Happens to me too, and now even more then a few months ago
1 Like
As found out by another ha user from nabu casa support apparently it’s a known issue as they’re upgrading network infrastructure, see the github comment.
2 Likes
I am having this same issue… I finally after a year of planning and playing to get to know HA, I am migrating off SmartThings due to the CONSTANT outages and un reliable setup, to HA with a USB GoControl Zwave/Zigbee dongle running on an ESXi host with the VM having 12GB RAM/8vCPU/60GB HDD. Got all 35 switches moved over and the numerous ZigBee devices and other Wi-Fi RGB controllers and such integrated. When I use the dashboard, everything is stupid quick. But when I use Alexa/Google, they either time out, or waits 10 seconds or so to perform the action… 10 seconds is LONG when you want the light on NOW… I am super disappointed by this and having regrets…
I even shut down the other 6 servers running on this ESXi host and even migrated to another ESXi host and same issues… I dedicated one of the 8 NIC’s to just this VM and put a path in my network to prioritize all traffic and even moved a vlan to make it go direct to internet and thats it… still same issues.
For the time being, now setup local google instead of nabucasa, now it’s instant
I mean , I setup this one now :
Thanks @pergola.fabio I am aware of that one, I decided to avoid the challenges of managing all that myself by using Nabu Casa but the family’s acceptance of it ‘not working all the time’ is wearing thin so I am reconsidering my approach.
Do I need to disable the nabu casa cloud google 1st? Will I have to redo all routines in google app? Could you post some rough steps you took to make it work locally?
?? All Steps are on that link, 2 posts above …
I used the nabucasa link for the auth, so no port forwarding is needed…
Thank you for this. Not only does it validate what I thought the issue was, but hopefully it will be fixed soon.
I also might look into setting up my own Alexa integration, but I’ve been trying to avoid this.
This has been getting worse and worse and it has been long enough to fix. Alexa now tells me dozens of times a day “…not responding”. It really is to the point of intolerable and I’ve moved getting off of Nabu Casa high on my list. Even HA itself is having Cloud timeout issues.
1 Like
Here’s the response I got when I contacted Nabu Casa support a couple days ago. (Completely separate from seeing this thread)
"
Yes, this is a known issue and it is being worked on.
We are re-writing the relayer which handles the communication between your instance and Amazon/Google and this is expected to be a large improvement over the current implementation. There should be much less latency and CPU usage on the backend and should be up between now February.
After that is tested and deployed, we will then be adding region specific relayer instances which should further improve the situation.
We have simply gotten to a point where we have many more users than when the service began and the same old resources and we need to make changes.
Thanks for your patience!
"
2 Likes
Ok so the new relayer has been deployed! This happened within the last hour as I write this.
This deployment didn’t take long at all and you should not have to restart HA or sign out/in to cloud on your server, but you can if you want to.
Most likely your HA server has connected to it already.
Try out your Google and/or Alexa voice assistants, apps, and see how it goes. Working well here.
6 Likes
Great job, thnx for the feedback!!
1 Like
Hi, want to open this old topic again, since I’m quite sure I am not the only one:
Using HA now for over 2 years and overall I am really happy with it! We have quite a few devices integrated with Alexa using this tutorial: Amazon Alexa Smart Home Skill - Home Assistant
Alexa commands basically work, but responses really slow / timeouts!
RaspPi with, KNX Gateway are connected with local ethernet cables, internet connection 100 Mbit/s fiber. Ping of DuckDNS domain from another host < 10ms. Alexa devic with 4/4 WiFi signal.
Any ideas how to speed up the process / reduce the timeouts with Alexa?
I once tested Google which was much faster, but I am not so happy with the other features of the Nest speakers.
Not sure what could be affecting your setup, but since they deployed the relayer when I originally reported this ~2y ago I’ve not had a single timeout. Currently using echo pop. Rest of my setup is the same from OP with few more devices.
I know this is not what you wanted to hear, but I got nothing else :<
edit:
The only thing I would try to change to test if you can on your end is swap out the pi. I had originally run the HA alone on 4gb, later 8gb rpi4 and while I thought I feel it’s snappy, I was cured of my misconception soon as I moved the HA container to a more powerful machine. Especially as my number of devices, automations and such grew I could feel the instance struggling at times.
This is just my opinion, please rpi enthusiasts do not stone me:
Rpi is perfectly capable of running stuff that you need once in a blue moon or just running on background regardless of speed. Yes it is even capable of running Kodi and such and yes it’s usable. But soon as you move your media server to more powerful machine you’ll realize there is usable and then there is “i click and it happens”. Especially heavy disk access brings pi to its knees even if you use usb drive. Just my 2 cents
FWIW: I saw occasional timeouts and similar interaction (like asking for an action/routine, Alexa taking a long time to respond, responding with a negative and then the action happening anyway) when I was using a UDI product at my previous residence. So there may be some part of this that is just purely AWS/Amazon.
It would be better if it were entirely local - and I guess the new Voice/LLM stuff coming out is intended to do that - but then I need to build my own microphone arrays??? (I still haven’t figured out what people are using for that.)
I want to add to this to help others. I’ve also noticed Alexa voice control become very flakey and uncooperative over the last 6 months or so, culminating in it barely working for the last month.
I was using duckdns and I did some web test DNS lookups and sometimes the reponse is horrific - like 15 seconds or more.
I’ve switched over to Dynu DNS (free ) and voice control is back to being almost instantaneous.
Switching DNS provider took a little head scratching and googling but I got there. In essence, you need to tell HA about the new dns address of HA, then I followed the YT tutorial I previously used to set up voice control and each time he mentioned the dns name/address - I edited/replaced it in AWS and in Amazon App/Dev manager. I had to recopy a key, from one to the other, and I had to re-edit a text file with one new line with that key, but it was just a case of rewatching the 18 min tutorial and just diving in when the DNS address was mentioned (Everything Smarthome is the tutorial I used).
This morning it was all dead again but realised my router reboots at 7am every day (inaccesible 5G router) and the DYNU DNS addon in HA had subsequently stopped itself at 6.58am. The external IP was now wrong. I’ve now enabled the watchdog in the DYNU DNS addon so hopefully it shoudl auto recover each morning instead of stopping!
I also realised my Asus router has its own free DDNS which is fast too, but, I reckon I’ll leave it alone now - rather not have my HA voice assistant totally reliant on one router in case I change it one day and am left scratching my head!
1 Like
Hi @dodgey99, I have also switched recently to Dynu DNS add-on and was happily surprised how much faster it is compared to DuckDNS when using Google Home voice commands.
I was not even using voice commands with Google Home with DuckDNS as I was always getting the “Home Assistant cannot be reached” reply or it was taking ages to execute the command. Now it is almost instantaneous… the way it should be!
I’m interested to know how you tested the DNS lookup to test the response times. Can you please share?
try this