Hello Team,
We are encountering a strange issue after recently upgrading our Weaviate version to 1.25. We are using the Python client (v3) for schema creation and data insertion. After inserting new data into the upgraded version, we are getting unexpected results during filter searches.
Specifically, one of our property names is “year.” When we attempt to filter based on this property, the search returns no documents, even though the records are present in the database. Interestingly, when we search using the UUID, the records are correctly found. Here is the filter we are using:
where_filter = {
“path”: [“year”],
“operator”: “Equal”,
“valueString”: year
}
query_result = (
client.query.get(index, properties).with_where(where_filter).with_limit(2).do())
Could you please advise on how to resolve this issue? — Let me know if you need further adjustments!
- Weaviate Server Version:
- Deployment Method: Kubernetes
- Number of Running Nodes: One node
- Weaviate Version: 1.25.0
hi @sanjeev1678 !!
That’s strange.
Can you reproduce this issue with the python v4 client?
Also, what is the exact version you are running?
Thanks!
Hello @DudaNogueira
Thanks for your response.
I am currently running the Python Weaviate client version 3.18.0.
I noticed an issue when trying to use an older backup file (version 1.18.0) with the new version of Weaviate (version 1.25.0). While the migration works, some column filtering no longer functions properly.
The data is still present in the object, but when I try to apply a filter, it returns an empty result. Is there any way to resolve this issue?
You shouldn’t do that.
If you backup on 1.18, you restore to 1.18, and not to 1.25 or any other different version.
This is because there are some migrations.
Also, you should not skip versions.
So in your case, you restore from 1.18 to 1.18, then upgrade to 1.19.latest, 1.20.latest, and so one.
Regarding the backup, the client will have no influence here, as the backup is a simple rest call to the server.
Let me know if this helps.