vnx-benchmark-throughput-switch on U280 failed

[csunclechen]

Exception in thread Thread-4: Traceback (most recent call last): File "/usr/lib/python3.6/threading.py", line 916, in _bootstrap_inner self.run() File "/usr/lib/python3.6/threading.py", line 864, in run self._target(*self._args, **self._kwargs) File "/home/cjj/.local/lib/python3.6/site-packages/pynq/pl_server/server.py", line 542, in server_proc server = Listener(self.socket_name, family='AF_UNIX', authkey=self.key) File "/usr/lib/python3.6/multiprocessing/connection.py", line 438, in init self._listener = SocketListener(address, family, backlog) File "/usr/lib/python3.6/multiprocessing/connection.py", line 576, in init self._socket.bind(address) FileNotFoundError: [Errno 2] No such file or directory

Exception in thread Thread-5: Traceback (most recent call last): File "/usr/lib/python3.6/threading.py", line 916, in _bootstrap_inner self.run() File "/usr/lib/python3.6/threading.py", line 864, in run self._target(*self._args, **self._kwargs) File "/home/cjj/.local/lib/python3.6/site-packages/pynq/pl_server/server.py", line 542, in server_proc server = Listener(self.socket_name, family='AF_UNIX', authkey=self.key) File "/usr/lib/python3.6/multiprocessing/connection.py", line 438, in init self._listener = SocketListener(address, family, backlog) File "/usr/lib/python3.6/multiprocessing/connection.py", line 576, in init self._socket.bind(address) FileNotFoundError: [Errno 2] No such file or directory

---

ConnectionError Traceback (most recent call last)

in 4 xclbin = '../benchmark.intf0.xilinx_u280_xdma_201920_3/vnx_benchmark_if0.xclbin' 5 #overlay_1 = pynq.Overlay(xclbin, device=pynq.Device.devices[0]) ----> 6 ol_w0 = pynq.Overlay(xclbin, device=daskdev_w0) 7 ol_w1 = pynq.Overlay(xclbin, device=daskdev_w1) ~/.local/lib/python3.6/site-packages/pynq/overlay.py in __init__(self, bitfile_name, dtbo, download, ignore_version, device) 342 343 if download: --> 344 self.download() 345 346 self.__doc__ = _build_docstring(self._ip_map._description, ~/.local/lib/python3.6/site-packages/pynq/overlay.py in download(self, dtbo) 401 Clocks.set_pl_clk(i) 402 --> 403 super().download(self.parser) 404 if dtbo: 405 super().insert_dtbo(dtbo) ~/.local/lib/python3.6/site-packages/pynq/bitstream.py in download(self, parser) 152 153 """ --> 154 self.device.download(self, parser) 155 156 def remove_dtbo(self): in download(self, bitstream, parser) 102 bitstream_data = f.read() 103 self._call_dask(_download, bitstream_data) --> 104 super().post_download(bitstream, parser) 105 106 def get_memory_by_idx(self, idx): ~/.local/lib/python3.6/site-packages/pynq/pl_server/device.py in post_download(self, bitstream, parser) 441 t.year, t.month, t.day, 442 t.hour, t.minute, t.second, t.microsecond) --> 443 self.reset(parser, bitstream.timestamp, bitstream.bitfile_name) 444 445 def has_capability(self, cap): ~/.local/lib/python3.6/site-packages/pynq/pl_server/device.py in reset(self, parser, timestamp, bitfile_name) 316 317 """ --> 318 self._client.reset(parser, timestamp, bitfile_name) 319 320 def clear_dict(self): ~/.local/lib/python3.6/site-packages/pynq/pl_server/server.py in reset(self, parser, timestamp, bitfile_name) 250 251 """ --> 252 self.client_request() 253 if parser is not None: 254 self._ip_dict = parser.ip_dict ~/.local/lib/python3.6/site-packages/pynq/pl_server/server.py in client_request(self) 489 except FileNotFoundError: 490 raise ConnectionError( --> 491 "Could not connect to PL server") from None 492 self._bitfile_name, self._timestamp, \ 493 self._ip_dict, self._gpio_dict, \ ConnectionError: Could not connect to PL server

[csunclechen]

I was trying to test the benchmark. I have two host servers and two alveo U280 cards. each host connect one card through pcie. And two cards connect the switch through qsfp28. Have you met this problem before?

[mariodruiz]

Hi @csunclechen

Can you run this piece of code in the jupyter notebook?

for i in range(len(pynq.Device.devices)):
    print("{}) {}".format(i, pynq.Device.devices[i].name))

If you get the a list of Alveo platforms everything is OK with pynq and XRT environment.

The next thing to consider is the the DaskDevice class, in particular the __init__ method. I am always giving name to the workers when creating the cluster. However, if no name is given to a worker, the default name is tcp://... This default name produces an error

Can you update the __init__ method of the DaskDevice class, import import re as well

def __init__(self, client, worker):
        """The worker ID should be unique

        """
        worker_id= re.sub(r'[^\w]', '_', worker)
        super().__init__("dask-" + worker_id)
        self._dask_client = client
        self._worker = worker
        self.capabilities = {
            'REGISTER_RW': True,
            'CALLABLE': True
        }
        self._streams = {}

[mariodruiz]

@csunclechen,

I was able to reproduce the issue and pushed a bug fix. Can you try the latest notebooks?