.local - url stops working

This has been up before, but i cannot find a solution anywhere.
When I reboot the R Pi 4 the .local - link works with MSIE 11 and Edge from my W10 laptop… This lasts for some eight hour after which attempts to visit the url ends up in “Canot reach the page”. The built in web server seems to shut down or just refuses to reply.
The problems are there with all my web browsers, that is: Firefox, MSIE 11 and Egde.

Trying to reach the R Pi using the ip returns the same.
It seems as if there is something wrong with the web server.
The inititative is great, but this crappy software kills my enthusiasm and makes me unsure that the job is getting done.
It looks as if the R Pi is doing some network traffic. The LED:s blink like they normally do when sending and recieving traffic. No special pattern.
And yes: Both my laptop and the R Pi are on the same network. As they were yesterday, so to speak…

So I rebooted the R Pi, and now it’s up and running and reachable with both the .local - url AND http://192.168.1.33/ for a few hours. Tomorrow I’ll have to reboot the Pi again.

Am I doing something wrong?
Is my R Pi chewing data even if I can’t login?
And when logged in: All menu choices: Are there any instructions on how the’re supposed to be set?
And finally: ar these reboots destroying something?

It would be helpful if someone from Balena confirms but…

I am running 4 x RPi4 (4 GB) and 2 x RPi3. The RPi3s are idle waiting for work which will fit in 1GB and their web interface is fine. The RPi4s have all lost their web interface but appear to be working as the router shows them occasionally sending and receiving data. Current stats say:

 upload / download
  1. 51 MB / 2.096 GB
  2. 51 MB . 2.1 GB
  3. 47 MB / 2.09 GB
  4. 47 MB / 2.12 GB

Some of the jobs run for many hours and every so often when I open all four web connections, one of them will work again which confirms that it is still doing useful work. Of my four, this has happened to two of them so it looks like while number crunching, the web interface is turned off or has a low scheduling priority so cannot respond in time when the computer is busy assuming each job runs in its own core.

When I did reboot before, it looked like the jobs continued where they left off so there must be checkpoints being kept, but I am not 100% sure about this. However, now I am leaving them to run hoping that they are still working and I just can’t monitor them properly.

So. Kind of “normal” behaviour?! I can live with that but sad that this is not described anywhere, as far as i’ve found. I mean: this is what yo do: Try to find info…

Ok. Thank you for your reply. Stay healthy!

R

Yes, normal behaviour:

Well… I don’t know. I think the built in web server shuts down. Even if the processor is heavily occupied it should reply. Slowly, but still…

R

Hi Vinntec, this is precisely the same behavior I have witnessed. When deep in the protein folding simulation (calculation), the webserver times out before your request is processed. However, as evidenced by your network traffic, the Pi’s are crunching. We’ll take a look at finding a way increase the timeout, increase the priority of the webserver, or both.

3 Likes

Hi @dtischler,

Would it be possible to configure the pi to use less resources to that the .local site is always viewable?

I’m running rosetta@home on a pi4 and every now and again I need to power it off to move it out the way. But I want to be able to shut it down or reboot it via the .local url rather than pulling the power plug.

Is there away to reduce the processor load so it’s always viewable?

Thanks,

Dom