LocalAI API endpoint not working #129
Comments
|
|
|
|
|
|
I confirm there is also something wrong on my Local-ai setup with Nextcloud. Everything works fine when is using the GUI from Local-AI, or even a curl API command : results come really fast (1-2 second, I have 2x GPU with CUDA enabled) But when querying through the assistant the answer takes a very long time (usually at least 2-3 minutes). I had a look on a Local-AI with Debug enabled, and meanwhile Local-AI does... nothing. It just waits... and at some points return the answer really quickly. I had a look to the Nextcloud Assistant page, and I can also see this 417 answer doing nothing, and finally gets a 200 answer (the top "new_message" is when the Chat Button is pressed to send the message)
|
|
I forgot to mention my versions:
|
|
I found more information about my issue. The "Chat with AI" (and also other AI sections) queries are only processed once every 5 minutes (at 9:00 / 9:05 / 9:10 / ...), that explains the time before getting an answer... which of course is not the expected behavior. Can someone have a look to reproduce? Thank you |
|
@ctft1 its totally nextcloud or assistant app problem bro, because localAI working perfect and it integrates to any other frameworks with its API without any problem. |
|
@ctft1 that it get processed faster then the 5 minutes (which is your default cronjob schedule) you need to ALWAYS run: But i can confirm the Expectation Failed issue |
|
Since the update of the assistant today it works on my case. |
|
I just updated to 2.1.1 and for me the problem still continues |
|
@ramezquitao they trolling bro, nextcloud just focus on sales not on fix errors |
I can confirm the problem persists, this could be a wonderful feature for Nextcloud but as @userbox020 says, they are focusing all the efforts in other thing. I have my own "ChatGPT", anyone can use it by the way: chat.privategpt.es and I cannot connect Nextcloud assistant using OpenAI nor my own app, both are failing and getting same errors: I hope they pay attention to us because this application should not be hard to build and it could help a lot. |
|
Are you running There is almost no warning about this change. We'll add it in the admin settings of Assistant and integration_openai. Let's try to stay constructive here please. |
I don't know exactly what you mean. First, I am running a dockerized Nextcloud. I am not an expert in PHP, but it is clear that NC is not working as the user expects. I mean, if I send a request to OpenAI or any other AI service like chat.privategpt.es, the user expects to get a streaming response as accustomed. In any case, it is clear the call from NC to the AI service is not arriving, as I see in the logs of chat.privategpt.es. This is my docker compose file: version: '3.7'
services:
nextcloud:
image: nextcloud
container_name: nextcloud
ports:
- 8080:80
volumes:
- nextcloud_data:/var/www/html/data
- nextcloud_config:/var/www/html/config
- nextcloud_apps:/var/www/html/apps
- /home/artivant/Public:/mnt/public
environment:
- NEXTCLOUD_ADMIN_USER=myuser
- NEXTCLOUD_ADMIN_PASSWORD=mypass
restart: unless-stopped
volumes:
nextcloud_data:
name: nextcloud_data_volume
nextcloud_config:
name: nextcloud_config_volume
nextcloud_apps:
name: nextcloud_apps_volumeCould you please try any AI service and reproduce this error? |
|
@hdnh2006 I can't reproduce the issue. and launch a task (or send a message in the chat UI) while the occ command is running?
If the admin sets up a worker that executes the background jobs without delay, the assistant + integration_openai should work as a user expects 😁 |
|
For me, solution give by @julien-nc , works. Thanks |
Running your code while asking a question with "Chat wit AI", this is my terminal log: It worked once or twice - but more often than not, it didn't. (PS: I use Apache with mySQL) |
|
->> It worked with |
|
Solution give by @julien-nc works but I have to keep the terminal open. Is there a way to hardcode the worker process? |
|
@SchBenedikt Awesome! Works like a champ now. Thanks much. |
|
Just running occ background-job:worker works and the AI Chat answers within a couple of seconds. Invalid job class: OCTaskProcessingSynchronousBackgroundJob I am running linuxserver's NC container. |
Which version of assistant are you using?
1.0
Which version of Nextcloud are you using?
v30
Which browser are you using? In case you are using the phone App, specify the Android or iOS version and device please.
Firefox and Google Latest
Describe the Bug
I have setting up LocalAI API endpoint working fine, and NextCloud even detect the models list of LocalAI but when I use any of the features to talk with the LLM it doesnt fetch anything, it get the follow error
Expected Behavior
Interact with the LLM normally
To Reproduce
Setup LocalAI, input the URL in AI settings and interact with the AI
The text was updated successfully, but these errors were encountered: