return cmpBool(!p1Found, !p2Found)
		}

		p1Usage, p1Err := podDiskUsage(p1Stats, p1, fsStatsToMeasure)
		p2Usage, p2Err := podDiskUsage(p2Stats, p2, fsStatsToMeasure)
		if p1Err != nil || p2Err != nil {
			// prioritize evicting the pod which had an error getting stats
			return cmpBool(p1Err != nil, p2Err != nil)
		}

		p1Disk := p1Usage[diskResource]
		p2Disk := p2Usage[diskResource]

		// promptly ejected from its queue.  The goroutine doing the APF handling writes an HTTP
		// response message with status 429.
		// The timeout handler invokes its inner handler in one goroutine while reacting to the
		// passage of time in its original goroutine.  That reaction to a time out consists of either
		// (a) writing an HTTP response message with status 504 to indicate the timeout or (b) doing an