Container eventually slows down and becomes unresponsive

Product support balenaOS
1

I am running a balenaOS image on a Variscite DART-MX8M Mini. Eventually the python software running within the container slows down. Wait even longer and the docker goes completely unresponsive. I originally thought that this was an issue with the python code however I’m now thinking that the resin_supervisor container memory usage is looking suspect.

Here is what the docker stats look like if left running for a couple hours. Its worth noting that if I CTRL-C out of the docker stats and renter the mem usage stays the same but the randomly created containers are removed.

After a power cycle:
*MULTI IMAGE NOT ALLOWED BUT IT LOOKS LIKE THE FOLLOWING
CONTAINER ID NAME CPU % MEM USAGE / LIMIT MEM % NET I/O BLOCK I/O PIDS
4a24cc7ec292 main_1_1 15.46% 138MiB / 1.952GiB 6.79% 0B / 0B 9.07MB / 180kB 31
028bb97565eb resin_supervisor 3.84% 63.11MiB / 1.952GiB 3.16% 0B / 0B 36.5MB / 172kB 11

30m later

*MULTI IMAGE NOT ALLOWED BUT THE resin_superviosr CPU up to 83.37MiB

Another 15m later

*MULTI IMAGE NOT ALLOWED BUT THE resin_superviosr CPU up to 91.49MiB

Another 15m later

*MULTI IMAGE NOT ALLOWED BUT THE resin_superviosr CPU up to 102.9MiB

Another 15m later

*MULTI IMAGE NOT ALLOWED BUT THE resin_superviosr CPU up to 116.7MiB

After 45minutes

image
image1920×837 104 KB

2

After about a day the system is unresponsive. Here is what balena stats look like then:

image
image1407×121 26.7 KB

3

Another update. Here is what the system looks like after 24h. As you can see the memory of the supervisor has been steadily growing.

image
image1249×102 17.3 KB

4

Bumping this in hopes to get a reply.

Thanks!