Release 2.4.3.133 Added automatic hub reboot on CPU load Question

📢 Feedback
1

In the 2.4.3.133 release im curious about the automatic hub reboot if CPU load remains at a critical level for 15 minutes or more Platform Change.

Is this a single reboot or triggered everytime the condition is met?
Is this disableable?

My concern is, is it possible to get stuck in a ~15min reboot loop if a device/app is misbehaving due to the cpu load trigger? ie Can 1 or more spammy or dying devices cause the cpu on the hub to spike and stay spiked and keep a hub in a reboot loop?

2

I'll add a disable endpoint.

I'd put it as "extremely unlikely". There are other mechanisms to limit the load that a single app or device can create on the hub, although they're not ironclad.

One, no. A dozen, maybe.

11 Likes
3

What is the criteria for "Critical" ?

image
image711×151 16.3 KB

1 Like
4

Are there any configurable settings and if yes where are they ?

5

This didn't make it into the release notes...

1 Like
6

@gopher.ny please answer @kampto question about the critical CPU Load value.

7

Can someone explain to me on using this endpoint. I think I replace the “hub” with my hubs ip address but I’m stuck from there

8

I haven't used these particular endpoints, but you typically put your IP in front of the "/hub"... e.g.

your.ip.here/hub/advanced/disableHubProcessMonitor

I always do these sort of "process" links in a new tab; that way, the results aren't overriding the display of some thing else and I can just close that tab when done.

1 Like
9

hmmm what else is there that this endpoint controls?

10

Thanks but that did not work. Hubitat gave me a Error 500

11

did you use a full URL in your web browser like http://192.168.1.100/hub/advanced/disableHubProcessMonitor (where 192.168.1.100 would be replaced with the local LAN IP of your particular HE hub.)

I just tried it on my hub and it worked without any error.

4 Likes
12

Just curious as to why you would want to disable this?

13

Warning: Most of this is philosophical

I would much rather be notified via my own rules and notifications (triggered from a Hub Info Device (V3)) about resource constraints (CPU, Memory, external PING check, etc.). Then I can check what's going on, and decide on a course of action - I have Wireguard VPN access to my LAN and a WiFi plugs on hubs, if I need to "hard boot" (To date, I never have had to do this while remote).

I just don't like the possibility of some repeated 15M automated reboot cycle, if a hub is in some runaway loop. I would much rather investigate myself and then decide what to do (hard reboot, trying to disable some device or App, etc.).

Fortunately, this has never happened to me (yet?), I've only once had Zwave Mesh repeatedly crash due to a bad Jasco device, regardless, I generally don't like "automagical" fixes, potentially running for days on end, that can mask and make problems worse (corrupted DB from hundreds of reboots?). -

I much prefer alerts, and then human troubleshooting and action. - And this goes for most "things" outside of the HE realm. IMHO, overly smart tools, without options to manually intercede will be the downfall of our species (not HE in this particular case - more like thinking of Boeing and MCAS)...

As always, YMMV, but I have it disabled on all my hubs.

1 Like
14

FWIW, auto reboot is disabled for one hour after hub boots up...provides plenty of time to use the endpoint to turn off the reboot if you're getting into a (exceedingly rare) reboot loop from this option. :man_shrugging:

I get it...

2 Likes
15

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.