Slowdown of home assistant

neox387 commented 9 months ago

Since installation my home assistant web pages can sometimes take up to 10 seconds to load This is also causing issue with other services for example a kodi ping starts to fail

The registers that failed to read can vary while other times they are reading normal and happends on these: read_input_register / read_holding_register

2023-12-31 14:05:22.828 DEBUG (MainThread) [custom_components.solis_modbus.sensor] register_key = 33019, value = 0 2023-12-31 14:05:25.839 ERROR (MainThread) [homeassistant] Error doing job: Exception in callback _run_async_call_action(, >) at /usr/src/homeassistant/homeassistant/helpers/event.py:1450 Traceback (most recent call last): File "/usr/local/lib/python3.11/asyncio/events.py", line 80, in _run self._context.run(self._callback, self._args) File "/usr/src/homeassistant/homeassistant/helpers/event.py", line 1454, in _run_async_call_action hass.async_run_hass_job(job, time_tracker_utcnow()) File "/usr/src/homeassistant/homeassistant/core.py", line 671, in async_run_hass_job hassjob.target(args) File "/usr/src/homeassistant/homeassistant/helpers/event.py", line 1532, in interval_listener hass.async_run_hass_job(job, now) File "/usr/src/homeassistant/homeassistant/core.py", line 671, in async_run_hass_job hassjob.target(*args) File "/config/custom_components/solis_modbus/sensor.py", line 903, in async_update values = controller.read_input_register(start_register, count) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/config/custom_components/solis_modbus/modbus_controller.py", line 31, in read_input_register raise ValueError(f"Failed to read Modbus register ({register}): {result}") ValueError: Failed to read Modbus register (33029): Modbus Error: [Input/Output] Modbus Error: [Invalid Message] No response received, expected at least 8 bytes (0 received) 2023-12-31 14:05:25.842 WARNING (MainThread) [homeassistant.components.kodi.media_player] Unable to ping Kodi via websocket 2023-12-31 14:05:25.867 ERROR (MainThread) [homeassistant] Error doing job: Task exception was never retrieved

Pho3niX90 commented 9 months ago

Could you give me some insights? What hardware are you using? Have you built charts with the data? What does the hardwares utilization look like before and after installing (if available)

Pho3niX90 commented 9 months ago

Another thing, it seems you have debug logging enabled, this repo has a lot of debugging logs, so turning this on will spam the logs, and likely slow down devices that write to SD cards, so my first suggestion would be to disable debug logging, and see if this resolves the issue.

https://www.home-assistant.io/integrations/logger/

neox387 commented 9 months ago

I had debug loggin enabled from the integration page of modbus to hopefully see more what was wrong its not always on.

Hardware wise I should not have any bottlenecks 19 integrations running no problem everything snappy loading within a second of clicking, have not setup up anything like graphs or automations yet just the integrations

But when I enable this one clicking on something or freshing the page can take up to 10 seconds

cpu utilization have not seen it go over 4%, ram at 7-12%

Fresh Home Assistant OS 11.2 vdi (virtualbox) CPU 2vcores from a Intel N5105 (tested with 4 cores) RAM 4GB ddr4-2933 Storage: 32GB on a raid 5 nvme gen 4 ssds

Datalogger stick: Solis-DLS-L inverter: RHI-5K-48ES-5G

Pho3niX90 commented 9 months ago

Okay it's a peculiar issue. Do you know how to modify the integrations files? Inside the const file you can change the interval of scans, if this works I will prioritize #15

neox387 commented 9 months ago

Ok i did try POLL_INTERVAL_SECONDS = 10 before i made the issue but at the time it did not seem to help

Testing again 10 seconds the value errors are 90% gone did have another ping service with a warning tough 15 after a long time i did get another new one ValueError: Failed to read Modbus register (43010): Modbus Error: [Input/Output] No Response received from the remote slave/Unable to decode response

Using 30 now because this way its less noticable, can still reproduce every 30 seconds doing a reload or interacting with home assistant it will hang for a few seconds.

Not sure how you get the data from the logger is it not 1 call or is it big that it takes a while, or is it my older datastick donno,

but maybe what you said in #15 about not updating all the field every times make a difference.

Pho3niX90 commented 9 months ago

So there are about 5 calls done, to avoid multiple calls. The menory used should be low since it's all U16 values.

NO response received will be noticed from time to time, this can be due to a timeout.

Could you do a ping test from your HA hardware to your inverter to see how stable it is? It could be a timeout issue if perhaps larger than 100ms (need to verify)

neox387 commented 9 months ago

Ping test where 2 times the polling happend and notice page loading for a few seconds but the ping test shows nothing strange and keeps pinging trough the hangs.

33 packets transmitted, 33 packets received, 0% packet loss round-trip min/avg/max = 0.819/1.362/1.841 ms

Tought I eliminated most errors with higher polling rate, running for 24hours+ now and sadly still errors:

Error doing job: Exception in callback _run_async_call_action(, >) at /usr/src/homeassistant/homeassistant/helpers/event.py:1450 11:28:35 PM – (ERROR) Solis Modbus (custom integration) - message first occurred at January 3, 2024 at 10:33:25 PM and shows up 127 times Error doing job: Exception in callback _run_async_call_action(, >) at /usr/src/homeassistant/homeassistant/helpers/event.py:1450 11:20:55 PM – (ERROR) Solis Modbus (custom integration) - message first occurred at January 3, 2024 at 10:32:19 PM and shows up 171 times>

neox387 commented 9 months ago

I'll test this weekend on my desktop with the VMware image to see if the same thing happens.

Did not test with new version maybe #29 helps

neox387 commented 9 months ago

Test on desktop with latest 11.3 vmware image with only this integration and can reproduce slowdown every 15 seconds

If others don't have this I suspect it's because of a different datalogger

Pho3niX90 commented 9 months ago

Are you able to do a short video of the slowdown?

Even if the datalogger is to blame, it's not suppose to slow down your system, so will have a look over the weekend to see what might be causing this.

neox387 commented 8 months ago

refreshing the page every second you see it at 14 and 30, just before the next polling at 45 I click on system info.

https://github.com/Pho3niX90/solis_modbus/assets/6033262/b361a498-7ab9-4726-bd52-ec8fc1dad068

Pho3niX90 commented 6 months ago

@neox387 it's been a while since you opened this issue, and not sure if you are still using the integration.

If you are, could you please see if version 1.4.0-beta fixes the issue for you?