Embedding fails - Not able to add documents

[julian-dymek]

When I try to add documents, I get the error down below. I installed Verba in a conda environment with pip. I am able to run llama3 and gpt4, but am not able to add documents.

(INFO) Importing... (ERROR) Delete object! Unexpected status code: 500, with response body: {'error': [{'message': 'could not delete object from vector repo: delete from index "verba_config": delete local object: shard="x8U9x6HOhDpr": store is read-only'}]}.

[donadelden]

I just got the same error, solved by running Verba from a new directory with permissions 777. Something like:

mkdir test
cp .env test
cd test
chmod 777 .
verba run

EDIT: the error came up again, and probably the folder that needs some changes in permissions is .local/share/weaviate/

[thomashacker]

Interesting! Thanks for the issue, we'll have a look

[jamesoflol]

Same.

I'm also getting this message: {"action":"read_disk_use","level":"warning","msg":"disk usage currently at 86.36%, threshold set to 80.00%"

Could that be related? Also, is it configurable?

[lucasrocha7111]

Same here issue here.

Apple M2 Pro MacOS Sonoma 14.5 Python 3.12.3

[lucasrocha7111]

One way to solve it is run using Docker

[samaphp]

I have confirmed that this issue is affected by several factors:

• Project Folder Permissions: Ensure that the repository folder has the proper permissions before running verba start. Incorrect permissions can cause various issues.
• Available Disk Space: Ensure that your disk space is below the threshold. If the disk is nearly full, the system may automatically switch to a read-only mode, which can lead to errors.
• Environment Variables: Make sure that the environment variables OLLAMA_URL and OLLAMA_MODEL are set. You can find an example of these in the .env.example file.

Here are the errors I encountered:

• flush halted due to shard READONLY status
• could not delete object from vector repo: delete from index "verba_config": delete local object: shard="ifrwh8eZAyOJ": store is read-only
• update vector: API Key: no api key found neither in request header: X-Openai-Api-Key nor in environment variable

The "store is read-only" error was misleading in my case because my disk was 98% full. and I was getting this error: Set READONLY, disk usage currently at 93.20%, threshold set to 90.00% You can check the actual disk availability using the command df -h. After reducing the disk usage to below the threshold, everything worked fine:

I hope this helps in resolving the issue.

I'm using verba command line installed using python instructions.