Hello community
I’d appreciate some help de-bugging simple project that’s giving me grief.
Problem: Headless Pi3b+ NAS running Manjaro ARM minimal (fully updated) crashes when I transfer multiple video files from desktop → NAS. Crash occurs between 10 and 15 minutes into transfers. Before this, remote SSH terminal becomes increasingly slow to respond. Behaviour is replicated across multiple file transfer protocols (Samba, SSHFS, NFS), but happens faster with Samba.
“Journalctl ¦ grep failure” post-crash shows this:
kernel: kworker/u8:3: page allocation failure: order:0, mode:0x40800(GFP_NOWAIT|__GFP_COMP), nodemask=(null),cpuset=/,mems_allowed=0
Context: NAS is a headless Pi3B+ and 3.5” 3TB SATA HDD (connected via a usb-SATA adapter with only the data lines connected – HDD power comes from a dedicated PSU). NAS is powered by a 100w PSU with separate 12v and 5v rails and sufficient current to power both Pi, HDD and fan cooler at peak load. NAS connects to router via 5ghz wifi and is situated close to the router (behaviour also happens on 2ghz and wired ethernet). OS is Manjaro ARM Minimal, fully updated, SD-card boot. Software firewall is disabled.
Assessment: I have ruled out power, router, HDD and thermal throttling as factors. I believe this may instead be a memory management issue. Here, alas, my competence ends.
Does anyone have any insight into this?
Many thanks