Skip to content

test(mm-router): move new MM-router tests to post_merge #9452

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dmitry-tokarev-nv merged 1 commit into from
May 12, 2026
+10 −10
Merged

test(mm-router): move new MM-router tests to post_merge #9452

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion tests/mm_router/test_router_rust_mm_frontend_decode_e2e.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,7 +53,7 @@
NAMESPACE = "router-rust-mm-fed"

pytestmark = [
pytest.mark.pre_merge,
pytest.mark.post_merge,
pytest.mark.e2e,
pytest.mark.vllm,
pytest.mark.multimodal,
Expand Down
2 changes: 1 addition & 1 deletion tests/mm_router/test_router_rust_mm_router_e2e.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,7 +51,7 @@
NAMESPACE = "router-rust-mm"

pytestmark = [
pytest.mark.pre_merge,
pytest.mark.post_merge,
pytest.mark.e2e,
pytest.mark.vllm,
pytest.mark.multimodal,
Expand Down
16 changes: 8 additions & 8 deletions tests/serve/multimodal_profiles/vllm.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -102,10 +102,10 @@
),
# `agg_router` exercises agg_multimodal_router.sh: Rust frontend
# with the `lightseek-mm` feature, MM-aware KV routing, multi-worker.
# Smoke-level on pre_merge so regressions to the script's plumbing
# Smoke-level on post_merge so regressions to the script's plumbing
# (worker boot order, ZMQ KV events, MM-routing build) surface in
# gating CI before they merge. The fine-grained routing-correctness
# assertions live in tests/mm_router/test_router_rust_mm_router_e2e.py.
# CI. The fine-grained routing-correctness assertions live in
# tests/mm_router/test_router_rust_mm_router_e2e.py.
#
# The payload sends two identical MM requests and asserts the
# second sees cached_tokens > 0 — proves the warm worker reused
Expand All @@ -114,7 +114,7 @@
# text-prefix only, both requests would still succeed but the
# second's cached_tokens would be 0 and this case would fail.
"agg_router": TopologyConfig(
marks=[pytest.mark.pre_merge],
marks=[pytest.mark.post_merge],
timeout_s=400,
profiled_vram_gib=18.7,
requested_vllm_kv_cache_bytes=1_719_075_000,
Expand All @@ -124,13 +124,13 @@
# The chat-processor variant of the MM-aware router: same routing
# architecture, but the frontend uses --dyn-chat-processor=vllm
# (Python preprocessor) instead of the Rust+lightseek path. Kept
# on pre_merge alongside the default so both entry points stay
# covered by gating CI; the routing assertions are equivalent.
# on post_merge alongside the default so both entry points stay
# covered by CI; the routing assertions are equivalent.
# SINGLE_GPU=true packs both workers onto GPU 0 to match the
# single-GPU CI environment (the chat-processor script's own
# default is false for production multi-GPU usage).
"agg_router_chat_processor": TopologyConfig(
marks=[pytest.mark.pre_merge],
marks=[pytest.mark.post_merge],
timeout_s=400,
profiled_vram_gib=18.7,
requested_vllm_kv_cache_bytes=1_719_075_000,
Expand All @@ -147,7 +147,7 @@
# CI; the content-hash correctness assertion lives in
# tests/mm_router/test_router_rust_mm_frontend_decode_e2e.py.
"agg_router_frontend_decode": TopologyConfig(
marks=[pytest.mark.pre_merge],
marks=[pytest.mark.post_merge],
timeout_s=400,
profiled_vram_gib=18.7,
requested_vllm_kv_cache_bytes=1_719_075_000,
Expand Down
Loading