I’m within the means of writing software program to check Linux based mostly embedded gadgets after manufacturing. Every of the gadgets has a number of bodily community interfaces, all internally bridged to a Linux bridge br0.
Drawback:
- I want to seek out out if all the NICs of the DUT are working correctly (i.e. nothing went unsuitable throughout manufacturing).
The check setup is as follows:
- All Ethernet NICs of the DUT (MAC addresses are recognized, STP deactivated, IPv4/IPv6 activated) are bodily related to separate ports (as an instance 1-3) of a managed change
- The system operating the check software program (additionally Linux) is bodily related to the identical change (as an instance on port 4).
What I did to this point:
- I made positive precisely one of many change ports (1-3) was enabled (
ifAdminStatus=up) by way of SNMP after which ship merely pinged the DUT to see if the related NIC is working and repeated the method for all change ports/NICs. This, nevertheless, is considerably gradual because the change wants ~3.5 seconds to (de-)activate a change port and introduces extra delay till a ping is then profitable. This provides as much as ~15 seconds which I contemplate solution to lengthy.
Ideally, I would prefer to activate all change ports 1-3 concurrently and do the check in parallel on layer 2 or 3.
Query:
What can be we one of the best ways to design a easy and quick (!) check to seek out out if all NICs are working and simultanously avoiding all the inherent issues (broadcast storms as a consequence of switching loops, …)?