Updates over the weekend have increased PB run times exponentially

[Sycamore967]

Describe the bug
Plan time now takes >10 mins, plans are going stale and PB is no longer reactive. Predbat showing as 'Active' all of the time.

Expected behaviour
This wasn't happening before the updates over the weekend, so that included HA OS and core updates as well as all of the PB updates released.

Predbat version

8.41.1

Environment details

• Inverter and battery setup - GE 2x 9.5 batteries
• Standard HAOS
• Anything else? - NO cloud, all on prem.

I've changed my plan interval to 10 mins (from 5) to try and allow PB to clear it's back log of processing but it hasn't really helped.

I went back to v8.40.14 as a test and processing did speed up a little - it only took 9 mins for PB to go to 'inactive'

Screenshots

Screen shot below shows how long the plan is taking to update - it was taken at 08:05

Next one from the Web UI taken at 08:15

Log file
Full set:

predbat_debug.yaml.txt
predbat_plan.html
predbat.log
apps.yaml.txt

[johnleeshore]

I have noticed the same as well and have changed my plan interval to 15 mins. I have noticed that predbat seems to be constantly churning. I do have load ML on.

[bowen159]

I've found the same with my setup. My friend has as well. Both are HA setups sit on Proxmox and this has happened since the OS 18 install. My friend has restored his to OS17.X and it's resolved it.
My dad uses HA without Proxmox and just directly installed to an SSD and V18 and Predbat are fine. No increase in cycle runtime.

[jrlegg]

I've also noticed the same issue when I updated HA OS to 18 over the weekend, I've found reverting back to 17.3 fixes predbats run time. I'm running HA in a proxmox vm.

[nickgee31]

I found commenting out the givtcp_rest section in apps.yaml fixed it for me:

#givtcp_rest:
#- 'http://homeassistant.local:6345'
#- 'http://homeassistant.local:6346'

[bowen159]

Yes, I'd noticed that Predbat was getting stuck on the rest API line in the logs for 2-3 mins. I've commented it out and now it's computing within the expected time on V18.

[johnleeshore]

I have also noted that Predbat was getting stuck on the rest API line for 3 to 5 mins. I have commented out the rest section in apps.yaml, restarted predbat and it is back to normal and has actually stopped churning over

[Adrian57]

See ”HAOS v18.0 oddness?" In discussions. Changing givtcp_rest to use IPaddresses fixes this.

[johnleeshore]

See ”HAOS v18.0 oddness?" In discussions. Changing givtcp_rest to use IPaddresses fixes this.

Perfect. Works perfectly - thanks for pointing this out