WARNING:Modbus connection failed WARNING:Failed to scrape inverter, sleeping until next scan

[kisa72]

Hi,

I have seen the previous questions/answers relating to the discussuions #34 and #48, and although I downgraded pymodbus, I still get the warnings for more than half the attempts to connect to the inverter and it normally occurs in sequences of say 5-6 failed attempts in a row.

One thing I found that may provide a reason is that it only occurs when I'm running my code in Debian Linux on my Laptop (which doubles as my Plex server).

When I run the same code in parallel on my Windows 11 PC through PYCharm, I'm lucky to get 1 in 30 attempts produce a warning.

Linux
45714.653348
WARNING:Modbus connection failed
WARNING:Failed to scrape inverter, sleeping until next scan
45773.712014
WARNING:Modbus connection failed
WARNING:Failed to scrape inverter, sleeping until next scan
45832.740916
WARNING:Modbus connection failed
WARNING:Failed to scrape inverter, sleeping until next scan
45891.819905
WARNING:Modbus connection failed
WARNING:Failed to scrape inverter, sleeping until next scan
45958.880418
WARNING:Modbus connection failed
WARNING:Failed to scrape inverter, sleeping until next scan

Windows
48417.061711
now.minute 26
48476.695655
now.minute 27
48536.333087
now.minute 28
48596.128118
now.minute 29
48655.688589
now.minute 30

The 'now.minute' is a print statement in both the Linux and Windows codes, but because the Linux one failed to connect, it restarts it's main loop and doesn't get to my 'now.minute' print statement.

I think this is quite significant and seems to point at an underlying Linux related issue.

Understanding the lower level communication methods/libraries of Linux or Windows is beyond my knowledge, but if someone has any ideas of why the same code is failing (50% of the time) on a Linux machine compared to never on a Windows machine, I'd be keen to understand why and it may resolve some others issues as as well.

Thanks,

KISA

[michael-robbins]

What's your sync frequency? I've seen it sometimes get really flaky when it's set like <10s