Tests for the Llama-stack integration have been failing for a few days: https://github.com/deepset-ai/haystack-core-integrations/actions/runs/20904352674/job/60055198562
The recent v0.4.0 release contains several breaking changes that could be related.