Name L4 and above by network hop, and close the #1877 review

.claude/skills/test-all-device/SKILL.md

@@ -19,22 +19,22 @@ Detection / isolation procedures referenced below live in
 5. **Read the marker selector out of the same job** you took the timeout from.
    `st-onboard-a2a3` is two pytest passes, not one: the sweep deselects
    `-m "not sdma"` and a later step runs `-m sdma`. `st-onboard-a5` uses
-   `-m "not pod"` and does not exclude `sdma`. Take the expression from
+   `--exclude-level 4` and does not exclude `sdma`. Take the expression from
    `ci.yml` rather than assuming, so this skill cannot drift from the job it
    reproduces.
 6. **Run through `task-submit`** (§E). On a5, the underlying command excludes
-   only pod tests, so SDMA remains in the sweep:
+   only level-4 tests, so SDMA remains in the sweep:
 
    ```bash
-   pytest examples tests/st -m "not pod" --platform a5 \
+   pytest examples tests/st --exclude-level 4 --platform a5 \
      --device <range-or-$TASK_DEVICE> --pto-session-timeout <timeout> -v
    ```
 
    On a platform whose job separates SDMA tests, the underlying sweep command
    is:
 
    ```bash
-   pytest examples tests/st -m "not sdma" --platform <platform> \
+   pytest examples tests/st -m "not sdma" --exclude-level 4 --platform <platform> \
      --device <range-or-$TASK_DEVICE> --pto-session-timeout <timeout> -v
    ```
 

.claude/skills/test-runtime-device/SKILL.md

@@ -20,15 +20,15 @@ Detection / isolation procedures referenced below live in
    `task-submit`, let it pick via `--device auto --device-num <range size>`.
 6. **Read the marker selector out of the same job** you took the timeout from.
    `st-onboard-a2a3` is two pytest passes: the sweep deselects `-m "not sdma"`
-   and a later step runs `-m sdma`. `st-onboard-a5` uses `-m "not pod"` and
+   and a later step runs `-m sdma`. `st-onboard-a5` uses `--exclude-level 4` and
    does not exclude `sdma`. Both quarantined tests are
    `tensormap_and_ringbuffer`, so the second pass is only needed when
    `$ARGUMENTS` names that runtime.
 7. **Run through `task-submit`** (§E). On a5, the underlying command excludes
-   only pod tests, so SDMA remains in the sweep:
+   only level-4 tests, so SDMA remains in the sweep:
 
    ```bash
-   pytest examples tests/st -m "not pod" --platform a5 --runtime $ARGUMENTS \
+   pytest examples tests/st --exclude-level 4 --platform a5 --runtime $ARGUMENTS \
      --device <range-or-$TASK_DEVICE> \
      --pto-session-timeout <timeout> -v
    ```
@@ -37,7 +37,7 @@ Detection / isolation procedures referenced below live in
    is:
 
    ```bash
-   pytest examples tests/st -m "not sdma" --platform <platform> --runtime $ARGUMENTS \
+   pytest examples tests/st -m "not sdma" --exclude-level 4 --platform <platform> --runtime $ARGUMENTS \
      --device <range-or-$TASK_DEVICE> \
      --pto-session-timeout <timeout> -v
    ```

.claude/skills/testing/SKILL.md

@@ -15,20 +15,20 @@ description: Testing guide and pre-commit testing strategy for simpler. Use when
 
 **Important**: Always read `.github/workflows/ci.yml` first for the current
 `--pto-session-timeout` values. Quarantines are marker-based, so mirror the
-a2a3 sweep with `-m "not sdma"` rather than copying a path list. PTO-ISA
-reproducibility comes from the repo-root `pto_isa.pin`.
+a2a3 sweep with `-m "not sdma" --exclude-level 4` rather than copying a path
+list. PTO-ISA reproducibility comes from the repo-root `pto_isa.pin`.
 
 **CI does not run one flat sweep on a2a3.** Marked tests are quarantined out of
 the general onboard sweep and run in a step of their own, after it, because
-they are only correct in isolation. A5 runs the non-pod corpus, including
+they are only correct in isolation. A5 runs the corpus below level 4, including
 SDMA tests, on both x86_64 and ARM64. Reproducing a2a3 CI means reproducing that
 shape — a bare `pytest examples tests/st --platform a2a3` is *not* what CI runs
 and will report failures that CI never sees:
 
 | Marker | Tests | CI behavior |
 | ------ | ----- | ----------- |
 | `@pytest.mark.manual` / `CASES[*]["manual"]` | Standalone pytest tests / individual scene-test cases; optionally scoped to a platform list | Per-PR main sweep: excluded by default on the selected platforms; dedicated DFX steps: included; `daily.yml`: full sweep with `--manual include` |
-| `@pytest.mark.sdma` | a2a3: `sdma_async_completion_demo`, `prefetch_async_demo`; a5: `sdma_async_completion_demo` | a2a3: the dedicated SDMA step; a5: included in the non-pod sweep |
+| `@pytest.mark.sdma` | a2a3: `sdma_async_completion_demo`, `prefetch_async_demo`; a5: `sdma_async_completion_demo` | a2a3: the dedicated SDMA step; a5: included in the non-network1 sweep |
 
 The a2a3 SDMA demos provision 48 device-only STARS streams, which makes an
 AICore fault take ~306 s to tear down instead of ~0.3 s — so they must not
@@ -74,7 +74,7 @@ pytest examples tests/st --platform a2a3sim \
 
 # All hardware scene tests — mirror ci.yml: deselect the quarantined marker, or
 # those tests fail here and nowhere else
-pytest examples tests/st -m "not sdma" --platform a2a3 --device <range> \
+pytest examples tests/st -m "not sdma" --exclude-level 4 --platform a2a3 --device <range> \
     --pto-session-timeout <timeout>
 
 # The quarantined tests, the way CI runs them — same corpus, selected by the
@@ -83,8 +83,8 @@ pytest examples tests/st -m "not sdma" --platform a2a3 --device <range> \
 pytest examples tests/st -m sdma \
     --platform a2a3 --device <2 devs> --pto-session-timeout <timeout>
 
-# A5 runs the non-pod corpus, including SDMA tests, on both host architectures.
-pytest examples tests/st -m "not pod" --platform a5 --device <range> \
+# A5 runs the corpus below level 4, including SDMA tests, on both host architectures.
+pytest examples tests/st --exclude-level 4 --platform a5 --device <range> \
     --pto-session-timeout <timeout>
 
 # Single runtime

.github/actions/pod-run-pytest/action.yml → .github/actions/network1-run-pytest/action.yml

@@ -1,7 +1,7 @@
-name: Run pod pytest across the pair
+name: Run network1 pytest across the pair
 description: >-
-  Start the peer's L3 daemon, run pod-marked pytest against it, then stop the
-  daemon and pull its logs back. pod-stage has already put the tree and venv on
+  Start the peer's L3 daemon, run network1-marked pytest against it, then stop the
+  daemon and pull its logs back. network1-stage has already put the tree and venv on
   the peer.
 
 inputs:
@@ -21,37 +21,37 @@ runs:
       working-directory: ${{ github.workspace }}
       run: |
         set -euo pipefail
-        source "$POD_SSH_HELPER"
-        DAEMON_PORT="$POD_L3_DAEMON_PORT"
-        LOCAL_LOGS="$RUN_DIR/pytest/daemon-${POD_REMOTE_MACHINE}"
-        REMOTE_LOGS="output/pod-ci/pytest/daemon-${POD_REMOTE_MACHINE}"
-        mkdir -p "$LOCAL_LOGS" "$RUN_DIR/pytest/parent-${POD_MACHINE}/ascend"
+        source "$NETWORK1_SSH_HELPER"
+        DAEMON_PORT="$NETWORK1_L3_DAEMON_PORT"
+        LOCAL_LOGS="$RUN_DIR/pytest/daemon-${NETWORK1_REMOTE_MACHINE}"
+        REMOTE_LOGS="output/network1-ci/pytest/daemon-${NETWORK1_REMOTE_MACHINE}"
+        mkdir -p "$LOCAL_LOGS" "$RUN_DIR/pytest/parent-${NETWORK1_MACHINE}/ascend"
         {
-          echo "POD_REMOTE_ENDPOINT=${REMOTE_DAEMON_HOST}:${DAEMON_PORT}"
-          echo "POD_PYTEST_REMOTE_LOGS=$REMOTE_LOGS"
+          echo "NETWORK1_REMOTE_ENDPOINT=${REMOTE_DAEMON_HOST}:${DAEMON_PORT}"
+          echo "NETWORK1_PYTEST_REMOTE_LOGS=$REMOTE_LOGS"
         } >> "$GITHUB_ENV"
 
-        pod_ssh "
+        network1_ssh "
           pkill -f 'python -m simpler.remote_l3_worker --host ${REMOTE_DAEMON_HOST} --port ${DAEMON_PORT}' || true
         " || true
 
-        pod_ssh "
+        network1_ssh "
           set -eo pipefail
           cd '$REMOTE_WORKDIR'
           export PYTHONPATH=\"\${PYTHONPATH:-}\"
           export CMAKE_PREFIX_PATH=\"\${CMAKE_PREFIX_PATH:-}\"
-          source '$POD_REMOTE_CANN_ENV'
+          source '$NETWORK1_REMOTE_CANN_ENV'
           set -u
           source .venv/bin/activate
           export SIMPLER_SCHEDULER_TIMEOUT_MS='$SIMPLER_SCHEDULER_TIMEOUT_MS'
           export SIMPLER_OP_EXECUTE_TIMEOUT_US='$SIMPLER_OP_EXECUTE_TIMEOUT_US'
           export SIMPLER_STREAM_SYNC_TIMEOUT_MS='$SIMPLER_STREAM_SYNC_TIMEOUT_MS'
           mkdir -p '$REMOTE_LOGS'
           export ASCEND_PROCESS_LOG_PATH=\"\$PWD/$REMOTE_LOGS\"
-          echo '[pod-daemon] machine${POD_REMOTE_MACHINE} listening on ${REMOTE_DAEMON_HOST}:${DAEMON_PORT}'
+          echo '[network1-daemon] machine${NETWORK1_REMOTE_MACHINE} listening on ${REMOTE_DAEMON_HOST}:${DAEMON_PORT}'
           python -m simpler.remote_l3_worker --host '${REMOTE_DAEMON_HOST}' --port '${DAEMON_PORT}'
         " > "$LOCAL_LOGS/daemon.ssh.log" 2>&1 &
-        echo $! > "$RUNNER_TEMP/pod-daemon-pytest.pid"
+        echo $! > "$RUNNER_TEMP/network1-daemon-pytest.pid"
 
         # A daemon that dies during import never opens the port, so on its own
         # the connect probe only reports that after the full wait. The ssh
@@ -63,7 +63,7 @@ runs:
         import sys
         import time
 
-        ssh_pid = int(open("$RUNNER_TEMP/pod-daemon-pytest.pid").read().strip())
+        ssh_pid = int(open("$RUNNER_TEMP/network1-daemon-pytest.pid").read().strip())
 
         def ssh_alive():
             try:
@@ -76,7 +76,7 @@ runs:
             except OSError:
                 return False
 
-        deadline = time.time() + ${POD_DAEMON_WAIT_S}
+        deadline = time.time() + ${NETWORK1_DAEMON_WAIT_S}
         while time.time() < deadline:
             if not ssh_alive():
                 print(
@@ -94,7 +94,7 @@ runs:
         sys.exit(1)
         PYWAIT
 
-    - name: Run pod pytest
+    - name: Run network1 pytest
       shell: bash
       working-directory: ${{ github.workspace }}
       env:
@@ -103,39 +103,39 @@ runs:
         set -eo pipefail
         export PYTHONPATH="${PYTHONPATH:-}"
         export CMAKE_PREFIX_PATH="${CMAKE_PREFIX_PATH:-}"
-        source "$POD_CANN_ENV"
+        source "$NETWORK1_CANN_ENV"
         source .venv/bin/activate
         set -u
-        mkdir -p "$RUN_DIR/pytest/parent-${POD_MACHINE}/ascend"
-        timeout "${POD_SMOKE_TIMEOUT_S}s" python -m pytest $PYTEST_ARGS
+        mkdir -p "$RUN_DIR/pytest/parent-${NETWORK1_MACHINE}/ascend"
+        timeout "${NETWORK1_SMOKE_TIMEOUT_S}s" python -m pytest $PYTEST_ARGS
 
     # Runs even when pytest failed, since a device-side failure names its reason
     # only in the daemon or ASCEND logs. The staging tree itself stays for
-    # pod-teardown.
+    # network1-teardown.
     - name: Stop the daemon and collect peer logs
       if: always()
       shell: bash
       working-directory: ${{ github.workspace }}
       run: |
         set +e
-        [ -r "${POD_SSH_HELPER:-}" ] || exit 0
-        source "$POD_SSH_HELPER"
+        [ -r "${NETWORK1_SSH_HELPER:-}" ] || exit 0
+        source "$NETWORK1_SSH_HELPER"
 
-        PID_FILE="$RUNNER_TEMP/pod-daemon-pytest.pid"
+        PID_FILE="$RUNNER_TEMP/network1-daemon-pytest.pid"
         if [ -r "$PID_FILE" ]; then
           kill "$(cat "$PID_FILE")" 2>/dev/null
           rm -f "$PID_FILE"
         fi
-        DAEMON_PORT="${POD_L3_DAEMON_PORT:-19073}"
-        pod_ssh "
+        DAEMON_PORT="${NETWORK1_L3_DAEMON_PORT:-19073}"
+        network1_ssh "
           pkill -f 'python -m simpler.remote_l3_worker --host ${REMOTE_DAEMON_HOST} --port ${DAEMON_PORT}' || true
           for _ in \$(seq 1 20); do
             pgrep -f 'python -m simpler.remote_l3_worker --host ${REMOTE_DAEMON_HOST} --port ${DAEMON_PORT}' >/dev/null || break
             sleep 0.5
           done
         "
-        mkdir -p "$RUN_DIR/pytest/daemon-${POD_REMOTE_MACHINE}"
+        mkdir -p "$RUN_DIR/pytest/daemon-${NETWORK1_REMOTE_MACHINE}"
         rsync -a -e "$RSYNC_SSH" \
-          "$REMOTE_TARGET:$REMOTE_WORKDIR/${POD_PYTEST_REMOTE_LOGS:-output/pod-ci/pytest/daemon-${POD_REMOTE_MACHINE}}/" \
-          "$RUN_DIR/pytest/daemon-${POD_REMOTE_MACHINE}/" 2>/dev/null
+          "$REMOTE_TARGET:$REMOTE_WORKDIR/${NETWORK1_PYTEST_REMOTE_LOGS:-output/network1-ci/pytest/daemon-${NETWORK1_REMOTE_MACHINE}}/" \
+          "$RUN_DIR/pytest/daemon-${NETWORK1_REMOTE_MACHINE}/" 2>/dev/null
         exit 0