Topology dependent waiting times

[kleinreact]

Clock control simulation fails regularly during the nighty CI.

Investigating the simulated topologies reveals that the values for the reframing waiting times are not appropriate. The problem is that choosing good values for these waiting times depends on the topology. Mostly the hourglass and linear topologies fail to re-center, which is caused by the reframing waiting times being too short.

Especially for the linear topologies the propagation of clock synchronization between both ends takes long causing the requirement for longer reframing waiting times. We observe that the longer the linear topology, the larger must be the reframing waiting time in general.

It has to be investigated how to better determine reframing waiting times depending on the topology to avoid the issue.

[lmbollen]

When you say "waiting" times, do you mean reframing waiting times or waiting times for stability detection?

[kleinreact]

I mean reframing waiting times. I'll make it more precise in the description.