test: Add Dynamo launch mode to L0_infer#8851
Merged
Merged
Conversation
Add a SERVER_LAUNCH_MODE=dynamo path that runs L0_infer through the Dynamo frontend (KServe gRPC) plus an in-process Triton worker, backed by etcd + NATS discovery. Skips test cases that exercise KServe features the Dynamo gRPC path does not yet support (FP16/BF16, classification output, model version selection) and disables the HTTP client paths the Dynamo frontend does not expose.
whoisj
approved these changes
Jun 24, 2026
Contributor
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does the PR do?
Adds a
SERVER_LAUNCH_MODE=dynamopath to the L0_infer test so it can run Triton through the Dynamo frontend (KServe gRPC) plus an in-process Triton worker, backed by etcd + NATS discovery, instead of the standalonetritonserverbinary.qa/common/dynamo_util.sh: sourced in place ofutil.shwhenSERVER_LAUNCH_MODE=dynamo. Overridesrun_server/kill_serverto launch the Dynamo frontend + Triton worker, bring up etcd + NATS (binaries shipped in the Dynamo container), translateSERVER_ARGSinto worker flags, and wait until all models in the repository are ready.qa/L0_infer/test.sh: selects the launch mode and sources the matching util script.qa/L0_infer/infer_test.py: skips cases that exercise KServe features the Dynamo gRPC path does not yet support (FP16/BF16 tensors, classification output, model version selection) and disables HTTP-client paths (the Dynamo frontend exposes only KServe gRPC).qa/common/infer_util.py,qa/common/shm_util.py: guard thetritonclient.httpimport behind non-dynamo mode.The default (
triton) path is unchanged.Checklist
<commit_type>: <Title>Commit Type:
Related PRs:
Where should the reviewer start?
Test plan:
Default mode unchanged:
qa/L0_infer/test.shwithSERVER_LAUNCH_MODEunset/triton.Dynamo mode: run inside the Dynamo Triton container with
SERVER_LAUNCH_MODE=dynamo; frontend + worker come up against etcd + NATS, all models report ready, and the supported subset of L0_infer passes (FP16/classification/version cases skipped).CI Pipeline ID: 55686086
Caveats:
Background
Enables L0_infer to validate the Dynamo Triton backend frontend/worker path alongside the standalone server.
Related Issues:
N/A