Check if gitlab telemetry issue still persists.

Author: mkhazraee

.gitlab/test_python.sh

@@ -73,20 +73,20 @@ python3 test/python/prep_xfer_perf.py array
 
 echo "==== Running python examples ===="
 cd examples/python
-python3 nixl_api_example.py
 python3 partial_md_example.py
 python3 partial_md_example.py --etcd
 python3 query_mem_example.py
 
 # Running telemetry for the last test
 blocking_send_recv_port=$(get_next_tcp_port)
-mkdir -p /tmp/telemetry_test
 
 python3 blocking_send_recv_example.py --mode="target" --ip=127.0.0.1 --port="$blocking_send_recv_port"&
 sleep 5
-NIXL_TELEMETRY_ENABLE=y NIXL_TELEMETRY_DIR=/tmp/telemetry_test \
 python3 blocking_send_recv_example.py --mode="initiator" --ip=127.0.0.1 --port="$blocking_send_recv_port"
 
+mkdir -p /tmp/telemetry_test
+NIXL_TELEMETRY_ENABLE=y NIXL_TELEMETRY_DIR=/tmp/telemetry_test python3 nixl_api_example.py &
+sleep 1
 python3 telemetry_reader.py --telemetry_path /tmp/telemetry_test/initiator &
 telePID=$!
 sleep 6

test/python/test_nixl_api.py

@@ -204,51 +204,51 @@ def test_incorrect_plugin_env(monkeypatch):
         nixl_agent("bad env agent")
 
 
-def test_get_xfer_telemetry():
-    os.environ["NIXL_TELEMETRY_ENABLE"] = "y"
-
-    agent1 = nixl_agent(str(uuid.uuid4()))
-    agent2 = nixl_agent(str(uuid.uuid4()))
-
-    mem_size = 128
-    addr1 = utils.malloc_passthru(mem_size)
-    addr2 = utils.malloc_passthru(mem_size)
-
-    try:
-        reg1 = agent1.get_reg_descs([(addr1, mem_size, 0, "")], mem_type="DRAM")
-        reg2 = agent2.get_reg_descs([(addr2, mem_size, 0, "")], mem_type="DRAM")
-        agent1.register_memory(reg1)
-        agent2.register_memory(reg2)
-
-        agent1.add_remote_agent(agent2.get_agent_metadata())
-        src = agent1.get_xfer_descs(
-            [(addr1, mem_size // 2, 0), (addr1 + mem_size // 2, mem_size // 2, 0)],
-            mem_type="DRAM",
-        )
-        dst = agent1.get_xfer_descs(
-            [(addr2, mem_size // 2, 0), (addr2 + mem_size // 2, mem_size // 2, 0)],
-            mem_type="DRAM",
-        )
-
-        handle = agent1.initialize_xfer("WRITE", src, dst, agent2.name)
-        st = agent1.transfer(handle)
-        assert st in ("DONE", "PROC")
-
-        while True:
-            st = agent1.check_xfer_state(handle)
-            assert st in ("DONE", "PROC")
-            if st == "DONE":
-                break
-
-        telem = agent1.get_xfer_telemetry(handle)
-        assert telem.descCount == 2
-        assert telem.totalBytes == mem_size
-        assert telem.startTime > 0
-        assert telem.postDuration > 0
-        assert telem.xferDuration > 0
-        assert telem.xferDuration >= telem.postDuration
-
-        agent1.release_xfer_handle(handle)
-    finally:
-        utils.free_passthru(addr1)
-        utils.free_passthru(addr2)
+# def test_get_xfer_telemetry():
+#     os.environ["NIXL_TELEMETRY_ENABLE"] = "y"
+#
+#     agent1 = nixl_agent(str(uuid.uuid4()))
+#     agent2 = nixl_agent(str(uuid.uuid4()))
+#
+#     mem_size = 128
+#     addr1 = utils.malloc_passthru(mem_size)
+#     addr2 = utils.malloc_passthru(mem_size)
+#
+#     try:
+#         reg1 = agent1.get_reg_descs([(addr1, mem_size, 0, "")], mem_type="DRAM")
+#         reg2 = agent2.get_reg_descs([(addr2, mem_size, 0, "")], mem_type="DRAM")
+#         agent1.register_memory(reg1)
+#         agent2.register_memory(reg2)
+#
+#         agent1.add_remote_agent(agent2.get_agent_metadata())
+#         src = agent1.get_xfer_descs(
+#             [(addr1, mem_size // 2, 0), (addr1 + mem_size // 2, mem_size // 2, 0)],
+#             mem_type="DRAM",
+#         )
+#         dst = agent1.get_xfer_descs(
+#             [(addr2, mem_size // 2, 0), (addr2 + mem_size // 2, mem_size // 2, 0)],
+#             mem_type="DRAM",
+#         )
+#
+#         handle = agent1.initialize_xfer("WRITE", src, dst, agent2.name)
+#         st = agent1.transfer(handle)
+#         assert st in ("DONE", "PROC")
+#
+#         while True:
+#             st = agent1.check_xfer_state(handle)
+#             assert st in ("DONE", "PROC")
+#             if st == "DONE":
+#                 break
+#
+#         telem = agent1.get_xfer_telemetry(handle)
+#         assert telem.descCount == 2
+#         assert telem.totalBytes == mem_size
+#         assert telem.startTime > 0
+#         assert telem.postDuration > 0
+#         assert telem.xferDuration > 0
+#         assert telem.xferDuration >= telem.postDuration
+#
+#         agent1.release_xfer_handle(handle)
+#     finally:
+#         utils.free_passthru(addr1)
+#         utils.free_passthru(addr2)