So here’s the deal… I upgraded my HE to the latest firmware earlier this week and since it has built in performance counters, I started getting alerts of severe CPU loads and a significant slow down of the hub. The most consuming entities are my 15 snow devices.
Based on advice received on the HE forums, I deleted the Hubitat dashboards that I had (and never used as I switched to sharptools) and the load reduced significantly. However after 24 hours I still see a significant rise attributed to the Sonos devices. Since this is a built in integration I can’t tinker with it either.
So to the main question which relates to my HE dashboards experience - Are unrendered Sharptools dashboards still work in the background? All my Sonos devices are confined to a single dashboard with a panel tile on my main dashboard to access it. About 98% of the time that dashboard is not rendered on any of my devices (mostly use Sonos via Alexa voice commands) but I would still like to keep it. However it is not sustainable on the hub in cur4ent form.
Is there a way to have dashboards time out or go into a dormant state if they are not accessed via an open session or called upon in x minutes?
By the way my power shades stopped working this week. I am now, in real time, making the connection that my shades panel may have been polling them for status continuously and caused their batteries to drain? This has never been the case and perhaps also attributed to the HE firmware change and how dashboards are run.
Same question still persists - sharptools dashboards that are not rendered may operate the same way (always active) hence fixing behavior on the HE side by removing the dashboards may only “shift” the issue to Sharptools doing the same?
The dashboards themselves aren’t actively doing anything in the background, but if there are devices added to those dashboards then event subscriptions get setup.
Meaning the dashboards don’t ‘poll’ for state or actively interact with the devices or the hub in the background. The SharpTools app just listens to the events that are already being reported by the devices (even if the dashboard is not actively being viewed).
Thanks @josh then:
Can this behavior be changed or controlled. e.g. subscribe to events upon rendering of a tile and unsubscribe after session timeout?
In lieu of the above, what controls the frequency of the events - the host or the subscriber (when subscribing)? Otherwise asked, is there a way to reduce / optimize these subscriptions to be less frequent for the entire integration / dashboard / tile?
And yes, ideally everything would be real time if HE’s CPU could handle it but like Orwell said: “all tiles are created equal but some tiles are more important than others” so perhaps a more granular approach is necessitated.
I can’t imagine I am the only one who experiences a spike in CPU and it might be a matter of time before other reports start to correlate to this pattern?
Keep in mind that the Hubitat metrics are really primitive in nature - just showing the time spent processing, but not the actual impact on the hub’s performance. So in the case of SharpTools, it’s primarily just the time spent sending events across the wire which are not processor intensive items.
The behavior around event subscriptions is not controllable. The device that generates the events is responsible for the frequency of events.
You might take a look at what version you were on before and what version you are on now and the changes between those. Nothing has changed with the SharpTools ‘app’ on Hubitat in the recent hub versions. Which makes me wonder what else changed that could be impacting your experience…
The only thing that changed on this end that started this behavior is the change from 2.2.4.xxx to 220.127.116.11 firmware version on HE. I have been busy on other fronts and thus made no changes to my home automation setup apart from this upgrade.
Given that the Sonos integration is built-in to HE, I can’t trace these issues. I have a (mistakingly) closed thread on the Hubitat community side that I will revive and hopefully get some response from their developers on what has changed in the code of this release to make the devices significantly more chatty.