redacted redacted redacted
I think the issue you’re running into is actually two separate problems that coincidentally appeared after the upgrade.
1. Model changes not visible without disabling the cache
This happens because Data Services uses multiple caching layers (Hibernate L2 Cache + Spring Cache via Hazelcast), and when models are deployed at runtime, the caches need to be invalidated. The key thing here is that the mgmtp.a12.dataservices.query.reindexing.mode = REBUILD_INDEX only affects the search index (used for queries), not the model cache.
For development environments where you frequently deploy model changes, you can either:
Disable all caching (your current workaround, but cleaner):
# Use the dataservices-no_cache profile, or set these properties:
spring.cache.type=none
spring.datasources.dataservices.jpa.properties.hibernate.cache.use_second_level_cache=false
spring.datasources.dataservices.jpa.properties.hibernate.cache.use_query_cache=false
Or try to disable only the model graph cache (better for performance):
mgmtp.a12.dataservices.cache.modelGraph.enabled=false
For production, you’d want caching enabled and deploy models via a server restart or init-app.
2. SME Deploy button error: “Failed to delete the existing Data from the Server”
This error occurs because the SME’s “Deploy All” / “Upload Workspace” functionality tries to call the internal seed-data endpoint (/api/v2/internal/seed-data) to delete existing data before uploading new data. Since 2025.06, these endpoints are disabled by default for security reasons.
The solution is to enable the seed-data endpoints in your development environment:
mgmtp.a12.dataservices.seed-data.import.enabled=true
mgmtp.a12.dataservices.seed-data.delete.enabled=true
mgmtp.a12.dataservices.seed-data.export.enabled=true
Important: These should only be enabled in development environments. Never enable them in production, as they allow any user to download, upload, or delete the entire application data.
Further readings:
redacted redacted redacted
I can imagine that maybe the endpoint is still protected by uaa?
Or its somehow related to this sentence from DS docs:
Importing seed data is specifically designed for empty databases. Attempting to import into a non-empty database may result in unexpected errors or data corruption.
GetA12 - Data Services - Import Seed Data as Stream
So the main question what does the error stack says on data-service side when the issue appears?