ValueError: filedescriptor out of range in select()

Benj-HaRRison commented 5 days ago

System Health details

System Information

version	core-2024.11.1
installation_type	Home Assistant OS
dev	false
hassio	true
docker	true
user	root
virtualenv	false
python_version	3.12.4
os_name	Linux
os_version	6.6.54-haos
arch	x86_64
timezone	Europe/Kyiv
config_dir	/config

Home Assistant Community Store

Home Assistant Cloud

Home Assistant Supervisor

Dashboards

Recorder

Checklist

[X] I have enabled debug logging for my installation.
[X] I have filled out the issue template to the best of my ability.
[X] This issue only contains 1 issue (if you have multiple issues, open one issue for each issue).
[X] This issue is not a duplicate issue of any previous issues..

Describe the issue

Every few days (from 12 hours to two days) the problem “ValueError: filedescriptor out of range in select()” appears steadily, with all sensors falling into the “Unavailable” status or sensor values “freezing” at the last received data.

The problem cannot be reproduced by force; it appears randomly. I didn’t notice any connections: the problem can appear at 9 pm, 5 am or 12 noon. In this case, no manipulations are performed: no one reconfigures anything in Home Assistant or touches the inverter.

The following messages appear in the HA log: Регистратор: homeassistant Источник: /usr/src/homeassistant/homeassistant/runner.py:147 Первое сообщение: 11:59:56 (4848551 сообщений) Последнее сообщение: 15:45:02

Error doing job: Exception in callback SerialTransport._read_ready() (None) Traceback (most recent call last): File "/usr/local/lib/python3.12/asyncio/events.py", line 88, in _run self._context.run(self._callback, *self._args) File "/usr/local/lib/python3.12/site-packages/pymodbus/transport/transport_serial.py", line 128, in _read_ready if data := self.syncserial.read(1024): ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/local/lib/python3.12/site-packages/serial/serialposix.py", line 565, in read ready, , _ = select.select([self.fd, self.pipe_abort_read_r], [], [], timeout.time_left()) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ValueError: filedescriptor out of range in select()

To restore work ha-must-inverter I need to restart Home Assistant every time and only then all the data is read again. What helps is restarting Home Assistant:

Removing the cable from the USB socket of the inverter and reconnecting does not solve the problem
Removing the cable from the server’s USB socket and reconnecting does not solve the problem
Restarting the integration does not solve the problem
Replacing the USB cable does not solve the problem
Rebooting the inverter does not solve the problem

USB or Serial device path: /dev/ttyUSB0 Baud rate: 19200 Parity (‘E’ven, ‘O’dd or ‘N’one): N Stop bits (1, 1.5, 2): 1 Byte size (7, 8): 8

Windows Server is used as the server, on which the virtual machine VMware Workstation Pro 17.5.0 build-22583795 for Home Assistant is installed. Every time you turn on the virtual machine, the specified USB device is automatically forwarded to VmWare.

Interesting fact: on the Windows Server, a program from the manufacturer "Solar Monitor" is installed and when the problem is present, If you remove forwarding of the USB device to the virtual machine, Solar Monitor in Windows Server works and reads data. If you forward the device back to VmWare, nothing will work until you restart Home Assistant.

Reproduction steps

The problem cannot be reproduced by force; it appears randomly. I didn’t notice any connections: the problem can appear at 9 pm, 5 am or 12 noon. In this case, no manipulations are performed: no one reconfigures anything in Home Assistant or touches the inverter.

Debug logs

Регистратор: homeassistant
Источник: /usr/src/homeassistant/homeassistant/runner.py:147
Первое сообщение: 11:59:56 (4848551 сообщений)
Последнее сообщение: 15:45:02

Error doing job: Exception in callback SerialTransport._read_ready() (None)
Traceback (most recent call last):
  File "/usr/local/lib/python3.12/asyncio/events.py", line 88, in _run
    self._context.run(self._callback, *self._args)
  File "/usr/local/lib/python3.12/site-packages/pymodbus/transport/transport_serial.py", line 128, in _read_ready
    if data := self.sync_serial.read(1024):
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/serial/serialposix.py", line 565, in read
    ready, _, _ = select.select([self.fd, self.pipe_abort_read_r], [], [], timeout.time_left())
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: filedescriptor out of range in select()

Diagnostics dump

28.txt

p4block commented 4 days ago

I can add something to this, I am running out of file descriptors on my system within the <24hrs roughly ever since I added this integration, but I am not 100% sure the culprit is this integration yet.

Check out the output of lsof -w | wc and check out its output too lsof -w > openfiles.txt. The debug log of HA may also show problems with "too many open files"

p4block commented 4 days ago

@p4block are you also using HASS in a docker container?

Yes, with both a ZigBee stick and the rs485 dongle mapped with volumes. I could switch to a privileged container if that solves the problems. I get 100k+ 0t0 pipe files created by something in HA, right now I'm just restarting the container every 6h but it's far from optimal lol

mukaschultze commented 4 days ago

I've updated pymodbus to 3.6.9, this is the same version that HASS core is using and it should be more stable for long lasting connections (There were multiple reports of issues in the core integration very similar to the ones I received here).

I recently moved and haven't had time to setup my inverter in the new house yet, so please let me know if this update improves or worsens the situation. Thank you very much!

p4block commented 2 days ago

@mukaschultze problem seems fixed after yesterday's update, fds don't creep up to the bajillions anymore

Benj-HaRRison commented 9 hours ago

I've updated pymodbus to 3.6.9, this is the same version that HASS core is using and it should be more stable for long lasting connections (There were multiple reports of issues in the core integration very similar to the ones I received here).

I recently moved and haven't had time to setup my inverter in the new house yet, so please let me know if this update improves or worsens the situation. Thank you very much!

I have good news: after updating about 24 hours ago, I stopped restarting HA because HA Must Inverter was freezing. I believe that my problem "ValueError: filedescriptor out of range in select()" was solved by this! 😎😎

But I got a new warning in the HA log. I don’t see what it affects, but it is in the log exactly after the last update...

Регистратор: pymodbus.logging Источник: runner.py:189 Первое сообщение: 26 ноября 2024 г. в 03:53:36 (244163 сообщений) Последнее сообщение: 20:48:32

Failed to connect [Errno 11] Could not exclusively lock port /dev/ttyUSB0: [Errno 11] Resource temporarily unavailable

I'll be watching this. If indeed "ValueError: filedescriptor out of range in select()" is no longer observed, then I will create a new ticket regarding the warning problem...

mukaschultze / ha-must-inverter