@Richard Burts wrote:
If you are changing the configuration of the stack (adding a switch, and I assume also if you remove a switch from the stack) then I assume the remaining stack members have to logically restart to implement the change. So I am not surprised that there is some impact on connectivity.
Although my stack experience is (very much now) dated, I recall (?), at least on a 3750 stack, adding a "new" stack member (without any active ports), was hitless. However, dealing with a failed stacked member, and replacing it, is often a quite different in its impact because the original stack member, even if not the stack master, may have active edge port connections. Those going off-line or coming on-line, may cause a network hit. Consider, if rather than a port-channel for redundancy, we have multiple stack member connections providing a L2 loop, managed by STP. Then, as @Cristian Matei initially suggested, we might have a STP convergence interruption.
BTW, just did a quick look at the series, which touts High Availability as a feature; including its HSR (High-availability Seamless Redundancy). "Traditional" stacking appears to be a feature too, see Stacking and High Availability Configuration Guide, Cisco Catalyst IE9300 Rugged Series Switches (ha, even the persistent stack MAC feature is provided).
Unfortunately, I have no hands-on experience with IE switches, which, do seem to be a bit different from Cisco Catalyst switches.
Again, whether your described ping loss is "normal" and whether it might be mitigated, are unknown answers to me.
