Megatest

				   cmd-in
				   (string->symbol cmd-in)))
            (params            (vector-ref dat 1))
            (start-t           (current-milliseconds))
            (readonly-mode     (dbr:dbstruct-read-only dbstruct))
            (readonly-command  (member cmd api:read-only-queries))
            (writecmd-in-readonly-mode (and readonly-mode (not readonly-command)))
            #;(foo               (begin
                                 (common:telemetry-log (conc "api-in:"(->string cmd))
                                                       payload: `((params . ,params)))
                                 
                                 #t))
            (res    
             (if writecmd-in-readonly-mode
                 (conc "attempt to run write command "cmd" on a read-only database")

       ;; save all stats
       (let ((delta-t (- (current-milliseconds)
			 start-t)))
	 (hash-table-set! *db-api-call-time* cmd
			  (cons delta-t (hash-table-ref/default *db-api-call-time* cmd '()))))
       (if writecmd-in-readonly-mode
           (begin
             #;(common:telemetry-log (conc "api-out:"(->string cmd))
                                   payload: `((params . ,params)
                                              (ok-res . #t)))
	     (vector #f res))
           (begin
             #;(common:telemetry-log (conc "api-out:"(->string cmd))
                                   payload: `((params . ,params)
                                              (ok-res . #f)))
             (vector #t res))))))))

;; http-server  send-response
;;                 api:process-request
;;                    db:*

           (handle-exceptions
           exn
           #t ;; just ignore it, it might have died in the meantime so joining it will throw an exception
           (thread-join! thread))
           )))
   (hash-table-keys *common:thread-punchlist*)))

;; (define *common:telemetry-log-state* 'startup)
;; (define *common:telemetry-log-socket* #f)
;; 
;; (define (common:telemetry-log-open)
;;   (if (eq? *common:telemetry-log-state* 'startup)
;;       (let* ((serverhost (configf:lookup *configdat* "telemetry" "host"))
;;              (serverport (configf:lookup-number *configdat* "telemetry" "port"))
;;              (user (or (get-environment-variable "USER") "unknown"))
;;              (host (or (get-environment-variable "HOST") "unknown")))
;;         (set! *common:telemetry-log-state*
;;               (handle-exceptions
;;                exn
;;                (begin
;;                  (debug:print-info 0 *default-log-port* "common-telemetry-log open udp port failure")
;;                  'broken)
;;                (if (and serverhost serverport user host)
;;                    (let* ((s (udp-open-socket)))
;;                      ;;(udp-bind! s #f 0)
;;                      (udp-connect! s serverhost serverport)
;;                      (set! *common:telemetry-log-socket* s)
;;                      'open)
;;                    'not-needed))))))
;;   
;; (define (common:telemetry-log event #!key (payload '()))
;;   (if (eq? *common:telemetry-log-state* 'startup)
;;       (common:telemetry-log-open))
;; 
;;   (if (eq? 'open *common:telemetry-log-state*)
;;       (handle-exceptions
;;        exn
;;        (begin
;;          (debug:print-info 0 *default-log-port* "common-telemetry-log comms failure ; disabled (no server?)")
;;          ;;(define *common:telemetry-log-state* 'broken-or-no-server-preclose)
;;          ;;(common:telemetry-log-close)
;;          (define *common:telemetry-log-state* 'broken-or-no-server)
;;          (set! *common:telemetry-log-socket* #f)

;;          )
;;        (if (and *common:telemetry-log-socket* event) ;; TODO - filter on event against telemetry.want-events
;;            (let* ((user (or (get-environment-variable "USER") "unknown"))
;;                   (host (or (get-environment-variable "HOST") "unknown"))
;;                   (start (conc "[megatest "event"]"))
;;                   (toppath (or *toppath* "/dev/null"))
;;                   (payload-serialized
;;                    (base64:base64-encode
;;                     (z3:encode-buffer
;;                      (with-output-to-string (lambda () (pp payload))))))
;;                   (msg     (conc user":"host":"start":"(current-process-id)":"(car (argv))":"
;;                                  toppath":"payload-serialized)))
;;              (udp-send *common:telemetry-log-socket* msg))))))
;;   
;; (define (common:telemetry-log-close)
;;   (when (or (member *common:telemetry-log-state* '(broken-or-no-server-preclose open)) *common:telemetry-log-socket*)
;;     (handle-exceptions
;;      exn
;;      (begin
;;        (define *common:telemetry-log-state* 'closed-fail)
;;        (debug:print-info 0 *default-log-port* "common-telemetry-log closure failure")

;;        )
;;      (begin
;;        (define *common:telemetry-log-state* 'closed)
;;        (udp-close-socket *common:telemetry-log-socket*)
;;        (set! *common:telemetry-log-socket* #f)))))

    ;; (tests:set-full-meta-info test-id run-id (calc-minutes) work-area)
    (tests:set-full-meta-info #f test-id run-id (calc-minutes) work-area 10)

    (let loop ((minutes   (calc-minutes))
	       (cpu-load  (alist-ref 'adj-core-load (common:get-normalized-cpu-load #f)))
	       (disk-free (get-df (current-directory)))
               (last-sync (current-seconds)))
      #;(common:telemetry-log "zombie" (conc "launch:monitor-job - top of loop encountered at "(current-seconds)" with last-sync="last-sync))
      (let* ((over-time     (> (current-seconds) (+ last-sync update-period)))
             (new-cpu-load  (let* ((load  (alist-ref 'adj-core-load (common:get-normalized-cpu-load #f)))
                                   (delta (abs (- load cpu-load))))
                              (if (> delta 0.1) ;; don't bother updating with small changes
                                  load
                                  #f)))
             (new-disk-free (let* ((df    (if over-time ;; only get df every 30 seconds

             (do-sync       (or new-cpu-load new-disk-free over-time))

             (test-info   (rmt:get-test-info-by-id run-id test-id))
             (state       (db:test-get-state test-info))
             (status      (db:test-get-status test-info))
             (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
         ((test-get-kill-request run-id test-id)
          (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")
          (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)
        (launch:handle-zombie-tests run-id)
        (when do-sync
          ;;(with-output-to-file (conc (getenv "MT_TEST_RUN_DIR") "/last-loadinfo.log" #:append)
          ;;  (lambda () (pp (list (current-seconds) new-cpu-load new-disk-free (calc-minutes)))))
          #;(common:telemetry-log "zombie" (conc  "launch:monitor-job - dosync started at "(current-seconds)))
          (tests:update-central-meta-info run-id test-id new-cpu-load new-disk-free (calc-minutes) #f #f)
          #;(common:telemetry-log "zombie" (conc "launch:monitor-job - dosync finished at "(current-seconds))))
        
	(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

		  (client:setup areapath)
		  #f))))

(define *send-receive-mutex* (make-mutex)) ;; should have separate mutex per run-id

(define (rmt:send-receive cmd rid params #!key (attemptnum 1)(area-dat #f)) ;; start attemptnum at 1 so the modulo below works as expected

  #;(common:telemetry-log (conc "rmt:"(->string cmd))
                        payload: `((rid . ,rid)
                                   (params . ,params)))
                          
  ;; do all the prep locked under the rmt-mutex
  (mutex-lock! *rmt-mutex*)
  
  ;; 1. check if server is started IFF cmd is a write OR if we are not on the homehost, store in runremote

	  	  (debug:print-info 0 *default-log-port* "filtering initial test list with tagexpr: " (args:get-arg "-tagexpr") " => " allowed-tests)
		  ));; tests will be ANDed with this list

    ;; register this run in monitor.db
    (rmt:tasks-add "run-tests" user target runname test-patts task-key) ;; params)
    (rmt:tasks-set-state-given-param-key task-key "running")

    #;(common:telemetry-log "run-tests"
                          payload:
                          `( (target . ,target)
                             (run-name . ,runname)
                             (test-patts . ,test-patts) ) )

    
    ;; Now generate all the tests lists