@JeuFore I don’t have an answer to give you, in this case, I’m completely stuck! Intuitively, since we can’t see anything, and the bug only appears on your end, could it not be a corruption of the SD card? You say that the instance’s database is on a hard drive, but what about Docker and the images, are they on the SD?
Yes, that’s exactly it. So, regarding the DB corruption, does that seem unlikely?
My other containers work fine, and all other Gladys features work as well.
Also, when I write the request in the chat and then restart Gladys, the message is not written.
Perhaps the cause of the problem, but could switching from arm32 to arm64 without changing the DB be the cause?
Have you switched from arm32 to arm64? What does that mean?
If you want to test your theory, launch a fresh Gladys container (you don’t have to add a volume, it’s really just for testing, use a different port), and see if the chat works
I was running Hypriot OS arm32 on my Raspberry and I switched to Ubuntu Server arm64.
Yes, I’ll try that then 
In response to myself, I just tested on a new clean instance, and… same problem gladys crash…
Ahhh, I have some good news. Actually, Gladys isn’t crashing, but it takes trèèèès long to respond. I left my Gladys tab running in the background, and eventually, a response came.
Oh, that’s good, I was surprised.
Are you sure about the performance of your HDD? Is your Pi doing other things at the same time?
The new instance is on my SD card, it is almost new, my HDD is working quite well too, and the Raspberry is not overloaded (on htop CPU at ~5%).
Hello!!
For information, I tested last night and I have the same issue as the one now mentioned by @JeuFore. That is, a chat response time of about 10 to 15 seconds, sometimes 30 seconds on the first two questions. Then no more issues with the following questions.
I retested this morning, exactly the same thing. And this outside of Gladys Plus, from the local IP. Nothing particular in the logs.
EDIT: It actually works on my end (probably the cache)
However, the camera intent does not work for me.
2021-05-17T09:15:33+0200 <error> index.js:16 (process.<anonymous>) NotFoundError: Answer with intent camera.get-image-room.fail and language en not found
at Brain.getReply (/src/server/lib/brain/brain.getReply.js:18:11)
at MessageHandler.replyByIntent (/src/server/lib/message/message.replyByIntent.js:15:27)
at Camera.command (/src/server/lib/device/camera/camera.command.js:32:25)
at EventEmitter.emit (events.js:315:20)
at Event.emit (/src/server/lib/event/index.js:8:16)
at MessageHandler.create (/src/server/lib/message/message.create.js:47:14)
at async MessageHandler.handleEvent (/src/server/lib/message/message.handleEvent.js:17:5)
Oh good, if you’re not the only one @JeuFore
When you say « chat response in 10s », does that mean if you ask the question, refresh the page, and the answer is there?
If so, it’s actually a bug that’s not related to the chat but a known websocket bug (the chat responding too late is just a symptom since websockets are used by the chat)
We have an issue tracking the problem:
Interesting bug but unrelated, I checked the error is legitimate, it’s missing a translation of a sentence in English ^^
Well, I just tested on my old installation, the page (when I refresh) still takes 2 minutes and 58 seconds to load, but without an answer 
I don’t know if it’s just a websocket issue.
EDIT: Well, after waiting 3 minutes, the chat works correctly again, immediate response and no refresh needed.
It’s therefore well linked to what I said earlier… I’m very strong 
Indeed @Jean34, it didn’t seem like it at first, but that was it 
Indeed well spotted 
My bad then…
I had that before editing my message
I just launched a dev build with quite a few fixes and Zigbee2mqtt device additions:
https://github.com/GladysAssistant/Gladys/actions/runs/992968002
Hello!
Just a quick question as I see you’ve added Zigbee devices. I would like to add (or have added) devices in the Tasmota integration, where should I start/whom should I ask?
Have a great day everyone!