Problem/Justification
The justification is that many users (as seen in the previous Bring Back SMART GUI) desire to have a built in GUI interface where they can establish a SMART testing routine. However we know the old GUI will not come back and I personally agree, it was too limiting and had barely any features. This is a recommendation for an enhanced GUI interface for SMART.
Impact
The impact to the user community will be favorable. They want it. There is no detriment to having this feature and it does not use resources, except smartd which is already in use.
User Story
My vision is fairly simple yet I hope covers all the bases:
Implement a GUI that supports:
Scheduling SMART Short and Extended tests using an easy to adjust interface. Selectable Short/Extended, Day of Week, Frequency. There are many ways this setup could look, but if you include a few people in the community to provide constructive feedback, that would greatly help, or at least include @joeschmuck (myself) to review the GUI setup
Set thresholds for key values that indicate prefailure.
Set thresholds for alarm setpoints such as temperature. Warning and Critical values.
Be able to look at the SMART attributes via the GUI.
Retain the Email notification if smartd detects a problem.
This kind of request could get crazy however I feel including just some reasonable functionality would be good. I would not ask you to to determine if a drive is bad, failing, good. That needs a little more finesse as sometimes identifying the correct item can be tricky.
I assume others will comment here and offer their two cents, and they should. I would also suggest that my first entry here not be the only factor in the GUI design and implementation.
The user community wants this so if TrueNAS incorporates it, we might as well try to get the community involved and make sure we do not leave out key items.
With that said… To the community who will make comments. Please realize that we will never get it all. The request must be reasonable and reasonably maintainable for TrueNAS (the company).
If this proposal does not sound reasonable, I recommend you submit your own and let the people vote. There are a lot of good ideas out there, let’s hear them.
I suggest that the new feature record changes in all SMART metrics and display these changes in chart form, just like “Hard Disk Sentinel” does. Of course, we can’t expect TrueNAS to accurately analyze all metrics, but displaying charts of all metric changes is feasible; at least users can see the trends of all metrics.
A good response to those of us who mourn the loss of the SMART UI would be a rich reporting feature to show just how frickin’ kick ass TrueNAS is being at monitoring drive health and protecting our data. That would be visual proof of the rationale to remove the SMART UI and make drive health monitoring automatic.
A picture is worth bla bla bla
So: a UI feature that implicitly contrasts how shite SMART was at reporting true drive health to how awesome TrueNAS’s new approach is. Seriously - that would be a fantastic addition with the bonus of silencing many of the critics.
ZFS’s view on data integrity MUST be better than SMART - just provide a view of this from the UI.
I’ve also realized that voting is completely pointless.
Slapping a lock on the SMART feature request discussion just to choke the criticism might feel convenient now, but that kind of badwill has a way of coming back to bite you in the ass later.
TrueNAS realized they’d made a boo-boo, but instead of taking the overwhelming community criticism seriously, they treated it as nothing more than an annoyance.
The enshitification train keeps rolling on. toot toot
Closing a feature request thread acts to give back the (limited) votes to the people who voted for it. But yeah, iX have a habit of closing threads that get critical of them.
Since we’ve now realized (for the nth time) that they just don’t care, this thread isn’t blowing up like the last one did. In due course they’ll close it for “not enough interest.”
Two ways to look at it: those like us that like being able to schedule SMART tests on our own accord (now via cron jobs, I guess) and SMART tests getting triggered automagically in background (25.10 and beyond).
The messaging around this feature removal and alleged plan to upgrade SMART reporting in the future can only be described as pretty dismal. The latest tech talk takes pains to explain that SMART monitoring isn’t going away, that they plan on doing a better job of automatically flagging issues, and so on. All of which could have been communicated much better - if that was the plan from the outset.
I’m also not convinced that removing a scheduling feature before all these background upgrades were implemented passes the ??? test. Change for the sake of change is not a development strategy. After holding back for several days of radio silence, I expected a more coherent comms strategy from ixsystems than poo-pooing the validity of SMART test results, etc. none of which addressed the main issue, ie removal of the GUI SMART scheduling screen in 25.10.
The more recent news releases seem to be more coherent - ie “we are developing an automatic implementation system so sysadmins no longer have to set it up themselves”. They have also pledged to build a smarter back end to interpret test results, etc. Which, like “enterprise-ready” VMs will be a great feature if it actually ships.
But the jumble of justifications, rabbit holes, etc in the meantime do little other than erode customer goodwill.
See the latest video starting around the four minute mark. Kris is suggesting that all that SMART stuff will now be automatically triggered / executed and hence sysadmins will no longer have to do it themselves.
This is, IMHO, a very different comms strategy than what we heard initially. If only iXsystems had emphasized that SMART GUI scheduling was becoming obsolete because the appliance would do so automatically, etc.?
But why then the note that they’d automatically convert extant smart schedules into cron jobs? That makes zero sense if user-set SMART scans / tests are no longer needed.
Going down unrelated anecdotal SMART rabbit holes like “I’ve had bad disks live and allegedly-good disks die” is also unhelpful.
One cannot help but get the impression that management is batting cleanup without wanting to admit that they could have done a lot better.
But quite in line with the messaging around the removals of k3s and Incus.
I think that sake for sake of change (“move forward and break things”) actually passes for a great plan in the Linux world, not to mention a fair share of the Silicon Valley.
User defined (or at least “user vetted”) schedules are still needed. How is TrueNAS CE going to automagically guess that tests cannot be scheduled on *** because the server is shut down? Or how is it going to guess WHEN to schedule long tests of 20+ TB drives (these which take over one full day to complete…) while avoiding periods when users demand maximal performance from the server?
