I’m on 1.3.0 and it isn’t prompting me to upgrade to 1.3.1. Hmmm… I’ll get that figured out.
@NathanCu Nathan, please provide input in this discussion https://github.com/SpanPanel/span/discussions/178. The new SPAN API does not provide that particular sensor and while we derived other sensors that were also removed the case for dsm_state is a bit nuanced.
TL;DR We could add the sensor back but you may want to determine if that is really your best course based on the actual signals from the panel.
Thanks for your input.
@turtle2472 if you are referring to HACs, you need to pick the option to select anther version and then at the top of the list that appears pick the latest 2.0.x beta.
Have a backup in case you need to go back although I upgraded from 1.3.1 and did not have any issues.
I use it to inform power state across the entire power system of the home from the grid all the way in for the LLM… That signal basically acts as a single summary go no go… (just like a BMS) I’ve given it a map and it uses the DSM signals to give ‘power state’. Don’t make the LLMs ‘guess’ you tell them if it’s good. And that sensor was PERFECT
So if that specific sensor isn’t there I need a ‘we’re good boss’ signal. Observability… So if you want smart grid I need green yellow reds across the power system. If you remove it I MUST rebuild it with a template and then get into a potential issue where I may miss stuff. I’d rather the system tell me. (and yes I read the link I need grid up, overall state. I don’t even have a battery rn so it was a solid you have power signal…)
I just upgraded to 1.3.1 this morning after I posted. I might feel adventurous and attempt the 2.0.x beta later this week. I’ve got some things going on that I don’t know if I want to “chance” an issue with the beta.
I use HA + Span to shed my loads with everything in “Must Have” right now. Our power has been flaky so I don’t want to have issues when I can’t pay attention to it.
That’s an interesting question. Will the load shedding selectors start working in HA because they’re API driven now?
Thats a huge win…
The priorities now work as documented in the beta readme.
We derived dsm_state in v2.0.1b9, the discussion was updated so you know the semantics. The sensor was removed by in the eBus schema or at least not implemented. But we now use a combination of signals which should get you what you are after. If you have additional comments in the discussion we can adjust.
I read the change that’s PERFECT a derived state from the device controller is waaaay better than anything I could build.
LLMs all say thanks. (they natively understand DSM on grid and I don’t have to add context for the to behave)
If the grid goes down and you don’t have a battery how does the LLM work unless your’e running a local LLM on a host that is on UPS same as the router or 5G Internet? But even the 5G case relies on the LLM being there.
Maybe I misunderstood your comment or your install uses solar off-grid but of course that would only work in the day. Always interesting to see how folks are using LLM’s.
That my friend, has its own ring zero ups system. It’ll stay up for about an hour with internet access if service is available. It’s like a data center design but…
Diet pocketbook.
I am just about ready to publish this eBus/Pub-Sub release. Haven’t heard any feedback from those that might be running a version, if any.
@donaldcjackson has been great about answering some edge case questions regarding eBus so a big shout out to him and we are glad he joined us at SpanPanel.
The latest pre-release is 2.0.1RC1. Works for SPAN 32 at the moment.
I’ve tested my own upgrades from 1.3.1 and new installs in various configuraiton but I don’t have an integrated battery although simulator says it works. That’s all I can do. Make a backup, but I am relatively confident.
Res publica te vocat
Today might be my day to test then. I think I have some time to do so this afternoon as well.
Edit: :o How do I do the beta/RC again? I’m trying to do this while at work and struggling to recall.
Edit2: Figured it out. RC2 downloading now.
2.0.1RC2
Failed to set up:
Check the logs
Edit: I clicked reload and everything populated “normally”. I’ll poke at it through the day and see if anything stands out to me.
Edit2: I Appreciate seeing are more normal wording for priority. “Stays on in an outage”, “Stays on until battery threshold”, “Turns off in an outage”. No that the old wording was bad, this is just “clearer”.
Thanks for giving RCx a spin.
HACs requires a reboot when you change the integration, sort of a Windows model as opposed to Linux. Not sure why HA just doesn’t tell the custom integration to reload but HA should be setting a notification to reboot. It does for me.
I just upgraded to RC2 from 1.3.1. Generally everything is working. I did need to delete the priority and switch for the breaker used for the Drive unit. One thing that seems odd is HA is telling me I have 3 other SPAN units, all with the IP of 10.42.1.1, which I believe is the IP used so the SPAN can communicate w/ the Tesla PW3 / gateway, and it doesn’t provide a way to tell it to ignore them.
@sullencat
That is interesting…
IDK how the updated integration decides what SPAN panels exist, so I am going to hold my tongue and wait for somebody like @yodiyossarian to chime in first.
Depending on how that goes, I may have questions about exactly how your SPAN panel, HA, PW, etc are network connected, but let’s wait until the integration’s developers reply…
I suspect it’s related to mDNS, as it does look like the panel is advertising that IP in it’s broadcasts (along with the wired and wireless IPs on the main network).
Looking further at it, I think async_step_zeroconf in config_flow.py needs to abort if it sees that IP (it’s much more likely that that IP would be seen related to a Tesla / SPAN link than someone actually using 10.42.x.x and assigning the .1 IP to the SPAN panel).
And it’s now up to 5 instances of the SPAN panel with that IP it wants me to add. Hopefully an exclusion can be put in soon before I have 200 panels waiting to be added at the top of the Integrations page.
10.42.1.1 looks like a gateway address and it the would appear that zeroconf should perhaps ignore that pattern. Gateways are of course the path that mDNS takes when advertising across VLANS
What do you think?
