Using featureGates to enable EPP flowControl feature#973
Conversation
There was a problem hiding this comment.
Pull request overview
This PR switches EPP “flow control” enablement from an environment variable toggle to a flowControl feature gate configured via the EPP ConfigMap, and updates docs + deployment defaults accordingly.
Changes:
- Update docs to describe enabling EPP flow control via the
flowControlfeature gate. - Update deploy logic to patch the EPP image (when needed) and enable
flowControlin the EPP ConfigMap for scale-to-zero/e2e setups. - Bump the default
LLM_D_INFERENCE_SCHEDULER_IMGversion indeploy/install.sh.
Reviewed changes
Copilot reviewed 4 out of 4 changed files in this pull request and generated 4 comments.
| File | Description |
|---|---|
| docs/user-guide/scale-from-zero.md | Updates prerequisites to reference flowControl feature gate instead of env var. |
| docs/developer-guide/testing.md | Updates testing guidance to match new flowControl enablement mechanism. |
| deploy/lib/infra_llmd.sh | Implements ConfigMap-based flowControl enablement + conditional image patching for EPP. |
| deploy/install.sh | Updates default EPP image tag used for flowControl-capable scheduler. |
Signed-off-by: Braulio Dumba <Braulio.Dumba@ibm.com>
|
/ok-to-test |
|
🚀 Kind E2E (full) triggered by |
|
🚀 OpenShift E2E — approve and run ( |
GPU Pre-flight Check ✅GPUs are available for e2e-openshift tests. Proceeding with deployment.
|
|
/ok-to-test |
|
🚀 Kind E2E (full) triggered by |
|
🚀 OpenShift E2E — approve and run ( |
GPU Pre-flight Check ✅GPUs are available for e2e-openshift tests. Proceeding with deployment.
|
|
/ok-to-test |
|
🚀 Kind E2E (full) triggered by |
|
🚀 OpenShift E2E — approve and run ( |
GPU Pre-flight Check ✅GPUs are available for e2e-openshift tests. Proceeding with deployment.
|
This PR improves the E2E test with the following updates:
v0.7.0LLM_D_INFERENCE_SCHEDULER_IMGfeatureGatesto enable EPP flowControl feature (remove configuration via varENABLE_EXPERIMENTAL_FLOW_CONTROL_LAYER)ENABLE_EXPERIMENTAL_FLOW_CONTROL_LAYERThis PR is addressing the issues reported in the following PR: #968