I’ve had some issues recently with my devices failing with a “zram: Decompression failed!” error in the dmesg logs.
Sometimes the unit runs for a while before this issue occurs, sometimes it can happen quite quickly. I’m running 4 containers on a BalenaFin v1.1.1, including InfluxDB and Telegraf to collect system metrics to check on the health of my devices (ironically to monitor the memory usage of my device to catch/prevent OOM errors with InfluxDB). The other 2 containers are a GUI and a python based data uploader (small packets of data in the 10’s of kB), which don’t consume much memory (10-20MB each reported by Balena)
As far as I can tell, I’m not maxing out the memory, unless something big is happening between logs (every 15seconds).
- BalenaFin v1.1.1
- 4 Containers + Supervisor (And occasional health-check container the supervisor runs)
- OS version: balenaOS 2.80.3+rev1
- Supervisor version: 12.7.0
- Typical memory used around 60-65% when issue occurs
- Peak memory used has been as high as 75%
Below is a screen grab of the system metrics with a data gap in the middle (where the white markers are) where the unit stopped working, and then restarted a little later (although in this instance it didn’t seem to start 2 of the containers properly for some reason & I had to restart them manually)
Anyone run into this before?
I can provide some more dmesg output if needed.