Megatest

Diff
Login

Differences From Artifact [db6150ddbe]:

To Artifact [5cf026b1f8]:


406
407
408
409
410
411
412
















































































































































413
414
415
416
417
418
419
				  (cons hed reruns))))))
     (else
      (debug:print 4 "ERROR: No handler for this condition.")
      ;; TRY (thread-sleep! (+ 1 *global-delta*))
      (set! loop-list (list (car newtal)(cdr newtal) reg reruns)))) ;; END OF IF CAN RUN MORE
    loop-list))


















































































































































;; test-records is a hash table testname:item_path => vector < testname testconfig waitons priority items-info ... >
(define (runs:run-tests-queue run-id runname test-records keyvals flags test-patts required-tests reglen-in)
  ;; At this point the list of parent tests is expanded 
  ;; NB// Should expand items here and then insert into the run queue.
  (debug:print 5 "test-records: " test-records ", flags: " (hash-table->alist flags))
  (let ((run-info              (cdb:remote-run db:get-run-info #f run-id))







>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>







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
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
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
560
561
562
563
				  (cons hed reruns))))))
     (else
      (debug:print 4 "ERROR: No handler for this condition.")
      ;; TRY (thread-sleep! (+ 1 *global-delta*))
      (set! loop-list (list (car newtal)(cdr newtal) reg reruns)))) ;; END OF IF CAN RUN MORE
    loop-list))

(define (runs:process-expanded-tests hed tal reg reruns reglen regfull test-record runname test-name item-path jobgroup max-concurrent-jobs run-id waitons item-path testmode)
  (let* ((run-limits-info         (runs:can-run-more-tests jobgroup max-concurrent-jobs)) ;; look at the test jobgroup and tot jobs running
	 ;; (open-run-close runs:can-run-more-tests #f jobgroup max-concurrent-jobs)) ;; look at the test jobgroup and tot jobs running
	 (have-resources          (car run-limits-info))
	 (num-running             (list-ref run-limits-info 1))
	 (num-running-in-jobgroup (list-ref run-limits-info 2))
	 (max-concurrent-jobs     (list-ref run-limits-info 3))
	 (job-group-limit         (list-ref run-limits-info 4))
	 (prereqs-not-met         (mt:get-prereqs-not-met run-id waitons item-path mode: testmode))
	 (fails                   (runs:calc-fails prereqs-not-met))
	 (non-completed           (runs:calc-not-completed prereqs-not-met))
	 (loop-list               (list hed tal reg reruns)))
    (debug:print-info 4 "have-resources: " have-resources " prereqs-not-met: (" 
		      (string-intersperse 
		       (map (lambda (t)
			      (if (vector? t)
				  (conc (db:test-get-state t) "/" (db:test-get-status t))
				  (conc " WARNING: t is not a vector=" t )))
			    prereqs-not-met) ", ") ") fails: " fails)
    (debug:print-info 4 "hed=" hed "\n  test-record=" test-record "\n  test-name: " test-name "\n  item-path: " item-path "\n  test-patts: " test-patts)
    
    ;; Don't know at this time if the test have been launched at some time in the past
    ;; i.e. is this a re-launch?
    (debug:print-info 4 "run-limits-info = " run-limits-info)
    
    (cond ;; INNER COND #1 for a launchable test
     
     ;; Check item path against item-patts, 
     ;;
     ((not (tests:match test-patts (tests:testqueue-get-testname test-record) item-path required: required-tests)) ;; This test/itempath is not to be run
      ;; else the run is stuck, temporarily or permanently
      ;; but should check if it is due to lack of resources vs. prerequisites
      (debug:print-info 1 "Skipping " (tests:testqueue-get-testname test-record) " " item-path " as it doesn't match " test-patts)
      ;; (thread-sleep! *global-delta*)
      (if (or (not (null? tal))(not (null? reg)))
	  (set! loop-list
		(list (runs:queue-next-hed tal reg reglen regfull)
		      (runs:queue-next-tal tal reg reglen regfull)
		      (runs:queue-next-reg tal reg reglen regfull)
		      reruns))))
     
     ;; Register tests 
     ;;
     ((not (hash-table-ref/default test-registry (runs:make-full-test-name test-name item-path) #f))
      (debug:print-info 4 "Pre-registering test " test-name "/" item-path " to create placeholder" )
      (let ((th (make-thread (lambda ()
			       (mutex-lock! registry-mutex)
			       (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'start)
			       (mutex-unlock! registry-mutex)
			       ;; If haven't done it before register a top level test if this is an itemized test
			       (if (not (eq? (hash-table-ref/default test-registry (runs:make-full-test-name test-name "") #f) 'done))
				   (cdb:tests-register-test *runremote* run-id test-name ""))
			       (cdb:tests-register-test *runremote* run-id test-name item-path)
			       (mutex-lock! registry-mutex)
			       (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'done)
			       (mutex-unlock! registry-mutex))
			     (conc test-name "/" item-path))))
	(thread-start! th))
      (runs:shrink-can-run-more-tests-count)   ;; DELAY TWEAKER (still needed?)
      (if (and (null? tal)(null? reg))
	  (set! loop-list (list hed tal (append reg (list hed)) reruns))
	  (set! loop-list (list (runs:queue-next-hed tal reg reglen regfull)
				(runs:queue-next-tal tal reg reglen regfull)
				;; NB// Here we are building reg as we register tests
				;; if regfull we must pop the front item off reg
				(if regfull
				    (append (cdr reg) (list hed))
				    (append reg (list hed)))
				reruns))))
     
     ;; At this point hed test registration must be completed.
     ;;
     ((eq? (hash-table-ref/default test-registry (runs:make-full-test-name test-name item-path) #f)
	   'start)
      (debug:print-info 0 "Waiting on test registration(s): "
			(string-intersperse 
			 (filter (lambda (x)
				   (eq? (hash-table-ref/default test-registry x #f) 'start))
				 (hash-table-keys test-registry))
			 ", "))
      (thread-sleep! 0.1)
      (set! loop-list (list hed tal reg reruns)))
     
     ;; If no resources are available just kill time and loop again
     ;;
     ((not have-resources) ;; simply try again after waiting a second
      (debug:print-info 1 "no resources to run new tests, waiting ...")
      ;; Have gone back and forth on this but db starvation is an issue.
      ;; wait one second before looking again to run jobs.
      (thread-sleep! 1) ;; (+ 2 *global-delta*))
      ;; could have done hed tal here but doing car/cdr of newtal to rotate tests
      (set! loop-list (list (car newtal)(cdr newtal) reg reruns)))
     ;; (loop hed tal reg reruns))
     
     ;; This is the final stage, everything is in place so launch the test
     ;;
     ((and have-resources
	   (or (null? prereqs-not-met)
	       (and (eq? testmode 'toplevel)
		    (null? non-completed))))
      (run:test run-id run-info keyvals runname test-record flags #f test-registry)
      (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'running)
      (runs:shrink-can-run-more-tests-count)  ;; DELAY TWEAKER (still needed?)
      ;; (thread-sleep! *global-delta*)
      (if (or (not (null? tal))(not (null? reg)))
	  (set! loop-list (list (runs:queue-next-hed tal reg reglen regfull)
				(runs:queue-next-tal tal reg reglen regfull)
				(runs:queue-next-reg tal reg reglen regfull)
				reruns))))
     
     ;; must be we have unmet prerequisites
     ;;
     (else
      (debug:print 4 "FAILS: " fails)
      ;; If one or more of the prereqs-not-met are FAIL then we can issue
      ;; a message and drop hed from the items to be processed.
      (if (null? fails)
	  (begin
	    ;; couldn't run, take a breather
	    (debug:print-info 4 "Shouldn't really get here, race condition? Unable to launch more tests at this moment, killing time ...")
	    ;; (thread-sleep! (+ 0.01 *global-delta*)) ;; long sleep here - no resources, may as well be patient
	    ;; we made new tal by sticking hed at the back of the list
	    (set! loop-list (list (car newtal)(cdr newtal) reg reruns)))
	  ;; the waiton is FAIL so no point in trying to run hed ever again
	  (if (or (not (null? reg))(not (null? tal)))
	      (if (vector? hed)
		  (begin 
		    (debug:print 1 "WARN: Dropping test " (db:test-get-testname hed) "/" (db:test-get-item-path hed)
				 " from the launch list as it has prerequistes that are FAIL")
		    (runs:shrink-can-run-more-tests-count) ;; DELAY TWEAKER (still needed?)
		    ;; (thread-sleep! *global-delta*)
		    (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'removed)
		    (set! loop (list (runs:queue-next-hed tal reg reglen regfull)
				     (runs:queue-next-tal tal reg reglen regfull)
				     (runs:queue-next-reg tal reg reglen regfull)
				     (cons hed reruns))))
		  (begin
		    (debug:print 1 "WARN: Test not processed correctly. Could be a race condition in your test implementation? " hed) ;;  " as it has prerequistes that are FAIL. (NOTE: hed is not a vector)")
		    (runs:shrink-can-run-more-tests-count) ;; DELAY TWEAKER (still needed?)
		    (set! loop-list (list hed tal reg reruns))))))))
    loop-list)) ;; END OF INNER COND

;; End of INNER COND for launchable test.


;; test-records is a hash table testname:item_path => vector < testname testconfig waitons priority items-info ... >
(define (runs:run-tests-queue run-id runname test-records keyvals flags test-patts required-tests reglen-in)
  ;; At this point the list of parent tests is expanded 
  ;; NB// Should expand items here and then insert into the run queue.
  (debug:print 5 "test-records: " test-records ", flags: " (hash-table->alist flags))
  (let ((run-info              (cdb:remote-run db:get-run-info #f run-id))
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
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
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
	 ;; items is #f then the test is ok to be handed off to launch (but not before)
	 ;; 
	 ((not items)
	  (debug:print-info 4 "OUTER COND: (not items)")
	  (if (and (not (tests:match test-patts (tests:testqueue-get-testname test-record) item-path required: required-tests))
		   (not (null? tal)))
	      (loop (car tal)(cdr tal) reg reruns))
	  (let* ((run-limits-info         (runs:can-run-more-tests jobgroup max-concurrent-jobs)) ;; look at the test jobgroup and tot jobs running
		 ;; (open-run-close runs:can-run-more-tests #f jobgroup max-concurrent-jobs)) ;; look at the test jobgroup and tot jobs running
		 (have-resources          (car run-limits-info))
		 (num-running             (list-ref run-limits-info 1))
		 (num-running-in-jobgroup (list-ref run-limits-info 2))
		 (max-concurrent-jobs     (list-ref run-limits-info 3))
		 (job-group-limit         (list-ref run-limits-info 4))
		 (prereqs-not-met         (mt:get-prereqs-not-met run-id waitons item-path mode: testmode))
		 (fails                   (runs:calc-fails prereqs-not-met))
		 (non-completed           (runs:calc-not-completed prereqs-not-met)))
	    (debug:print-info 4 "have-resources: " have-resources " prereqs-not-met: (" 
			      (string-intersperse 
			       (map (lambda (t)
				      (if (vector? t)
					  (conc (db:test-get-state t) "/" (db:test-get-status t))
					  (conc " WARNING: t is not a vector=" t )))
				    prereqs-not-met) ", ") ") fails: " fails)
	    (debug:print-info 4 "hed=" hed "\n  test-record=" test-record "\n  test-name: " test-name "\n  item-path: " item-path "\n  test-patts: " test-patts)

	    ;; Don't know at this time if the test have been launched at some time in the past
	    ;; i.e. is this a re-launch?
	    (debug:print-info 4 "run-limits-info = " run-limits-info)

	    (cond ;; INNER COND #1 for a launchable test

	     ;; Check item path against item-patts, 
	     ;;
	     ((not (tests:match test-patts (tests:testqueue-get-testname test-record) item-path required: required-tests)) ;; This test/itempath is not to be run
	      ;; else the run is stuck, temporarily or permanently
	      ;; but should check if it is due to lack of resources vs. prerequisites
	      (debug:print-info 1 "Skipping " (tests:testqueue-get-testname test-record) " " item-path " as it doesn't match " test-patts)
	      ;; (thread-sleep! *global-delta*)
	      (if (or (not (null? tal))(not (null? reg)))
		  (loop (runs:queue-next-hed tal reg reglen regfull)
			(runs:queue-next-tal tal reg reglen regfull)
			(runs:queue-next-reg tal reg reglen regfull)
			reruns)))

	     ;; Register tests 
	     ;;
	     ((not (hash-table-ref/default test-registry (runs:make-full-test-name test-name item-path) #f))
	      (debug:print-info 4 "Pre-registering test " test-name "/" item-path " to create placeholder" )
	      (let ((th (make-thread (lambda ()
				       (mutex-lock! registry-mutex)
				       (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'start)
				       (mutex-unlock! registry-mutex)
				       ;; If haven't done it before register a top level test if this is an itemized test
				       (if (not (eq? (hash-table-ref/default test-registry (runs:make-full-test-name test-name "") #f) 'done))
					   (cdb:tests-register-test *runremote* run-id test-name ""))
				       (cdb:tests-register-test *runremote* run-id test-name item-path)
				       (mutex-lock! registry-mutex)
				       (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'done)
				       (mutex-unlock! registry-mutex))
				     (conc test-name "/" item-path))))
		(thread-start! th))
	      (runs:shrink-can-run-more-tests-count)   ;; DELAY TWEAKER (still needed?)
	      (if (and (null? tal)(null? reg))
		  (loop hed tal (append reg (list hed)) reruns)
		  (loop (runs:queue-next-hed tal reg reglen regfull)
			(runs:queue-next-tal tal reg reglen regfull)
			;; NB// Here we are building reg as we register tests
			;; if regfull we must pop the front item off reg
			(if regfull
			    (append (cdr reg) (list hed))
			    (append reg (list hed)))
			reruns)))

	     ;; At this point hed test registration must be completed.
	     ;;
	     ((eq? (hash-table-ref/default test-registry (runs:make-full-test-name test-name item-path) #f)
		   'start)
	      (debug:print-info 0 "Waiting on test registration(s): "
				(string-intersperse 
				 (filter (lambda (x)
					   (eq? (hash-table-ref/default test-registry x #f) 'start))
					 (hash-table-keys test-registry))
				 ", "))
	      (thread-sleep! 0.1)
	      (loop hed tal reg reruns))

	     ;; If no resources are available just kill time and loop again
	     ;;
	     ((not have-resources) ;; simply try again after waiting a second
	      (debug:print-info 1 "no resources to run new tests, waiting ...")
	      ;; Have gone back and forth on this but db starvation is an issue.
	      ;; wait one second before looking again to run jobs.
	      (thread-sleep! 1) ;; (+ 2 *global-delta*))
	      ;; could have done hed tal here but doing car/cdr of newtal to rotate tests
	      (loop (car newtal)(cdr newtal) reg reruns))
	      ;; (loop hed tal reg reruns))

	     ;; This is the final stage, everything is in place so launch the test
	     ;;
	     ((and have-resources
		   (or (null? prereqs-not-met)
		       (and (eq? testmode 'toplevel)
			    (null? non-completed))))
	      (run:test run-id run-info keyvals runname test-record flags #f test-registry)
	      (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'running)
	      (runs:shrink-can-run-more-tests-count)  ;; DELAY TWEAKER (still needed?)
	      ;; (thread-sleep! *global-delta*)
	      (if (or (not (null? tal))(not (null? reg)))
		  (loop (runs:queue-next-hed tal reg reglen regfull)
			(runs:queue-next-tal tal reg reglen regfull)
			(runs:queue-next-reg tal reg reglen regfull)
			reruns)))

	     ;; must be we have unmet prerequisites
	     ;;
	     (else
	      (debug:print 4 "FAILS: " fails)
	      ;; If one or more of the prereqs-not-met are FAIL then we can issue
	      ;; a message and drop hed from the items to be processed.
	      (if (null? fails)
		  (begin
		    ;; couldn't run, take a breather
		    (debug:print-info 4 "Shouldn't really get here, race condition? Unable to launch more tests at this moment, killing time ...")
		    ;; (thread-sleep! (+ 0.01 *global-delta*)) ;; long sleep here - no resources, may as well be patient
		    ;; we made new tal by sticking hed at the back of the list
		    (loop (car newtal)(cdr newtal) reg reruns))
		  ;; the waiton is FAIL so no point in trying to run hed ever again
		  (if (or (not (null? reg))(not (null? tal)))
		      (if (vector? hed)
			  (begin 
			    (debug:print 1 "WARN: Dropping test " (db:test-get-testname hed) "/" (db:test-get-item-path hed)
					 " from the launch list as it has prerequistes that are FAIL")
			    (runs:shrink-can-run-more-tests-count) ;; DELAY TWEAKER (still needed?)
			    ;; (thread-sleep! *global-delta*)
			    (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'removed)
			    (loop (runs:queue-next-hed tal reg reglen regfull)
				  (runs:queue-next-tal tal reg reglen regfull)
				  (runs:queue-next-reg tal reg reglen regfull)
				  (cons hed reruns)))
			  (begin
			    (debug:print 1 "WARN: Test not processed correctly. Could be a race condition in your test implementation? " hed) ;;  " as it has prerequistes that are FAIL. (NOTE: hed is not a vector)")
			    (runs:shrink-can-run-more-tests-count) ;; DELAY TWEAKER (still needed?)
			    (loop hed tal reg reruns))))))))) ;; END OF INNER COND

	 ;; End of INNER COND for launchable test.

	 ;; case where an items came in as a list been processed
	 ((and (list? items)     ;; thus we know our items are already calculated
	       (not   itemdat))  ;; and not yet expanded into the list of things to be done
	  (debug:print-info 4 "INNER COND: (and (list? items)(not itemdat))")
	  (if (and (debug:debug-mode 1) ;; (>= *verbosity* 1)
		   (> (length items) 0)







<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
|
<
<
<
|
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<







645
646
647
648
649
650
651


















652



653




















































































































654
655
656
657
658
659
660
	 ;; items is #f then the test is ok to be handed off to launch (but not before)
	 ;; 
	 ((not items)
	  (debug:print-info 4 "OUTER COND: (not items)")
	  (if (and (not (tests:match test-patts (tests:testqueue-get-testname test-record) item-path required: required-tests))
		   (not (null? tal)))
	      (loop (car tal)(cdr tal) reg reruns))


















	  (let ((loop-list (runs:process-expanded-tests hed tal reg reruns reglen regfull test-record runname test-name item-path jobgroup max-concurrent-jobs run-id waitons item-path testmode)))



	    (apply loop loop-list)))





















































































































	 ;; case where an items came in as a list been processed
	 ((and (list? items)     ;; thus we know our items are already calculated
	       (not   itemdat))  ;; and not yet expanded into the list of things to be done
	  (debug:print-info 4 "INNER COND: (and (list? items)(not itemdat))")
	  (if (and (debug:debug-mode 1) ;; (>= *verbosity* 1)
		   (> (length items) 0)