Memory test and study CPU and verify memory configuration.
DIMM config impacts DMA bandwidth manual is not always 
clear so experiments might be needed. 

Test performance from different and between bus-slots,.
Check routing performance corresponds to bus-type/width speed.


Check so system interrupts can give interrupt vectors. Modern boards should
use MSI-X see /proc/interrupts see thats system can supply enough for
vectors for the maximum number of NIC's in the config


Check BIOS and verify that system can be set so it boots after power-loss.
There use to be different settings.

Stress test for at least 24-hours with high/"rDoS"-like traffic load


Monitor temperatures when system is "fully" configured and with rDoS
like traffic load. Interfaces have very different power-specs

Check system clock-options performs decent. Timestamp is used in the 
kernel "fastpath" in some conditions. pktgen is a heavy user of
do_gettimeofday() and can be used to test check that performance is 
what expected. /sys/devices/system/clocksource/clocksource0


Check USB probing/boot try some different USB-brands

If redundant power-supply verify that system runs on single power.
also verify that power supply is hot-swappable.

Redirect console to serial port (Needed to track kernel oopses) 
normally in bootloader config (grub).

Verify that interface are mounted firmly in the bus slots. Especially 
for interfaces with pluggable modules .


After lab testing deployment in less critical missions...