Webpage freezing/not updating

[Guy559]

Unsure if this is a device issue (iPhone/iPad) or a coding issue.

On multiple occasions, when I open the diy-battery.local page to seek app status, the page is either frozen or the data is blanked out (photos). This has happened across iPad/iPhone/Win desktop/Android tablet. Unsure if it’s something with these devices or resident within the code itself.

I’ve left the pages open for 10-15min with no update. Other times, when I use a new computer to access diy-battery.local, the website opens correctly there…and seems to spur the original device to finally update the information. And, sometimes not.

On a positive note, this seems limited to web interface. When I log into HA, I can see MQTT information for diy-battery device updating at around 1-2Hz, so all good there.

[kommando828]

On windows ctrl F5 refreshes the page, all these browsers are now using data saving for pages that are not being used, so have 20 tabs open and the data for most will be put to sleep after a set period, going back into the tab and you get issues until you refresh the page.

[sijones]

If you leave a page for a while the websocket which the data is sent over will be closed, you would need to refresh the page (F5) and then it will reconnect.

On a side note, please use the discussions area to check things before opening an issue.

[Guy559]

@sijones erroneously closed this issue without resolution.

Obviously F5 / Command-R refreshes a webpage. While I didn’t originally state it, webpages across multiple decides were refreshed with no resolution from the app, which is most certainly an issue if websocket/controller isn’t updating/communicating.

Typical response is to clarify whether an issue is actual or not before saying “refresh the page” and then closing the issue without confirmation. Discussions are typically for just that — discussions, as is clarified by available topics like “General Discussion” or “Show and Tell.” None specify “Troubleshooting” or similar.

As of now, diybatterybms app still not providing a webpage - regardless of restarting browser, refreshing page, using an entirely different web browser on a different computer for the first time…or standing on one foot while pouring a cup of coffee. The ONLY way to get one back is to restart the device. THEN the device will offer a webpage again.

[sijones]

Please write the build you are using, the board and CAN chip in use, how long does it take before the page doesn't work, and then you'll need to then have a serial log of the event.

This is the first time anyone has brought this up and I have a lot of these devices running builds, so it's unique for you at the moment, so I'll need all the details.

[Guy559]

Lilygo T-2CAN board:

• ESP32-S3-WROOM-1U（MCN16R8）
• MCP2515 CAN chip (x2)
• Lilygo repository for reference: [https://github.com/Xinyuan-LilyGO/T-2Can](url)

DIYBATTERYBMS Build 2.7 release

Webpage becomes unresponsive after 6-18 hours. I just power cycled 45 mins ago, so I will periodically check to provide a better estimate of time-to-unresponsiveness.

I've not pulled a serial log before. Is this accomplished through USB-C using Visual Studio? Let me do some searching to find best way to produce this for you, as the device is inside my inverter's cable box. Will require some minor disassembly to reach serial port.

Thanks for assisting. Definitely a low-priority issue given that required functionality continues to work as expected (MQTT/CANbus/VE.direct comms).

[sijones]

Using VS Code, there is a button on the bottom to open the serial port, it will then start showing all logging information, once it occurs, copy the screen by using mouse to select what's there, practice a bit, you'll get the hang of it.

Can also use Putty.

[Guy559]

As I understand it, mDNS is for IP address resolution, correct? If so, that doesn’t appear to be the issue, as the problem occurs when typing in either diy-battery.local OR the assist end IP address.

The webpage still loads for the data presented is essentially zero’ed out or nil. Photos attached:

[kommando828]

Check your routers DHCP lease time and max it out, if your router has a MAC to IP address table to effectively make a DHCP assigned IP address fixed then add the Lilygo's MAC and IP.

[Guy559]

Thanks for suggestion. This has nothing to do with DNS, mDNS, or DHCP leasing. It’s not a networking issue per se, as my browser can resolve DIYBATTERYBMS wither with direct IP or through mDNS.

The issue is that, periodically, once the website for DIYBATTERYBMS is pulled up, the information within is zero’ed out or nil.

I’ll keep monitoring and update if I can pinpoint behavior. For example, last night I noticed it had flaked out again, so I shrugged and figured I’d get to it today. Then, when I unlocked my ipad with webpage already open, it automagically updated. So, I can’t say what’s causing it.

Please feel free to throw this one on the back burner until I can provide additional info. Appreciate the assist all around.

[sijones]

try one of the builds from this issue: #38

I uploaded a new build and it might be similar issue.

[Guy559]

Observing some interesting behavior.

Stepped away for a couple hours as system continues to charge via solar. Two instances open: an iPad and a Windows PC running the Edge browser. When I returned and opened the iPad, the webpage was frozen and would not update. I next checked the free heap, which had dipped from 200k to ~175k. Refreshed page on iPad, used manual IP address, etc., to no avail.

I then went to the Windows PC. Edge browser had also frozen. This was the original websocket instance. When I refreshed the page, the information began displaying correctly, free heap climbed back up to 200k, and the iPad webpage began updating automagically.

[Guy559]

Does an open instance of DIYBATTERYBMS webpage take up free heap?

Was able to replicate this tonight: Fresh instance of webpage for DIYBATTERYBMS would load but all information zeroed out or nil (data, settings, etc). When I looked at reported free heap, it was back down to ~175k.

I remember that I had the diy-batterly.local webpage up on my garage computer. As soon as I switched back to the diy-battery.local tab, the free heap jumped back up to ~195-200k and the already open webpage on iPad immediately updated. Odd behavior.

Makes me wonder if this issue is tied to memory utilization.

Also makes me wonder: is having the diy-battery.local tab open (but timed out) on a computer somehow dragging this functionality into a sleep state of some sort. Put another way, if I don't keep this tab open, will the issue occur? Or, will the issue still occur and now I'll be unable to restore w/o power cycling because, til now, I can just refresh the original webpage instance and bring the web-based reporting back across all platforms.

[Guy559]

Seems this only occurs when a webpage is left open. I have not had the problem reoccur when closing all webpages referencing the device's webapp.

If I do keep a webpage open, there appears to be a small memory leak of some sort, as free heap decreases by around 10-15%. At that point, new instances cannot be opened unless original webpage is accessed and refreshed.