I struggle with this. Part of the problem is that engineering (therefore the group of senior engineers) somehow always become the first-tier technical triage for problems. A customer reports a problem, a PM gathers more info, and then it's immediately thrown onto the queue of headaches that every single senior engineer is on the hook to fix. The ticket invariably gets round-robin assigned to a random senior engineer, who now has to drop everything and context switch over to debug mode.

Your oncall engineers should be doing the primary debugging. If they run into something they can’t debug, then they can call in the seniors.