Error during request to Context Chat Backend (ExApp): Expected keys "loaded_sources" and "sources_to_retry" in response.

[troed]

Which version of Nextcloud are you using?

32.0.6.1

Which version of PHP context_chat are you using?

5.3.1

Which version of backend context_chat are you using?

5.3.0

Nextcloud deployment method?

AIO

Describe the Bug

I've set up the backend to use a local ollama installation and model, and after lots of installing/reinstalling I've now ended up with seemingly everything working (after waiting for a while documents have now started to be indexed). However, when I tried to speed this up by enforcing a scan I got what I think is a strange error message.

# sudo -E -u www-data php occ context_chat:scan troed
[troed] Scanned files/The Anomaly - Director's Cut.epub
[troed] Scanned files/redovisningsavdelning_marvik.epub
[troed] Scanned files/stjarnklart.epub

In LangRopeService.php line 309:
                                                                                                                                                
  Error during request to Context Chat Backend (ExApp): Expected keys "loaded_sources" and "sources_to_retry" in response. Please upgrade the   
  Context Chat Backend app to the latest version.                                                                                               
                                                                                                                                                

context_chat:scan [-m|--mimetype MIMETYPE] [-d|--directory DIRECTORY] [--] <user_id>

(I'm unable to find a 5.3.1 version of the backend, thus the mismatch between it and the app)

This was what stats showed at the time:

# sudo -E -u www-data php occ context_chat:stats
ContextChat statistics:
The indexing is not complete yet.
Total eligible files: 5077
Files in indexing queue: 5075
New files in indexing queue (without updates): 5075
Queued documents (without files):array (
)
Files successfully sent to backend: 0
Indexed documents: array (
)
Actions in queue: 0
File system events in queue: 0

The indexing is currently running and "numbers go up" so this report is just about the strange looking error message for the manual scan.

To Reproduce

I've configured the backend via deployment options to use a local ollama installation. I've previously had a default configured backend running, but changed it out. I had difficulties getting nextcloud to understand that I wanted to reset the vectordb and so all context/assistant components were disabled, removed incl. data and then reinstalled again.

I also ran into an issue where the context_backend appid wasn't removed from Nextcloud's database and had to manually go in and delete that row before I could reinstall the backend successfully again.

I don't know if any of this is relevant though.

PHP logs (Warning these might contain sensitive information)

No response

Ex-App logs (Warning these might contain sensitive information)

No response

Server logs (if applicable)

No response

[kyteinsky]

hello,

Error during request to Context Chat Backend (ExApp): Expected keys "loaded_sources" and "sources_to_retry" in response. Please upgrade the

that's an error handling issue, has been fixed and will be released soon. It is not the actual error.
the real error message should be in the ex-app logs and the "data directory"/context_chat.log file. Both these logs can be downloaded from the "admin settings -> Assistant -> Context Chat" section, but as said below only the docker container logs should be enough in this case.

I've configured the backend via deployment options to use a local ollama installation.

there might be some issue with the config, easiest would be to get the docker container's logs like so:

docker logs nc_app_context_chat_backend

if the output is too long,

docker logs nc_app_context_chat_backend |& head -n400
docker logs -n400 nc_app_context_chat_backend

I also ran into an issue where the context_backend appid wasn't removed from Nextcloud's database and had to manually go in and delete that row before I could reinstall the backend successfully again.

would've been nice to know that then so it could be investigated in the app_api repo but good to hear it was solved.

I don't know if any of this is relevant though.

yes, it is, thanks for including them.