Megatest

                                   (delta (abs (- df disk-free))))
                              (if (and (> df 0)
                                       (> (/ delta df) 0.1)) ;; (> delta 200) ;; ignore changes under 200 Meg
                                  df
                                  #f)))
             (do-sync       (or new-cpu-load new-disk-free over-time))

	     ;;
	     ;; MOVE THIS TO A FILE FLAG BASED APPROACH (FOR NOW)
	     ;;
	     
             (test-info   (rmt:get-test-state-status-by-id run-id test-id))
             (state       (car test-info));; (db:test-get-state test-info))
             (status      (cdr test-info));; (db:test-get-status test-info))
	     (killreq     (equal? state "KILLREQ"))
             ;; (test-info   (rmt:get-test-state-status-by-id run-id test-id))
             ;; (state       (car test-info));; (db:test-get-state test-info))
             ;; (status      (cdr test-info));; (db:test-get-status test-info))
	     (killreq     (file-exists? (conc work-area"/kill-test"))) ;; (equal? state "KILLREQ"))
             (kill-reason  "no kill reason specified")
             (kill-job?    #f))
        ;; (common:telemetry-log "zombie" (conc "launch:monitor-job - decision time encountered at "(current-seconds)" with last-sync="last-sync" do-sync="do-sync" over-time="over-time" update-period="update-period))
        (cond
         (killreq
          (set! kill-reason "KILLING TEST since received kill request (KILLREQ)")
          (set! kill-job? #t))
         ((and runtlim (> (- (current-seconds) start-seconds) runtlim))
          (set! kill-reason (conc "KILLING TEST DUE TO TIME LIMIT EXCEEDED! Runtime=" (- (current-seconds) start-seconds) " seconds, limit=" runtlim))
          (set! kill-job? #t))
         ((equal? status "DEAD")
         #;((equal? status "DEAD") ;; NEED ALTERNATIVE MECHANISM FOR THIS.
          (tests:update-central-meta-info run-id test-id new-cpu-load new-disk-free (calc-minutes) #f #f)
          (rmt:set-state-status-and-roll-up-items run-id test-id 'foo "RUNNING" "n/a" "was marked dead; really still running.")
          ;;(set! kill-reason "KILLING TEST because it was marked as DEAD by launch:handle-zombie-tests (might indicate really overloaded server or else overzealous setup.deadtime)") ;; MARK RUNNING
          (set! kill-job? #f)))

        (debug:print 4 *default-log-port* "cpu: " new-cpu-load " disk: " new-disk-free " last-sync: " last-sync " do-sync: " do-sync)

	;; revisit logic in zombie handling.
	;;
        (if (common:low-noise-print 600 "run zombie") ;; every five minutes is plenty
	    (launch:handle-zombie-tests run-id))
	
        (when do-sync
          (tests:update-central-meta-info run-id test-id new-cpu-load new-disk-free (calc-minutes) #f #f))
          ;; (tests:update-central-meta-info run-id test-id new-cpu-load new-disk-free (calc-minutes) #f #f)

	  (let ((oup (open-output-file (conc work-area"/.run-logging-stats.csv") :append))
		(csv (conc run-id","test-id","new-cpu-load","new-disk-free","(calc-minutes))))
	    (debug:print 0 *default-log-port* "Updating run log, csv="csv)
	    (with-output-to-port oup
	      (lambda ()
		(print csv)))
	    (close-output-port oup)))
        
	(if kill-job? 
	    (begin
              (debug:print-info 0 *default-log-port* "proceeding to kill test: "kill-reason)
	      (mutex-lock! m)
	      ;; NOTE: The pid can change as different steps are run. Do we need handshaking between this
	      ;;       section and the runit section? Or add a loop that tries three times with a 1/4 second