We discoverd a bug in performance_timing.c which causes servald to crash in our 100 nodes hub scenario. The config enabled debug.timing=true.
Servald crashed on two nodes, which produced almost the same FATAL report.
qemu7-serval.txt
qemu10-serval.txt