The non-Linux semaphore ticketholder implementation has two waiting paths: if an OperationContext is provided, we use waitForConditionOrInterrupt. If not, we use a regular condition variable wait() or wait_until().

Due to the problem described in SERVER-86269, this causes starvation for non-interruptible callers without an OperationContext. Specifically, the ThroughputProbingTicketholderMonitor get starved because it does not have an interruptible OpCtx.