Many connection issues with Venus A v144/v147

[Jojo-A]

Hi @MarstekEnergy,
as already mentioned in the HA core PR (home-assistant/core#156012 (comment)) I have many connection issues, that lead to flooding the error logs with these messages:

Logger: pymarstek.udp
Quelle: custom_components/marstek/coordinator.py:77
Integration: Marstek
Erstmals aufgetreten: 10:54:39 (83 Vorkommnisse)
Zuletzt protokolliert: 11:03:54

Request timeout: 192.168.0.117:30000
Send request to 192.168.0.117:30000: {"id": 7758, "method": "ES.GetMode", "params": {"id": 0}}
Send request to 192.168.0.117:30000: {"id": 7759, "method": "PV.GetStatus", "params": {"id": 0}}
Send request to 192.168.0.117:30000: {"id": 7760, "method": "ES.GetMode", "params": {"id": 0}}
Send request to 192.168.0.117:30000: {"id": 7761, "method": "PV.GetStatus", "params": {"id": 0}}

as well as

Logger: custom_components.marstek.coordinator
Quelle: custom_components/marstek/coordinator.py:102
Integration: Marstek
Erstmals aufgetreten: 10:55:29 (11 Vorkommnisse)
Zuletzt protokolliert: 11:03:39

No valid data received from device at 192.168.0.117 (device_mode=Unknown, soc=0, power=0) - connection failed
No valid data received from device at 192.168.0.117 (device_mode=Unknown, soc=14, power=0) - connection failed

and

Logger: custom_components.marstek.coordinator
Quelle: custom_components/marstek/coordinator.py:123
Integration: Marstek
Erstmals aufgetreten: 10:55:29 (11 Vorkommnisse)
Zuletzt protokolliert: 11:03:39

Device 192.168.0.117 status request failed: No valid data received from device at 192.168.0.117. Scanner will detect IP changes automatically

This leads to complete connection losses and for some reason the Venus A then "forgets" its configuration (output power settings, API setting,...) and gets permanently unresponsive.

This "feels" like the many API requests make the Venus' firmware crash and restart and it forgets the configs. But I might be wrong...

After changing the request delay and the timeouts, it works much more stable, although I still get thousands of errors per day.

Here is what I changed in the coordinator.py:

        try:
            # Use library method to get complete device status
            # Note: get_device_status catches exceptions internally and returns default values
            # So we need to check the returned data to detect failures
            device_status = await self.udp_client.get_device_status(
                current_ip,
                port=DEFAULT_UDP_PORT,
                #timeout=5.0,  # Increased timeout for device status requests
                timeout=10.0,  # Increased timeout for device status requests
                include_pv=True,
                #delay_between_requests=5.0,
                delay_between_requests=60.0,
            )

As said, this does NOT FIX the communication issue. It just reduces the errors to a level where the Venus does not become permanently unresponsive (reduction from thousands of errors per hour to only a few thousands per day).

Please let me know if I can provide additional data what might help you to trace down that issue!

Greetings and thank you!

[Jojo-A]

This leads to complete connection losses and for some reason the Venus A then "forgets" its configuration (output power settings, API setting,...) and gets permanently unresponsive.

This behavior is also described here:
jaapp/ha-marstek-local-api#85

And they seem to share the same opinion, that the API requests might make the device crash/reset completely which leads to forgetting all settings.
I am quite sure, that this needs to be fixed both in the integration but also in the firmware...

[SHA-NDY]

That's what I observe even with firmware 147. But the connection is more stable now.
Settings getting lost is still a problem especially when the API turns off...

[gizahNL]

It's honestly completely baffling me, you buy a expensive battery for thousands of euros, and it is just crashing daily. And Marstek is doing shit even though being aware of this for months. You'd expect a hotfix within a week, and a proper fix within a month.

[Tsjoecha]

I don't encounter issues at all, but I'm at V147. You can ask the update through their help section in the app Groeten, *Dries Feys* Op do 5 mrt 2026 om 19:49 schreef Gijs Peskens ***@***.***>:

…

*gizahNL* left a comment (MarstekEnergy/ha_marstek#8) <#8 (comment)> It's honestly completely baffling me, you buy a expensive battery for thousands of euros, and it is just crashing daily. And Marstek is doing shit even though being aware of this for months. You'd expect a hotfix within a week, and a proper fix within a month. — Reply to this email directly, view it on GitHub <#8 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/A3Y3YNCFKJS5X2X2XTI3VHT4PHD5DAVCNFSM6AAAAACSA5TDS6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DAMBWHE4TCOJUGQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[Jojo-A]

It's honestly completely baffling me, you buy a expensive battery for thousands of euros, and it is just crashing daily. And Marstek is doing shit even though being aware of this for months. You'd expect a hotfix within a week, and a proper fix within a month.

That's very true, athough one could argue, that Marsteks batteries have still an unbeatable value for the money.
On the other hand: the open API was a definite buying argument for me.
To see, that this feature is so unstable is not only sad: it is impressing that they managed to program it SO BAD, that it behaves as it currently does (crash, reset)!
And what's even more embarrassing is, that they seem to have no one who is willing and able to fix that!

But hey, they added totally irrelevant "features" like the possibility to switch off the LED...
But a fix for a totally screwed up key feature? How dare you...

I am an embedded software engineer since over 15 years. So this is not a random rant. It is justified criticism of FAIL that should have been FIXED SINCE MONTH!

Did you know? Before the unit stalls and crashes, it stops all MPPTs but keeps discharging the battery. So in worst case, you'll lose your day's sun energy harvest AND end up with an empty battery after crash/reset.
Happens to me twice...

[Jojo-A]

I don't encounter issues at all, but I'm at V147. You can ask the update
through their help section in the app
Groeten,

Dries Feys

Op do 5 mrt 2026 om 19:49 schreef Gijs Peskens @.***>:
…

Being on v147 is no solution at all. I am at v147 too (as many in the community) and my unit still has this bad behavior.

[Jojo-A]

I just wrote a quite long (and angry) bug report to Marstek through their app. Let's see what happens. I will keep you updated.

[SHA-NDY]

I just wrote a quite long (and angry) bug report to Marstek through their app. Let's see what happens. I will keep you updated.

They don't reply to my reports for weeks now. Before they did reply in terms of hours, now they seem to be disappeared...

[gizahNL]

I just wrote a quite long (and angry) bug report to Marstek through their app. Let's see what happens. I will keep you updated.

They don't reply to my reports for weeks now. Before they did reply in terms of hours, now they seem to be disappeared...

Same. I've got multiple support requests in the app open, zero response. I've send them an e-mail a few weeks ago, again: zero response. It's like the company disappeared from the internet.

[Jojo-A]

Interesting enough: we are publicly complaining about how bad Marsteks support currently is -> in one of THEIR OWN repositories -> with absolutely NO reaction at all!

Maybe we should try to make it bigger like posting warnings in the comment sections on youtube and encouraging more people to create support requests (even if these won't get replied either). We can post comments and warnings in the comment sections of the seller websites, where we bought the stuff. We can contact these sellers to also contact Marstek in order to get rid of all those bad comments... We should increase the pressure.

[SebastianC90]

Same here, the Open API was the main reason for bying this storage. Now I cannot use it without checking every few hours if it's still alive.
@MarstekEnergy, this Integration ist worthless without fixing this in your Firmware. Please prioritize!

[Jojo-A]

My Venus A crashed just today again. It stopped the MPPTs and discharged the battery and remained in that state until I have factory-resetted the device through the app.

I have of course forwarded this to the Marstek support which again did not reply. I will ask my legal assistance if we can do something about is because 1. the device does not work as intended and 2.) I regularly lose kWh (money).

So for all who are planning to use the local Open API: be warned! It is a total mess and you should look for another solution/product!

[Wvdbeng]

Hi all!

Just shy of completing my first 2 weeks of having my Venus E 3.0, started out doing manual schedules, tried AI mode for a while, an duse this and also the other HA integration for monitoring. I'm on V144 I believe, and normaly i see the unit "crash" after doing a full charge or a full discharge... Although this night it crashed and resetted itself around 00.30. on 25%. After resetting this morning it emptied itself to the factory 12% and..... resetted itzelf agian. Which means i'm losing my charge window around lunch today... again. I considered just disconnecting and using AI charge but that modus sometimes just idle's and does nothing for the full day.... I wrote my obervings to the dutch distributor this morning. Curious to find out what they say.

[Jojo-A]

Sure you can avoid these problems by using Modbus or deactivating the API.
But the API is something we paid for! And the bug is so severe, that it MUST be fixed. So in my opinion all alternatives (like Modbus - which was never an advertised feature for Venus A in the first place) are more a workaround or a forced retreat, not a solution... 👎