Cloud dashboard slow to load on 2.0.6

I did not perform any upgrades. However I did notice yesterday my cloud based dashboards started loading very slow, and sometimes not at all. They are still exhibiting the symptoms this morning. I have not rebooted if any staff would like to review my logs. My local dashboard is loading instantly and working flawlessly.

Cloud

Local

It's likely you have a device that has been authorized that has a large attribute in size. Some 3rd party drivers create very large attribute states and cause the cloud device endpoint to fail.

Remove or disable any 3rd party drivers and see if the problem goes away.

@patrick I have made no changes. This started yesterday. All on its own. It seems like their is an issue with the cloud server. It is responding very slowly. At times it is not responding at all. I would think if it were an issue with a driver the local dashboard would be slow to.

There is a limit to the size the cloud servers allow for an endpoint. Check your devices size, if it exceeds 128k it will error on the cloud side. What you are showing is indicative of this. Or that the cloud endpoint is unable to reach your hub, in which a reboot may solve that issue.

@patrick I rebooted and saw no improvement. I will review my device sizes and see if I can find one over the size you mentioned.

It's the devices endpoint, the red one at the bottom of your screenshot. That size can't exceed 128k for cloud dashboards.

Check the same dashboard via its local link and verify the size of that endpoint is under 128k

You mean like the built-in Sonos integration? :wink:

Yes, in Dashboard v2 a blacklist of attributes was created to exclude trackData from the Sonos driver. If we find other attributes that are causing issues we will update this list with those as well.

There has also been implemented a 1024 character limit to the size of attributes, anything going over that limit will be excluded from the device and not sent to dashboard to avoid this issue.

1 Like

So, if one wants to show track data from devices, what would be an acceptable workaround?

https://docs.hubitat.com/index.php?title=Driver_Capability_List#MusicPlayer

No need for a workaround just use the Track Description attribute.

1 Like

Duh. Gotcha. I see it now. Thanks @patrick!

7.12k

Looks like it also loaded?

Yeah it is, with an 8 second delay, however it is hit and miss. Something degraded since yesterday. The initial get has an 8 second delay also. The other dashboard I have setup for my wife wont load at all either. I am also getting a response intermittently that says "Internal server error".

I did not perform the hub upgrade trying to avoid this. I am thinking at this point I might as well go ahead and upgrade it.

@patrick load times have returned to normal with no intervention on my end. Do you guys know what the cause for the slow down on the hosted dashboards earlier ?

Nothing showed on our end. Keep in mind cloud dashboards need to have access to the hub and depends on both connections in between.

If either ISP or connections (hops) in between both routes are having issues, the remote relay will time out (approx 15 seconds).

Well... I work at the NOC for the ISP. We weren't having any issues today and non of my other web based servers/apps were experiencing any trouble. This was a strange one. I digress it works now. Now to decide whether or not to move forward with the upgrade. Thanks for your help with this one.

1 Like

The other possibility is that the hub was under enough of a load that it did not respond within the 15 seconds. Rare, but it does happen. Using any 3rd party apps or drivers?

Also, can't control the route between the ISP and AWS as well, which handles the relay back to your hub. We monitor the performance of the AWS side, but that doesn't always show the peering to and from if they are having latency or packet loss along the way.

Out of curiosity, if you work for the NOC at an ISP, why not just set up a VPN and use that for remote access? A VPN connection that you control is going to be far more reliable and faster.

There is an ISP node in "Washington" dropping 70% of traffic today. Not sure that's unusual :slight_smile: someplace is always 'broken,' right? :slight_smile: