sshiv012 opened a new issue, #9081: URL: https://github.com/apache/texera/issues/9081
### Task Summary Pin the LiteLLM image in the Kubernetes Helm chart to the same version the single-node deployment uses. Right now the k8s chart uses the moving tag `litellm/litellm:main-stable` (`bin/k8s/values.yaml` and `bin/k8s/values-development.yaml`). Every time a pod restarts it can pull a different LiteLLM build, without anyone changing Texera. That is how #9080 reached our k8s deployment: a newer LiteLLM started compressing chat replies, and the agent broke. The single-node deployment already pins `ghcr.io/berriai/litellm:v1.83.10-stable` (`bin/single-node/docker-compose.yml`), so k8s and single-node currently run different LiteLLM versions. | Deployment | Today | After | | --- | --- | --- | | single-node | `ghcr.io/berriai/litellm:v1.83.10-stable` | unchanged | | k8s (`values.yaml`, `values-development.yaml`) | `litellm/litellm:main-stable` (moving) | `ghcr.io/berriai/litellm:v1.83.10-stable` | This does not replace the proxy fix for #9080; it only stops surprise LiteLLM upgrades. Future LiteLLM upgrades become a deliberate version bump in both places. ### Task Type - [ ] Refactor / Cleanup - [x] DevOps / Deployment / CI - [ ] Testing / QA - [ ] Documentation - [ ] Performance - [ ] Other -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
