DASHBOARD Disk Activity widget

Okay, I’ve noticed something that I think can be improved or implemented as an option with the disk activity widget.

  1. Keep in mind I suspect a majority of users use some sort of RAID in their system.
  2. Also think about what data is actually useful to know when testing things internally.


While the top graphic portion is impressive, it is also somewhat less than optimal because it is showing individual disk activity; BUT, there is no way to fix a color to any disk. The disks and colors associated with them jump all over the place.

When transferring to/from a RAID setup, that lack of fixed color association is frustrating. In addition, I would rather get better and easier to understand POOL and/or RAID information. Possibly we could use check boxes for source information selection? We could then select our RAID disks individually to be combined as one.

Lastly, the throughput display on the bottom and the various selections in the scroll box are kind of small and not 100% understood by me. I would rather see something akin to a pool or raid selection option possibly combined with disk check box selection mentioned above.

Just my two cents and half a cup of tea…

(Is it possible for someone to program their own widget???)

Cheers everyone!



@Tex1954 Thanks for the feedback.
Yes our Dashboard has had very little attention since it was completely re-written by @Flyer many years ago. But it has, since then, received an additional widget, via @sfranzen (see later for details):

Yes, of course. The directory in the source to look at for the view part is here:

Your suggestions re raid drive flagging within the disk section could introduce some confusion, but as always the proof is in the pudding/pie. So do dive in and see what you can come up with. The disk’s widget is in that same directory here:


These views gain their input by connecting to the ‘feeds’ from:

And specific to the disk widget we have class DisksWidgetNamespace(RockstorIO):
and internal to that the ‘engine’ is here:

As you can see we simply parse /proc/diskstats

There is some awarness of larger elements but not much. Hence it seeming disconnected from other stuff; such as pools and raid. The concern is that adding too much complexity, and pulling from too many models may make the display unworkable slow. But again a proof of concept may be all we need to see here.

As to the more recent addition, post @Flyer’s major contribution in this area we have @sfranzen’s contribution/addition:
See forum thread/post:

and commit:

via pr:

As always pull requests welcome and we have the following recently updated doc section for how to get a development environment setup ready:

Community contributions -> developer section.

But our dashboard is likely to remain fairly basic given there are projects solely focused on such things. Have you for instance tried out our netdata Rock-on. You might like that.

Hope that helps.


Thanks for the input! I see visions of yet another system build in my future.


1 Like