Megatest

Diff
Login

Differences From Artifact [1ecd29c566]:

To Artifact [a93ed3a08e]:


369
370
371
372
373
374
375


376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391

392
393
394

395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410

411
412

413
414
415
416








417
418
419
420
421
422
423
424
425
			  (debug:print-info 0 *default-log-port* "Waiting for server alive signature")
                          (mutex-lock! *heartbeat-mutex*)
                          (set! sdat *server-info*)
                          (mutex-unlock! *heartbeat-mutex*)
                          (if (and sdat
				   (not changed)
				   (> (- (current-seconds) start-time) 2))


			      sdat
                              (begin
				(debug:print-info 0 *default-log-port* "Still waiting, last-sdat=" last-sdat)
                                (sleep 4)
				(if (> (- (current-seconds) start-time) 120) ;; been waiting for two minutes
				    (begin
				      (debug:print-error 0 *default-log-port* "transport appears to have died, exiting server " server-id " for run " run-id)
				      (tasks:server-delete-record (db:delay-if-busy tdbdat) server-id "failed to start, never received server alive signature")
				      (exit))
				    (loop start-time
					  (equal? sdat last-sdat)
					  sdat)))))))
         (iface       (car server-info))
         (port        (cadr server-info))
         (last-access 0)
	 (server-timeout (server:get-timeout)))

    (let loop ((count         0)
	       (server-state 'available)
	       (bad-sync-count 0))


      ;; Use this opportunity to sync the tmp db to megatest.db
      (if *dbstruct-db* 
 	  (let ((start-time (current-milliseconds))
 		(sync-time  #f)
 		(rem-time   #f))
	    (thread-sleep! 4))
	  ;; Removed code is pasted below (keeping it around until we are clear it is not needed).
	  ;; no *dbstruct-db* yet, set running after our first pass through and start the db
	  (if (eq? server-state 'available)
	      (let ((new-server-id (tasks:server-am-i-the-server? (db:delay-if-busy tdbdat) run-id))) ;; try to ensure no double registering of servers
		(if (equal? new-server-id server-id)
		    (begin
		      (tasks:server-set-state! (db:delay-if-busy tdbdat) server-id "dbprep")
		      (thread-sleep! 0.5) ;; give some margin for queries to complete before switching from file based access to server based access
		      (set! *dbstruct-db*  (db:setup)) ;;  run-id))

		      (tasks:server-set-state! (db:delay-if-busy tdbdat) server-id "running")
		      (server:write-dotserver *toppath* (conc iface ":" port)))

		    (begin ;; gotta exit nicely
		      (tasks:server-set-state! (db:delay-if-busy tdbdat) server-id "collision")
		      (http-transport:server-shutdown server-id port))))))
      








      (if (< count 1) ;; 3x3 = 9 secs aprox
	  (loop (+ count 1) 'running bad-sync-count))
      
      ;; Check that iface and port have not changed (can happen if server port collides)
      (mutex-lock! *heartbeat-mutex*)
      (set! sdat *server-info*)
      (mutex-unlock! *heartbeat-mutex*)
      
      (if (or (not (equal? sdat (list iface port)))







>
>
|














|
>


|
>


|
<
<
<
<
|
|
|
|
|
|
|
|
|
>
|
|
>
|
|
|
|
>
>
>
>
>
>
>
>

|







369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401




402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
			  (debug:print-info 0 *default-log-port* "Waiting for server alive signature")
                          (mutex-lock! *heartbeat-mutex*)
                          (set! sdat *server-info*)
                          (mutex-unlock! *heartbeat-mutex*)
                          (if (and sdat
				   (not changed)
				   (> (- (current-seconds) start-time) 2))
			      (begin
				(debug:print-info 0 *default-log-port* "Received server alive signature")
				sdat)
                              (begin
				(debug:print-info 0 *default-log-port* "Still waiting, last-sdat=" last-sdat)
                                (sleep 4)
				(if (> (- (current-seconds) start-time) 120) ;; been waiting for two minutes
				    (begin
				      (debug:print-error 0 *default-log-port* "transport appears to have died, exiting server " server-id " for run " run-id)
				      (tasks:server-delete-record (db:delay-if-busy tdbdat) server-id "failed to start, never received server alive signature")
				      (exit))
				    (loop start-time
					  (equal? sdat last-sdat)
					  sdat)))))))
         (iface       (car server-info))
         (port        (cadr server-info))
         (last-access 0)
	 (server-timeout (server:get-timeout))
	 (server-going  #f))
    (let loop ((count         0)
	       (server-state 'available)
	       (bad-sync-count 0)
	       (start-time     (current-milliseconds)))

      ;; Use this opportunity to sync the tmp db to megatest.db
      (if (not server-going) ;; *dbstruct-db* 




	    ;; Removed code is pasted below (keeping it around until we are clear it is not needed).
	    ;; no *dbstruct-db* yet, set running after our first pass through and start the db
	    (if (eq? server-state 'available)
		(let ((new-server-id (tasks:server-am-i-the-server? (db:delay-if-busy tdbdat) run-id))) ;; try to ensure no double registering of servers
		  (if (equal? new-server-id server-id)
		      (begin
			(tasks:server-set-state! (db:delay-if-busy tdbdat) server-id "dbprep")
			(thread-sleep! 0.5) ;; give some margin for queries to complete before switching from file based access to server based access
			(set! *dbstruct-db*  (db:setup)) ;;  run-id))
			(set! server-going #t)
			(tasks:server-set-state! (db:delay-if-busy tdbdat) server-id "running")
			(server:write-dotserver *toppath* (conc iface ":" port))
			(delete-file* (conc *toppath* "/.starting-server")))
		      (begin ;; gotta exit nicely
			(tasks:server-set-state! (db:delay-if-busy tdbdat) server-id "collision")
			(http-transport:server-shutdown server-id port))))))

      ;; when things go wrong we don't want to be doing the various queries too often
      ;; so we strive to run this stuff only every four seconds or so.
      (let* ((sync-time (- (current-milliseconds) start-time))
	    (rem-time  (quotient (- 4000 sync-time) 1000)))
	(if (and (<= rem-time 4)
		 (>  rem-time 0))
	    (thread-sleep! rem-time)))
      
      (if (< count 1) ;; 3x3 = 9 secs aprox
	  (loop (+ count 1) 'running bad-sync-count (current-milliseconds)))
      
      ;; Check that iface and port have not changed (can happen if server port collides)
      (mutex-lock! *heartbeat-mutex*)
      (set! sdat *server-info*)
      (mutex-unlock! *heartbeat-mutex*)
      
      (if (or (not (equal? sdat (list iface port)))
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
	      ;;
	      ;; Consider implementing some smarts here to re-insert the record or kill self is
	      ;; the db indicates so
	      ;;
	      ;; (if (tasks:server-am-i-the-server? tdb run-id)
	      ;;     (tasks:server-set-state! tdb server-id "running"))
	      ;;
	      (loop 0 server-state bad-sync-count))
	    (http-transport:server-shutdown server-id port))))))

;; code cut out from above
;;
;; (condition-case
;;  ;; (if (and (member (mutex-state *db-sync-mutex*) '(abandoned not-abandoned))
;;  ;;	      (> (- (current-seconds) *db-last-sync*) 5)) ;; if not currently being synced nor recently synced







|







465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
	      ;;
	      ;; Consider implementing some smarts here to re-insert the record or kill self is
	      ;; the db indicates so
	      ;;
	      ;; (if (tasks:server-am-i-the-server? tdb run-id)
	      ;;     (tasks:server-set-state! tdb server-id "running"))
	      ;;
	      (loop 0 server-state bad-sync-count (current-milliseconds)))
	    (http-transport:server-shutdown server-id port))))))

;; code cut out from above
;;
;; (condition-case
;;  ;; (if (and (member (mutex-state *db-sync-mutex*) '(abandoned not-abandoned))
;;  ;;	      (> (- (current-seconds) *db-last-sync*) 5)) ;; if not currently being synced nor recently synced
521
522
523
524
525
526
527




528
529
530
531
532
533
534
535
536
537
538
539
540


541
542
543
544
545
546
547
548
549
550
551
552

553
554
555
556
557
558
559
    (exit)))

;; all routes though here end in exit ...
;;
;; start_server? 
;;
(define (http-transport:launch run-id)




  (let* ((tdbdat (tasks:open-db)))
    (set! *run-id*   run-id)
    (if (args:get-arg "-daemonize")
	(begin
	  (daemon:ize)
	  (if *alt-log-file* ;; we should re-connect to this port, I think daemon:ize disrupts it
	      (begin
		(current-error-port *alt-log-file*)
		(current-output-port *alt-log-file*)))))
    (if (server:check-if-running run-id)
	(begin
	  (debug:print 0 *default-log-port* "INFO: Server for run-id " run-id " already running")
	  (exit 0)))


    (let loop ((server-id (tasks:server-lock-slot (db:delay-if-busy tdbdat) run-id))
	       (remtries  4))
      (if (not server-id)
	  (if (> remtries 0)
	      (begin
		(thread-sleep! 2)
		(loop (tasks:server-lock-slot (db:delay-if-busy tdbdat) run-id)
		      (- remtries 1)))
	      (begin
		;; since we didn't get the server lock we are going to clean up and bail out
		(debug:print-info 2 *default-log-port* "INFO: server pid=" (current-process-id) ", hostname=" (get-host-name) " not starting due to other candidates ahead in start queue")
		(tasks:server-delete-records-for-this-pid (db:delay-if-busy tdbdat) " http-transport:launch")

		))
	  (let* ((th2 (make-thread (lambda ()
				     (debug:print-info 0 *default-log-port* "Server run thread started")
				     (http-transport:run 
				      (if (args:get-arg "-server")
					  (args:get-arg "-server")
					  "-")







>
>
>
>












|
>
>












>







531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
    (exit)))

;; all routes though here end in exit ...
;;
;; start_server? 
;;
(define (http-transport:launch run-id)
  (with-output-to-file
      (conc *toppath* "/.starting-server")
    (lambda ()
      (print (current-process-id) " on " (get-host-name))))
  (let* ((tdbdat (tasks:open-db)))
    (set! *run-id*   run-id)
    (if (args:get-arg "-daemonize")
	(begin
	  (daemon:ize)
	  (if *alt-log-file* ;; we should re-connect to this port, I think daemon:ize disrupts it
	      (begin
		(current-error-port *alt-log-file*)
		(current-output-port *alt-log-file*)))))
    (if (server:check-if-running run-id)
	(begin
	  (debug:print 0 *default-log-port* "INFO: Server for run-id " run-id " already running")
	  (exit 0))
	(begin ;; ok, no server detected, clean out any lingering records
	   (tasks:server-force-clean-running-records-for-run-id  (db:delay-if-busy tdbdat) run-id "notresponding")))
    (let loop ((server-id (tasks:server-lock-slot (db:delay-if-busy tdbdat) run-id))
	       (remtries  4))
      (if (not server-id)
	  (if (> remtries 0)
	      (begin
		(thread-sleep! 2)
		(loop (tasks:server-lock-slot (db:delay-if-busy tdbdat) run-id)
		      (- remtries 1)))
	      (begin
		;; since we didn't get the server lock we are going to clean up and bail out
		(debug:print-info 2 *default-log-port* "INFO: server pid=" (current-process-id) ", hostname=" (get-host-name) " not starting due to other candidates ahead in start queue")
		(tasks:server-delete-records-for-this-pid (db:delay-if-busy tdbdat) " http-transport:launch")
		(delete-file* (conc *toppath* "/.starting-server"))
		))
	  (let* ((th2 (make-thread (lambda ()
				     (debug:print-info 0 *default-log-port* "Server run thread started")
				     (http-transport:run 
				      (if (args:get-arg "-server")
					  (args:get-arg "-server")
					  "-")