C-8 Loosing connection

🛎️ Get Help
1

Hello, hopefully I am using the correct process to ask for help. I have two new C8 hubs, one in each of my homes. The one where I am not located at present seems to lose connection at least once a day. I am running the 2.3.6.146 platform.

My setup is pretty simple, I have an ASUS RT-AX58U router connected to an Xfinity cable modem. The C8 is connected via Ethernet cable to one of the router ports (no switch). I use the provided power supply.

I have basic and rule machine rules to turn on a few lights at sunset (for which the basic rule is used), then turn them off at different times using a RM rule later (e.g. after 10 PM).

The last few evenings I've checked out my IP cameras and noticed that the lights were not switched on. I also noticed I couldn't reach the dashboard I set up to control those lights manually. From there, I logged into the router remotely and determined that the Hubitat did not appear in the list of connected devices. Rebooting the router does not seem to have any effect and the C8 still does not appear as a connected device. I assume since the lights are not turned on after sunset that the local processing is not happening. I suppose the C8 still needs to be connected to the router in order to execute rules?

Fortunately, I have Kasa Wifi plug such that I can turn the C8 off, and then on again. The Kasa plug to which the C8 is connected is shown as a connected device, so I am confident the Wifi plug is not causing the problem. Turning the plug off and then on again seems to restore the connection to the router and I can then access the dashboard.

I have the C8 set up with a static IP address both on the router and on the hub. I double checked to ensure that the IP address is the same on both. In reading some other posts, I have jumbo frames disabled on the router, and have the ethernet speed on the C8 set to 100 fixed.

I have installed Hub Connect in order to enable the Amazon Alexa integration to access devices on both hubs without having to specify which hub. The C8 at my present location is the server hub, the remote C8 (with the connection issues) is the client. Hub Connect seems pretty solid, from what I can tell thus far.

In addition to Hub Connect and Alexa integration, I also have the Notification, Hubitat Package Manager, Kasa Integration, Sensibo Integration, and Poller apps installed (some of my devices are pre-Z-wave Plus).

For now, I have no way to physically access the C8 in question, since I won't be there for a few weeks. Hence, I have no idea what color the status lights are and what not. Also, I will not be able to change the ethernet cable or the port to which it is connected for now.

If anyone has any further troubleshooting ideas, please let me know and I'l try them out, either from my current location or when I get there in person in a few weeks.

Thanks in advance!

2

I split your post to a new topic.

Check your logs for an errors or warnings. I am not sure how intense Hub Connect is when working across the internet vs local, have never used it myself. Anything using IP / Internet connection though, if they get disconnected and repeatedly try to reconnect they can eventually crash the hub. You should see errors in the past logs if this is happening.

This is a bad practice, could be causing an issue. If you have the IP set on the router side, switch the hub back to DHCP.

Not for local Z-wave and Zigbee devices, they can work totally offline.

3

Thanks much for the assist and quick reply! I switched the hub back to DHCP, so I’ll see if that helps.

I’ll also check the logs once I figure out how to do that…

I do have my cameras setup with static ip addresses on the router and the cameras and haven’t had a problem in years. Just curious as to if this is a bad practice in general or in the context of HE hubs.

Given that the hub should continue to execute rules when not connected to the router, is it safe to assume it crashed? That would explain why the rules hadn’t executed to turn the lights on after sunset.

Thanks again! I’m new to Hubitat and finally punted on my Vera Plus controller a month or so ago. The learning curve has not been very steep, I’m quite pleased thus far.

The community is terrific, I’ve learned a lot by reading through various posts.

4

In general, never a good idea.

Here is the docs on the logs page: Logs | Hubitat Documentation

5

OK, found the menu. I’ll read the documentationNext question, should I enable logs for all devices and apps? I turned most logging off assuming it is a resource hog.

I did find this error in the logs, however. It seems to repeat every 10 minutes or so.

app:52023-12-11 08:44:02.136 PMerrorError making Call to Alexa message gateway: {"header":{"namespace":"System","name":"Exception","messageId":"30f10ade-0210-46a3-b3c3-2b6b3dda278f"},"payload":{"code":"INVALID_ACCESS_TOKEN_EXCEPTION","description":"Access token is not valid."}}

app:52023-12-11 08:44:01.829 PMerrorError making Call to refresh Token: [error:invalid_grant, error_description:The request has an invalid grant parameter : refresh_token. User may have revoked or didn't grant the permission.]

6

Sorry for all the replies - I think one issue was that I had devices enabled for the Alexa integration on both hubs. Turns out this is unnecessary when using Hub Connect. I turned them all off in the Alexa Integration on the remote hub. The dashboards and Alexa integration still work, since I added all the devices from the main C8. As I mentioned, that was the rationale for implementing Hub Connect.

Maybe that was messing things up. Why you cannot access devices via Alexa from multiple hubs is beyond me - you could that from Vera hubs many years ago…

7

I think the way the Hubitat skill is setup in Alexa you can only connect it to one hub at a time.

Sounds like the Alexa app on the remote hub is still installed and trying to connect. You should disable or delete the app. Its constant reconnection attempts may be gumming up the hubs memory.

Speaking of which, you may also want to try to "clean" the database by downloading a local backup then restoring it right away. This has been shown to fix rapid memory decline issue with 2.3.6 firmware, which would cause the hub to eventually crash. It is explained here: ‼ READ FIRST - Before Posting in Get Help

8

Turning the hub off and on like that can corrupt the database. I would start by downloading the latest backup using the create and download button under backups to your pc. Then do a soft reset and when prompted. This will ensure a clean and healthy database.

Now when the hub is unreachable, can you reach it from a pc by going to yourhubsipaddress:8081 ?

If so then there is likely something in your mesh that is killing the hub.

Do you have Jumbo Frames turned on anywhere?

This is not good practice. Make it either or. If Static on the hub vs a dhcp reservation make sure the ip is outside the DHCP pool.

What is your frequency on polling??/

Can you post your z-wave details page in its entirety? (Use windows snip)

1 Like
9

You only need to download a local backup and then restore it right away. This does the same thing as a soft reset and restore but with less steps and no need to go into the diagnostic tool.

2 Likes
10

Hi Rick,
Thanks for the assist! I’m not in the same location as the hub. However, I do have remote admin and can access it (after doing another off/on via a Kasa plug this afternoon). From what I can tell, you have to on the LAN in order to access the diagnostic tool. It fails, showing the local hub ip address and port 8081. I can access other menus (including determining the MAC address). I don’t have a PC at the remote home, so there’s no way to access the hub on the LAN through a VPN or via TeamViewer.

I’ve never turned on jumbo frames for any devices, unless they use jumbo frames by default and I’m unaware of it.

I changed the setup to be DHCP on the hub, and static on the router.

Polling is performed via the poller app. I couldn’t find a way to specify the polling frequency. I’ve read that some users utilize WebCore for polling, but I am not at all familiar with it. If there is a way to set the frequency via the poller app, let me know. Otherwise, if there is a way to set polling via the device page, that would be great. I'm coming from Vera hubs, where you could easily specify the polling interval per device. On a side note, you could also arm and disarm motion and door sensors, I have yet to figure out how to do that on Hubitat.

Here is the Zwave info:

image
image2914Ă—1301 228 KB

image
image2877Ă—988 169 KB

image
image2897Ă—1399 255 KB

image
image2880Ă—1390 255 KB

image
image2881Ă—1382 245 KB

image
image2859Ă—1386 261 KB

image
image2890Ă—468 56.1 KB

11

Done! I'll see what happens tomorrow... Thank you!

12

You don't arm devices, you would use something like Hubitat Safety Monitor. The devices get selected, and you can arm/disarm the home via a dashboard tile.
Hubitat® Safety Monitor | Hubitat Documentation

You cannot. I am not sure about Webcore, but in Rule Machine you can specify intervals.

That said, what devices (brand/model) are you having to poll?

13

I'm only polling one in-wall switch that is likely an old GE 12722. I can't be certain without being at the remote location and taking the unit out of the wall (I won't be there again for a few weeks). I noticed that when I manually turned the light on or off, the status would not update in the app. A rule I had created would also not function properly. FYI, I have quite a few older devices that aren't specifically shown in the options when you use the find by Device Type or Brand menus, so I set those up as generic z-wave devices using the "Add Device Manually" option.

I figured out that the C8 seems to know which devices are candidates for polling based on which ones show up when you access the menu to poll switches or dimmers. After I selected that one switch to poll, the status is correct when pushing the toggle manually, and the rule works properly. That is the only device I am polling at present.

I am aware that Rule Machine can also do polling, although I am not certain of the context. My assumption is that you can build polling / refresh / capture into the rule as an action. Not clear how one would use RM to set up a generic polling interval similar to how I assume the poller app works. What would the trigger event be? How would you figure out what the polling interval should be? And, how would you set that up? A sample rule would be very helpful. I am new to Hubitat, having just replaced two Vera Plus hubs in the last couple of months. Vera handled polling automatically, and you could specify the polling interval per device, or tell it not to poll devices when not necessary.

14

Polling one switch shouldn't be a problem. I would not bother to move that to Rule Machine or any other polling method. I think that is a red herring.

There is no issue doing that. I normally don't even use anything but the generic "Add Zigbee" or "Add Z-wave" options. The correct driver is or should be selected no matter which method you use, they use "fingerprints" rather than the model number to select a correct driver.

In fact, earlier in the history of the Hubitat, the guided by device brand/model menus didn't even exist. Those guided menus just assist you by giving instructions rather than having to look them up or trying random button presses.

1 Like
15

Thanks to all of you who have responded and helped out on this topic. As it turns out, after I did the soft reset, the hub crashed again. Hence, the reset didn't seem to help. At that point, I decided to turn off anything relying upon the hub that might mess things up (e.g. HVAC) at my remote home if the hub weren't operational.

I had installed the Sensibo integration and was using it via RM to control a mini-split in my garage based on a temperature range as well as a humidity range. Sensibo has a feature called "Climate React" that only handles humidity. I found there would be wild swings in temperature when using Climate React.

The rules I set up were working great - I aslo have a Watchman device so I can monitor temperature and humidity pretty accurately. The rules involved the temperature / humidity readings of an Everspring ST814 battery-powered temperature / humidity sensor. Ideally, it would be great to monitor the Watchman device, but the Hubitat driver posted on their website doesn't work, so the Everspring was the only device I have that is suitable for this use-case.

Either a too-high temperature reading or too-high humidity reading would trigger the mini-split to turn on (via the Sensibo integration) and vice versa. This resulted in very consistent temp and humidity levels in the garage, way better than with Climate React.

Long story short, I reverted back to Climate React on Sensibo and disabled the rules described above. The plan was to just leave the hub off if it crashed again until I could get there (since powering the hub off and on via a Wifi Kasa plug could corrupt the database, as someone noted).

Lo and behold, the C8 has not crashed since then. I have a feeling that those rules and monitoring the Everspring device were somehow over-burdening the hub. I have yet to dig into the past logs, which I'll do soon. For now, crisis averted, I can leave the hub on and execute the other rules (mostly for turning lights on and off) and try again when I get to the other location in a couple of weeks.

I will post additional information on how I had things set up as well as logs or whatever else I can determine in a couple of weeks.

This whole episode has been somewhat frustrating, but also useful in that I now have a better idea of how to navigate through the logs and some other features.

Hope you all have a great holiday season and Happy New Year!

1 Like