actually downloading; and frequent abnormal disconnections from the Web Shell. I have never encountered such a wide range of problems that are so difficult to troubleshoot that I’ve even given up on submitting bug reports.
I’ve always wanted to give TrueNAS a nickname: TrueBug.
The feeling I’m getting is that there isn’t a reason to update from ElectricEel-24.10.2.2 Anyway, +1 from me. Never expected that I’d somewhat join the Core users & have a preferred build that I stick to religiously.
This feature request is not about providing SMART analysis in any possible configuration: It is about bringing back the ability to schedule SMART tests from GUI, as TrueNAS provided for years…
How hard would that be?
Further, it would be nice (but already outside of scope for this request), just for the sake of avoiding a trip to the shell, that the GUI could display the latest detailed SMART report (-x and/or -a) in a window which allows to easily copy the text for pasting elsewehre—e.g. in this forum, seeking human help interpreting the output. How hard would that minor nicety be?
I just set up fresh hardware and drives with 25.10. This was my out-of-the box experience: I see Scrutiny handles drive monitoring. I try to install the app, but I receive some inscrutable error about the catalog. Furious Google searching lets me know I need to ‘refresh’ the catalog of something I just installed moments ago.
The app starts up. There are no drives. Just a blank page with no settings or option to add any drives. I did more furious Google searching and just gave up. I’m guessing some mystery task eventually runs because the next morning the drives show up without explanation.
I don’t have a prexisting schedule to port over. I’m just left hanging. I did start the first tests manually and will lookup the cron syntax that I haven’t dealt with in years, but overall this is a really bad user experience.
But this experience is, apparently, what paying and community edition sysadmins were allegedly asking for? As opposed to making SMART scheduling better, SMART interpretation easier, and resolution simpler?
I suppose if the goal is to increase the incidence rate for paid support contract contacts, this may be the way to go.
If SMART is judged to be deficient then invest effort into improving it instead of ditching it from the UI. Don’t ask us to rely on a moribund 3rd party app instead.
Plenty of effort has been directed to TN features that no one asked for so why not listen to the community of users instead and direct effort to what they want?
Oh, and while you’re doing that, you can restore the Network link to its previous location instead of stuffing it into the System submenu.
@HoneyBadger, in under a week, this has more votes than any other feature request ever on this forum, by a margin of nearly 50%. iX’ silence is deafening.
