Some logic has been built in emhttp to keep track of the disk being active/stand-by and the updating of the disk temperature. Both monitor and webGUI rely on this logic for proper handling and display of the temperatures.
If/when something goes out of sync, it is usually restored when the poll_attribute timer kicks-in.
That sounds great, what *should* happen, but isn't. My poll_attribute has been set to either 180 or 60, and I've waited many hours, with no change. Happened again yesterday, 4 drives only. Happened to at least 2 or 3 of the drives when I was checking the Help text for SMART info, after clicking on the drive, with the Help enabled. It briefly seemed to indicate SMART info would not be available until drive was spun up, then in a few seconds the SMART info all appeared (looked really great by the way, with a new Download button!). Returned to the Main screen, and drive still showed spun down (grey ball), but a temp was showing! The temp continued to show and be updated for quite a few hours. (I know by now you must be getting tired of my posts!)
Yeah for every one line answer I get ten lines of questions back, that is a heavy burden
But ... you did make an interesting observation !
When users look at the SMART information in the GUI, the GUI will issue a smartctl command in the background to obtain the necessary information. This is done without knowledge to emhttp itself and it can bring emhttp out of sync if a disk got spun-up due to an executed SMART command (do you happen to have Seagate disks?)
It seems that the poll attribute timer updates the temperature reading for an active disk (it uses hdparm and smartctl for that), but emhttp itself doesn't get updated about the state of the disk if it has changed. I believe Tom made a remark about that, and advises to let control of disk spin-up/spin-down solely be done by emhttp.
A solution might be to extend the poll attribute functionality with a disk state update feature if changed state is detected, this approach doesn't give an immediate sync but at least at some point things get straightened. This however is up to LT to decide/implement. Perhaps a 'defect report' can help?