Swap-Related Segmentation Faults/Core Dumps on TonidoPlug 2

[necopinus]

I'm having a rather frustraiting, intermittent problem on my TonidoPlug 2, which has been made all the more frustrating by my inability to gather enough information to effectively diagnose what's going on. I'm posting here in hopes that someone might be able to offer me some pointers as to how I might further debug this.

I am experiencing intermittent segfaults/core dumps in my running services. Sometimes the TP2 will run for weeks without a problems, and sometimes it will experience so many faults that it becomes completely unusable, unresponsive to anything but simple network probes. Logs report nothing useful outside of the fact that processes are dumping core, at least until Journald itself dies...

So far, all I've been able to do is determine that the problem is related to swap. The TP2 will run fine if I disable all of my swap space; however, enabling swap using either a swap partition or a swap file will (eventually) cause services to start to die. I've checked my hard drive for bad blocks, and everything seems clean. Moreover, it is not the case that processes start dying as soon as swap is used - rather they don't die before I start swapping.

About the only lead I have is that there seems to be a correlation between high load (compiling something from AUR, lots of database access, etc.) and this instability; perhaps I'm only having problems when the kernel tries to swap out application memory, but not buffers or cache? (I honestly don't know enough about Linux memory management to knwo if that makes sense...) Setting vm.swappiness = 0 also seems to resolve the issue, but when I do this I never see swap used at all, so I think this amounts to side-stepping, rather than solving, the problem.

The symptoms are also superficially similar to those reported at http://archlinuxarm.org/forum/viewtopic.php?f=28&t=3263 on the Cubox.

Does anyone have an idea of where I can go next to try to figure out what might be going on?