The SD card is not old but that is probably irrelevant as I run the OS and data partitions on a USB spinning hard drive which is 12 months old approx.
RPi & USB HDD are powered from an official Raspberry PSU rated at 2.5A. The emonTx also has a shop AC-AC adapter.
The unit is ‘semi-down’ at the moment meaning I can see it with a network scanner but not access it. I’m confident that a reboot will clear the error but that’s not fixing it.
Would there be any value in pulling the power and examining the OS file system partition on the USB HDD? - but I’m not sure what to look for.
Also this problem is not unique to this particular emonTx/RPi instance. I have 3 other identical instances monitoring different parts of a largish PV installation. These have experienced this same problem occasionally in the past.
This tends to rule out a hardware problem?? … but not a sub-optimal overall hardware design concept, of course.
I have a main system and a couple of secondary ones that run constantly with no bother.
Do all 2 run on the same wireless network?
[edit] Have you made any changes to the base SD image? Added any other packages at all?
Only other thing might be a full disk. Linux is pretty robust so this is surprising. It is not anything to do wth the EmonTX as that could not impact on the Linux OS.
All you could do is connect up a console (monitor and keyboard) no next time SSH drops out you can more easily examine the system.
Are all 3 TXs within RF range? I presume the TXs are on different nodes so could not interfere with each other? Do you see just that node in each Pi?
The PV system is at my son’s place (an hour drive away). It’s a big site – the panels & inverter are 100m from the house/supply meter. The site is covered by wi-fi. No RF is used – each RPi is serially direct connected to its emonTx via the GPIO pins. It would be difficult to set up a console/monitor. I currently connect with the site using Dataplicity. The USB spinning disk with the OS & data partitions has a capacity of 100+ Gb. And each instance has a different Node number
If I think back, a few weeks ago I had problems with an 9.9.5 version update on this particular instance. I’m beginning to think – it will net save me time if I just start over and burn a fresh image.
As an aside - you helped me with a recent Forum post and I now get a daily email reporting the feed update delay from each running instance. If it’s more than a couple of mins then I know there’s a problem and can get on the case.
Ah OK, that makes sense. How do you move the OS from an SD card image to the HDD? Are you sure the HDD is getting enough power? I had an issue with an HDD many years ago and that was the problem I identified. If out in the open, is it getting too hot/cold?