Custom apps and devices cause 'Sever Load' post - Please help me troubleshoot!

[mabstrategy]

I've recently got a 'sever load' alert, please can someone advise as to what I should do? I have what I feel is a fairly 'moderate' setup, at least compared to what many others seem to have... I guess the only thing that is different is that I have 2 Hue hubs with circa 50 devices on each connected via CoCoHue. Do I need to buy another to try and off-set??

What I've done so far:

Unofficial Ring - I did have polling going for x3 camera snapshots in the dashboard, I know this can increase hub load, so I've now switched that off entirely, however, it has been running just fine for at least 2-3 months.

I've looked at the app log screen below, it seems 'timer-related' WebCore pistons are using the most resource, however, none seem to be too demanding. Obviously the Ring App's % of busy is high at 40%, but hopefully turning off the jpg polling will have fixed this...

Then this is my device driver page, and the CoCo Hue app dominates hub resources. I guess I want to know is this expected behaviour for 2 Hue Hubs that are there, or there abouts full?

They have 'enable push updates' activated, but also poll every 1min (default setting) as recommended by the developer (something to do with the Hue App API still not being great). Is it this 1min poll that is causing my issues? The reason I have this set is because I have some zigbee implants on each switch (all integrated into hue), however, it can take a while for the Hue hub to update if one of these gets switched off at the wall, hence the poll setting.

Nonetheless, all this considered, everything was seemingly working ok up until yesterday, only thing I have done is added a Fibaro ZW5 motion sensor and and Aqara Temp and Humidity sensor in the past 2 days, but they are not even registering on the log pages and are seemingly barely impacting the hub at all...

Lastly here's a pic of my current hub information, if anything stands out to you? 62 degrees seems hot, but I'm not convinced that is correct as the hub is only slightly warm to the touch - nothing feels 'hot'.

Many thanks in advance!

[jtmpush18]

Well, to start off with, may I suggest starting with the simple things:

1. Do you have any devices which do "power monitoring"? If so, disable those first.
   (Look at every device, and check to make sure that it doesn't have "power monitoring").
2. Turn on Zwave loggings: Settings-> Zwave Details-> Zwave Logs. See how many logs are going on, and what devices they are for. Might surprise you.
3. Do the same for Zigbee logs. Any big repeats? Lot's of activity?
4. Look at the general logs. Scroll backwards in time, and look for errors, repeats, etc. Anything unusual?

Please note: these are just the first "baby" steps... Hubitat does provide a lot more details, but please walk before you you run.

[brad5]

For a CoCoHue comparison I have three Hue hubs and run both the CoCoHue and native integration. Together they use about 25% of busy and about 1.5% of total CPU. One thing that's interesting is my average MS is only 8. I don't have nearly as many Hue devices though. I don't have push updates enabled.

Some folks have problems with Aqara devices and they are known to sometimes cause zigbee mesh issues. Not sure if that's related to your problem or not.

Are you using hub mesh, and if so how many devices are you syncing?

[mabstrategy]

Thanks for the response. No not using hub mesh....

[gopher.ny]

That's an interesting case... can you PM me your hub's id? I can take a look at the engineering logs on the hub, maybe there will be some clues there.

[mabstrategy]

Many thanks for the response. Re. power monitoring, I have a few Smartthings plugs, but all power reporting is switched off (I don't have any automations that use this function). I'll try your other suggestions asap and get back...

[mabstrategy]

Many thanks @gopher.ny I will do now...

[bobbyD]

@mabstrategy

CoCo Hue app dominates hub resources

CoCoHue is using a lot of resources, which is rather unusual. I would ask @bertabcd1234 for ideas on how to optimize the app better. I know there are some features that are still experimental, so that might be something contributing to your excessive load.

[mabstrategy]

Many thanks @bobbyD

@bertabcd1234 any thoughts on any potential tweaks I could make?

Thanks again for your help.

[gopher.ny]

I'm going to second the CoCoHue, if only because of the event stream disconnects every 30 seconds. There's a readTimeout option in the event stream interface.

[mabstrategy]

FYI, I've re-booted my hub and the alert has now gone the only change I made was to turn off the Ring Camera jpg polling for my x3 ring cameras. However, as previously mentioned, this had been running fine for months previously...

I've checked the Zigbee and Z-Wave logs and they both seem fairly quiet, with nothing obvious 'jumping out' at me...

CoCo Hue is still using the majority of my hubs resources according to the logs, though they are now slightly less than before:

As far as Apps are concerned. Webcore has now replaced Ring at the top spot with 25% usage following my Ring jpg polling switch-off:

[mabstrategy]

Thank you for this. I guess that this isn't something I can do anything about outside of not using the app?

[bobbyD]

@mabstrategy

re-booted my hub and the alert

Rebooting the hub resets the alert. There is only a matter of time before the hub reaches the alert threshold and you will see it again.

[bobbyD]

@mabstrategy

I guess that this isn't something I can do anything about outside of not using the app?

Event streaming is optional and experimental. There is an option to turn it off in the CoCoHue app.

[mabstrategy]

Ok, thanks...

[mabstrategy]

Ah yes, just saw that's the 'push' event side of things which updates statuses imediately - exactly what I use the app for, so that's a real bummer as it's not something I want to turn off...

[brad5]

What's the use case that requires immediate updates? Maybe there's another way to accomplish the same thing?

[mabstrategy]

Ah you've put me on the spot now, I remember when this feature came out it solved a few issues I was having at the time. I think it was primarily around status updates for motion lighting on the push notifications being instant vs. the slower polling method. But will have to dig back through my posts to properly remember my use case. I haven't had to think about it since because everything just worked!

[brad5]

Maybe see if turning off the push notifications solves your immediate problem and then worry about it if you need to go down that path

[mabstrategy]

Yeah, exactly what I've done. Will see what I notice (if anything) isn't working as it was before...

EDIT

Just found my first use case. In one of my bathrooms I have a light and an extractor fan on 2 different gangs controlled by x2 samotech Zigbee switches integrated into hue. When I switch on the light, I have a hubitat automation that also switches on the extractor fan. Here's what happens now...

I turn on the light via the switch - hue takes about 5-10 seconds to realise it is on
With event push on - extractor fan switches on at the exact same time hue realises it is on
With event push off - the fan comes on circa 30seconds later I'm guessing as it is waiting for CocHue to poll and report the switch status (my polling setting is currently set to 30seconds)