i’ve had/have the same problem. I disabled my mjpeg camera and it seems to have stopped…
before i noticed a gradual incline of CPU usage during 1 - 2 days before a spontaneous restart of hass. I export CPU and MEM usage to influx and visualise it in grafana…
@pvizeli
it seems to me hass.io has a memory issue somewhere?
its the power connector, depending on what you have installed, it consumes more, you need the appropriate cable and for it, read the rPi specs and make the adjustment. I was having that same problem by using one of my old phone chargers, once I got a better usb cable and better amp plug, it never happened again.
Or do like me and put a portable charger in between, so you never loose power.
I am using the AC adapter that came with the Pi in one of those popular all-in-one kits from Amazon. Is it really possible that it isn’t getting enough power from that?
Sounds more like a memory leak and all available RAM gets used then chews into the little 100MB swap.
I had this issue on non-HassIO install while a dependency was installing in the background.
If you have access to the hypervisor sysstat might be installed you could check via sar -w to get your swapfile usage.
Is there any way to see logs after this happens? When I had the issue, I was not able to access my RPi via any methods and a power cycle was the only thing that fixed it. When it starts back up, all the logs reset.
I agree with one poster that said it could be your power supply/cable. In my case, it wasn’t the issue, but worth checking into.
I have been having the same symtoms described above and have documented them here: Hass.io Access Crashing Overnight, Every Night
Any of you have any luck addressing this issue or coming up with a way to check the logs after the necesarry power cycle?
For anyone else that stumbles upon this thread with similar problems, read the solution that worked for me in @anndrewww’s thread.
I have never used WiFi and had the issue a few times.
First time, I had just started using HTTP OwnTracks. Disabled it and everything worked again.
Second time was recently when I enabled the updated Ring component. Disabled it and everything is rock solid since.
Possible I have a bad Ethernet port I guess, and the updated components increased network traffic, but I can’t prove it either way at the moment.
I had mine go offline two nights ago for some reason. Again, no way to figure out what the culprit is. Worst part is that when I power cycled since I could not SSH in, when it came up, my alarm went off. I’ve since added some safeguards so that a non-ready z wave network will not trigger it, but really pissed everyone off at 5am.
I do. I use it for the device tracker.
Could be a coincidence, but the only two (2) crashes since my re-install happend at or near when I left my “Home” Zone with all my devices. I had only one device tracker method enabled, the Bluetooth device tracker, so I could try to eliminate some variables without multiple methods (TP-Link Device Tracker; Owntracks; etc).
EDIT:
The other oddity that I have noticed is that the file size of my Snapshots balloons to 70-80MB with the “Dasshio” add-on segment making up nearly 70MB, which makes no sense.
Curious if it is. I do not have OwnTracks enabled anymore, but do use AsusWRT, GPS Logger, and Bluetooth.
Its very random when it happens so even if I disabled Bluetooth, it would be a good solid week of up time before I could say it might be the culprit. Plus I’m a tinkerer so I end up restarting Home Assistant quite a bit.
I’m having the same problem, or similar. I can still ping my Hassio and access the samba share, but I can’t access the web interface. It’s not consistant on how often it happens, could be an hour to 12 hours. I don’t have much connected to it. Just 2 TP-Link switches, a hikvision camera (with binary-sensors), a ring camera (with binary-sensors), rf-switches (but the transmitter is disconnected), and a gpio sensor (also disconnected).
I’m going to google how to check the cpu usage and the memory to see if I can find any clues there.
You think it could be because I have identified devices that are not connected?
I seem to be having a similar issue.
On a whim I also blew a bunch of cold air across both sides of the pi, it did not come back up.
Anyone have a way to 1) see some type of persistent log, 2) see cpu/mem use ? Curious whether this is only on Rpi3 or other hardware as well.
As neat as Hass.io is I feel we’re lacking some of these debugging options that are typically useful (or they’re hard to find). I like that maintenance and configuration is simpler (so I’m told, just got on to this in the past month), but I’m starting to wonder if I wouldn’t rather trade that for reliability.
I had the same issue and spent a few weeks troubleshooting it.
Your RPi wi-fi chip is busted. Switch to ethernet connection and Hassio will behave normally.
I’m doing an informal poll related to this issue. Did you purchase your RPi as part of a Vilros kit?
Have never used Wifi on RPi and still had the issue. I got rid of a bunch of services and things I didn’t need and the problem disappeared. I think its related to certain services, but can’t prove that.
Neither of you two got further with logging/mem usage? Did nothing come of the memory issue comments above?
I have nmap_tracker and zwave set up, it was happening with just those.
Recently moved and haven’t had a chance to run ethernet, this will be the longer-term goal. For now I’m just getting started and feeling the waters but would very much like it to be working reasonably reliably.
At this point I’ve got half a mind to set up a Pi2 with nmap_tracker and point it at this one to see how long it’s up for each time. I might also stick a scope on the power pin to see if there are still local brownouts, I know my supply has plenty of current capacity, but the line regulation might not be great for fast switching loads, worth a shot?
Unfortunately it seems they left U19 (which is the chip that appears to be connected to the antenna) off the schematics, so we won’t be getting many clues from there. I did see some Rpi forums talk about catching catching an undervoltage condition and displaying it to the screen, might be worth seeing if it’s something that can be incorporated into the logs?
I got it via Digi-key.ca, and use a separate power supply.
Hello, I have the exact same issue
Basically, raspberry doesn’t respond anymore (front, ssh, ping), leds are on, z-wave sensor keeps running in background…
I also have an official power supply with enough A and use wifi
Did you solve the issue or have a better understanding of it ?
Thanks in advance
Seems to be a software or hardware issue with the wi-fi onboard the Pi. Since switching to an Ethernet connection, I have not had the issues.