Hello.
The part with "No space left" seems quite clear to me. The emergency
with cache file being deleted and re-created should never happen when GC
works, though. Deleted files may get held a while by other kresd and
kres-cache-gc, and since recent versions we fully pre-allocate the new
cache. Increasing your tmpfs size would probably avoid the
out-of-space, but I'm way more concerned about the GC not working well
and those SIGSEGVs (80% is way too low to cause problems).
Can you confirm that you use a recent knot-resolver version? And
perhaps, where are your packages from and 32/64-bit?
Additionally, a coredump from any of those crashes would be nice, but
maybe we can find something even without it. (easiest via coredumpctl in
your case, I suppose)
--Vladimir