Megatest

Diff
Login

Differences From Artifact [0a93b5efd0]:

To Artifact [31523ae98e]:


215
216
217
218
219
220
221

222
223

224
225
226
227
228
229
230
231
232
	 (test-names         #f)  ;; (tests:filter-test-names all-test-names test-patts))
	 (required-tests     #f)  ;;(lset-intersection equal? (string-split test-patts ",") test-names))) ;; test-names)) ;; Added test-names as initial for required-tests but that failed to work
	 (task-key           (conc (hash-table->alist flags) " " (get-host-name) " " (current-process-id)))
	 (tasks-db           (tasks:open-db)))

    (set-signal-handler! signal/int
			 (lambda (signum)

			   (let ((tdb (tasks:open-db)))
			     (tasks:set-state-given-param-key tdb task-key "killed")

			     (sqlite3:finalize! tdb))
			   (print "Killed by sigint. Exiting")
			   (exit)))

    ;; register this run in monitor.db
    (tasks:add tasks-db "run-tests" user target runname test-patts task-key) ;; params)
    (tasks:set-state-given-param-key tasks-db task-key "running")
    (runs:set-megatest-env-vars run-id inkeys: keys inrunname: runname) ;; these may be needed by the launching process
    (if (file-exists? runconfigf)







>


>

|







215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
	 (test-names         #f)  ;; (tests:filter-test-names all-test-names test-patts))
	 (required-tests     #f)  ;;(lset-intersection equal? (string-split test-patts ",") test-names))) ;; test-names)) ;; Added test-names as initial for required-tests but that failed to work
	 (task-key           (conc (hash-table->alist flags) " " (get-host-name) " " (current-process-id)))
	 (tasks-db           (tasks:open-db)))

    (set-signal-handler! signal/int
			 (lambda (signum)
			   (signal-mask! signum)
			   (let ((tdb (tasks:open-db)))
			     (tasks:set-state-given-param-key tdb task-key "killed")
			     ;; (sqlite3:interrupt! tdb) ;; seems silly?
			     (sqlite3:finalize! tdb))
			   (print "Killed by signal " signum ". Exiting")
			   (exit)))

    ;; register this run in monitor.db
    (tasks:add tasks-db "run-tests" user target runname test-patts task-key) ;; params)
    (tasks:set-state-given-param-key tasks-db task-key "running")
    (runs:set-megatest-env-vars run-id inkeys: keys inrunname: runname) ;; these may be needed by the launching process
    (if (file-exists? runconfigf)
247
248
249
250
251
252
253



254
255
256
257





258
259
260
261
262
263
264
265
266
267
268
    (debug:print-info 0 "all tests:  " (string-intersperse (sort all-test-names string<) " "))
    (debug:print-info 0 "test names: " (string-intersperse (sort test-names string<) " "))

    ;; on the first pass or call to run-tests set FAILS to NOT_STARTED if
    ;; -keepgoing is specified
    (if (eq? *passnum* 0)
	(begin



	  ;; have to delete test records where NOT_STARTED since they can cause -keepgoing to 
	  ;; get stuck due to becoming inaccessible from a failed test. I.e. if test B depends 
	  ;; on test A but test B reached the point on being registered as NOT_STARTED and test
	  ;; A failed for some reason then on re-run using -keepgoing the run can never complete.





	  (for-each (lambda (state)
		      (rmt:general-call 'delete-tests-in-state run-id state))
		    (cons "NOT_STARTED" (string-split (or (configf:lookup *configdat* "setup" "allow-auto-rerun") ""))))
	  (rmt:set-tests-state-status run-id test-names #f "FAIL" "NOT_STARTED" "FAIL")))

    ;; Ensure all tests are registered in the test_meta table
    (runs:update-all-test_meta #f)

    ;; now add non-directly referenced dependencies (i.e. waiton)
    ;;======================================================================
    ;; refactoring this block into tests:get-full-data







>
>
>




>
>
>
>
>

|
|
<







249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270

271
272
273
274
275
276
277
    (debug:print-info 0 "all tests:  " (string-intersperse (sort all-test-names string<) " "))
    (debug:print-info 0 "test names: " (string-intersperse (sort test-names string<) " "))

    ;; on the first pass or call to run-tests set FAILS to NOT_STARTED if
    ;; -keepgoing is specified
    (if (eq? *passnum* 0)
	(begin
	  ;; Is this still necessary? I think not. Unreachable tests are marked as such and 
	  ;; should not cause problems here.
	  ;;
	  ;; have to delete test records where NOT_STARTED since they can cause -keepgoing to 
	  ;; get stuck due to becoming inaccessible from a failed test. I.e. if test B depends 
	  ;; on test A but test B reached the point on being registered as NOT_STARTED and test
	  ;; A failed for some reason then on re-run using -keepgoing the run can never complete.
	  ;;
	  ;; (rmt:general-call 'delete-tests-in-state run-id "NOT_STARTED")
	  
	  ;; Now convert FAIL and anything in allow-auto-rerun to NOT_STARTED
	  ;;
	  (for-each (lambda (state)
		      (rmt:set-tests-state-status run-id test-names state #f "NOT_STARTED" state))
		    (string-split (or (configf:lookup *configdat* "setup" "allow-auto-rerun") "")))))


    ;; Ensure all tests are registered in the test_meta table
    (runs:update-all-test_meta #f)

    ;; now add non-directly referenced dependencies (i.e. waiton)
    ;;======================================================================
    ;; refactoring this block into tests:get-full-data
354
355
356
357
358
359
360
361

362














363
364
365
366
367
368
369

    (if (not (null? required-tests))
	(debug:print-info 1 "Adding " required-tests " to the run queue"))
    ;; NOTE: these are all parent tests, items are not expanded yet.
    (debug:print-info 4 "test-records=" (hash-table->alist test-records))
    (let ((reglen (configf:lookup *configdat* "setup" "runqueue")))
      (if (> (length (hash-table-keys test-records)) 0)
	  (begin

	    (runs:run-tests-queue run-id runname test-records keyvals flags test-patts required-tests (any->number reglen) all-tests-registry)














	    ;; if run-count > 0 call, set -preclean and -rerun STUCK/DEAD
	    (if (> run-count 0)
		(begin
		  (if (not (hash-table-ref/default flags "-preclean" #f))
		      (hash-table-set! flags "-preclean" #t))
		  (if (not (hash-table-ref/default flags "-rerun" #f))
		      (hash-table-set! flags "-rerun" "STUCK/DEAD,n/a,ZERO_ITEMS"))







|
>
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>







363
364
365
366
367
368
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

    (if (not (null? required-tests))
	(debug:print-info 1 "Adding " required-tests " to the run queue"))
    ;; NOTE: these are all parent tests, items are not expanded yet.
    (debug:print-info 4 "test-records=" (hash-table->alist test-records))
    (let ((reglen (configf:lookup *configdat* "setup" "runqueue")))
      (if (> (length (hash-table-keys test-records)) 0)
	  (let* ((keep-going #t)
		 (th1        (make-thread (lambda ()
					    (runs:run-tests-queue run-id runname test-records keyvals flags test-patts required-tests (any->number reglen) all-tests-registry))
					  "runs:run-tests-queue"))
		 (th2        (make-thread (lambda ()				    
					    ;; (rmt:find-and-mark-incomplete-all-runs))))) CAN'T INTERRUPT IT ...
					    (let ((run-ids (rmt:get-all-run-ids)))
					      (for-each (lambda (run-id)
							  (if keep-going
							      (rmt:find-and-mark-incomplete run-id #f))) ;; ovr-deadtime)))
							run-ids)))
					  "runs: mark-incompletes")))
	    (thread-start! th1)
	    (thread-start! th2)
	    (thread-join! th1)
	    (set! keep-going #f)
	    (thread-join! th2)
	    ;; if run-count > 0 call, set -preclean and -rerun STUCK/DEAD
	    (if (> run-count 0)
		(begin
		  (if (not (hash-table-ref/default flags "-preclean" #f))
		      (hash-table-set! flags "-preclean" #t))
		  (if (not (hash-table-ref/default flags "-rerun" #f))
		      (hash-table-set! flags "-rerun" "STUCK/DEAD,n/a,ZERO_ITEMS"))
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
	(setenv "MT_RUNNAME"   runname)
	(runs:set-megatest-env-vars run-id inrunname: runname) ;; these may be needed by the launching process
	(let ((items-list (items:get-items-from-config tconfig)))
	  (if (list? items-list)
	      (begin
		(if (null? items-list)
		    (let ((test-id (rmt:get-test-id run-id test-name "")))
		      (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "ZERO_ITEMS" "Failed to run due to failed prerequisites")))
		(tests:testqueue-set-items! test-record items-list)
		(list hed tal reg reruns))
	      (begin
		(debug:print 0 "ERROR: The proc from reading the items table did not yield a list - please report this")
		(exit 1))))))

     ((and (null? fails)







|







497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
	(setenv "MT_RUNNAME"   runname)
	(runs:set-megatest-env-vars run-id inrunname: runname) ;; these may be needed by the launching process
	(let ((items-list (items:get-items-from-config tconfig)))
	  (if (list? items-list)
	      (begin
		(if (null? items-list)
		    (let ((test-id (rmt:get-test-id run-id test-name "")))
		      (if test-id (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "ZERO_ITEMS" "Failed to run due to failed prerequisites"))))
		(tests:testqueue-set-items! test-record items-list)
		(list hed tal reg reruns))
	      (begin
		(debug:print 0 "ERROR: The proc from reading the items table did not yield a list - please report this")
		(exit 1))))))

     ((and (null? fails)
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
	(if (and give-up
		 (not (and (null? tal)(null? reg))))
	    (let ((trimmed-tal (mt:discard-blocked-tests run-id hed tal test-records))
		  (trimmed-reg (mt:discard-blocked-tests run-id hed reg test-records)))
	      (debug:print 1 "WARNING: test " hed " has discarded prerequisites, removing it from the queue")

	      (let ((test-id (rmt:get-test-id run-id hed "")))
		(mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_DISCARDED" "Failed to run due to discarded prerequisites"))
	      
	      (if (and (null? trimmed-tal)
		       (null? trimmed-reg))
		  #f
		  (list (runs:queue-next-hed trimmed-tal trimmed-reg reglen regfull)
			(runs:queue-next-tal trimmed-tal trimmed-reg reglen regfull)
			(runs:queue-next-reg trimmed-tal trimmed-reg reglen regfull)







|







535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
	(if (and give-up
		 (not (and (null? tal)(null? reg))))
	    (let ((trimmed-tal (mt:discard-blocked-tests run-id hed tal test-records))
		  (trimmed-reg (mt:discard-blocked-tests run-id hed reg test-records)))
	      (debug:print 1 "WARNING: test " hed " has discarded prerequisites, removing it from the queue")

	      (let ((test-id (rmt:get-test-id run-id hed "")))
		(if test-id (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_DISCARDED" "Failed to run due to discarded prerequisites")))
	      
	      (if (and (null? trimmed-tal)
		       (null? trimmed-reg))
		  #f
		  (list (runs:queue-next-hed trimmed-tal trimmed-reg reglen regfull)
			(runs:queue-next-tal trimmed-tal trimmed-reg reglen regfull)
			(runs:queue-next-reg trimmed-tal trimmed-reg reglen regfull)
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
	    (debug:print-info 1 "no fails in prerequisites for " hed " but also none running, keeping " hed " for now. Try count: " (hash-table-ref/default *seen-cant-run-tests* hed 0))
	    ;; num-retries code was here
	    ;; we use this opportunity to move contents of reg to tal
	    (list (car newtal)(append (cdr newtal) reg) '() reruns)) ;; an issue with prereqs not yet met?
	  (begin
	    (debug:print-info 1 "no fails in prerequisites for " hed " but nothing seen running in a while, dropping test " hed " from the run queue")
	    (let ((test-id (rmt:get-test-id run-id hed "")))
	      (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "TIMED_OUT" "Nothing seen running in a while."))
	    (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))))

     ((and 
       (or (not (null? fails))
	   (not (null? prereq-fails)))
       (member 'normal testmode))
      (debug:print-info 1 "test "  hed " (mode=" testmode ") has failed prerequisite(s); "
			(string-intersperse (map (lambda (t)(conc (db:test-get-testname t) ":" (db:test-get-state t)"/"(db:test-get-status t))) fails) ", ")
			", removing it from to-do list")
      (let ((test-id (rmt:get-test-id run-id hed "")))

	(if (not (null? prereq-fails))
	    (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_DISCARDED" "Failed to run due to prior failed prerequisites")
	    (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_FAIL"      "Failed to run due to failed prerequisites")))
      (if (or (not (null? reg))(not (null? tal)))
	  (begin
	    (hash-table-set! test-registry hed 'CANNOTRUN)
	    (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)))







|













>
|
|
|







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
	    (debug:print-info 1 "no fails in prerequisites for " hed " but also none running, keeping " hed " for now. Try count: " (hash-table-ref/default *seen-cant-run-tests* hed 0))
	    ;; num-retries code was here
	    ;; we use this opportunity to move contents of reg to tal
	    (list (car newtal)(append (cdr newtal) reg) '() reruns)) ;; an issue with prereqs not yet met?
	  (begin
	    (debug:print-info 1 "no fails in prerequisites for " hed " but nothing seen running in a while, dropping test " hed " from the run queue")
	    (let ((test-id (rmt:get-test-id run-id hed "")))
	      (if test-id (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "TIMED_OUT" "Nothing seen running in a while.")))
	    (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))))

     ((and 
       (or (not (null? fails))
	   (not (null? prereq-fails)))
       (member 'normal testmode))
      (debug:print-info 1 "test "  hed " (mode=" testmode ") has failed prerequisite(s); "
			(string-intersperse (map (lambda (t)(conc (db:test-get-testname t) ":" (db:test-get-state t)"/"(db:test-get-status t))) fails) ", ")
			", removing it from to-do list")
      (let ((test-id (rmt:get-test-id run-id hed "")))
	(if test-id
	    (if (not (null? prereq-fails))
		(mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_DISCARDED" "Failed to run due to prior failed prerequisites")
		(mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_FAIL"      "Failed to run due to failed prerequisites"))))
      (if (or (not (null? reg))(not (null? tal)))
	  (begin
	    (hash-table-set! test-registry hed 'CANNOTRUN)
	    (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)))
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
		reruns)
	  #f))
     
     ;; 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" )
      (if (eq? *transport-type* 'fs) ;; no point in parallel registration if use fs
	  (begin
	    (rmt:general-call 'register-test run-id run-id test-name item-path)
	    (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'done))
	  (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)







|







667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
		reruns)
	  #f))
     
     ;; 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" )
      (if #t ;; always do firm registration now in v1.60 and greater ;; (eq? *transport-type* 'fs) ;; no point in parallel registration if use fs
	  (begin
	    (rmt:general-call 'register-test run-id run-id test-name item-path)
	    (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'done))
	  (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)
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
	  ;; 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 "WARNING: Dropping test " test-name "/" item-path
				 " from the launch list as it has prerequistes that are FAIL")
		    (let ((test-id (rmt:get-test-id run-id hed "")))
		      (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_FAIL" "Failed to run due to failed prerequisites"))
		    (runs:shrink-can-run-more-tests-count) ;; DELAY TWEAKER (still needed?)
		    ;; (thread-sleep! *global-delta*)
		    ;; This next is for the items
		    (mt:test-set-state-status-by-testname run-id test-name item-path "NOT_STARTED" "BLOCKED" #f)
		    (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'removed)
		    (list (runs:queue-next-hed tal reg reglen regfull)
			  (runs:queue-next-tal tal reg reglen regfull)







|







771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
	  ;; 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 "WARNING: Dropping test " test-name "/" item-path
				 " from the launch list as it has prerequistes that are FAIL")
		    (let ((test-id (rmt:get-test-id run-id hed "")))
		      (if test-id (mt:test-set-state-status-by-id run-id test-id "NOT_STARTED" "PREQ_FAIL" "Failed to run due to failed prerequisites")))
		    (runs:shrink-can-run-more-tests-count) ;; DELAY TWEAKER (still needed?)
		    ;; (thread-sleep! *global-delta*)
		    ;; This next is for the items
		    (mt:test-set-state-status-by-testname run-id test-name item-path "NOT_STARTED" "BLOCKED" #f)
		    (hash-table-set! test-registry (runs:make-full-test-name test-name item-path) 'removed)
		    (list (runs:queue-next-hed tal reg reglen regfull)
			  (runs:queue-next-tal tal reg reglen regfull)
885
886
887
888
889
890
891


892
893
894
895

896
897
898
899
900
901
902
	       (tal         (cdr sorted-test-names))
	       (reg         '()) ;; registered, put these at the head of tal 
	       (reruns      '()))

      (if (not (null? reruns))(debug:print-info 4 "reruns=" reruns))

      ;; Here we mark any old defunct tests as incomplete. Do this every fifteen minutes


      (if (> (current-seconds)(+ last-time-incomplete 900))
          (begin
            (set! last-time-incomplete (current-seconds))
            (rmt:find-and-mark-incomplete-all-runs)))


      ;; (print "Top of loop, hed=" hed ", tal=" tal " ,reruns=" reruns)
      (let* ((test-record (hash-table-ref test-records hed))
	     (test-name   (tests:testqueue-get-testname test-record))
	     (tconfig     (tests:testqueue-get-testconfig test-record))
	     (jobgroup    (config-lookup tconfig "test_meta" "jobgroup"))
	     (testmode    (let ((m (config-lookup tconfig "requirements" "mode")))







>
>



|
>







910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
	       (tal         (cdr sorted-test-names))
	       (reg         '()) ;; registered, put these at the head of tal 
	       (reruns      '()))

      (if (not (null? reruns))(debug:print-info 4 "reruns=" reruns))

      ;; Here we mark any old defunct tests as incomplete. Do this every fifteen minutes
      ;; moving this to a parallel thread and just run it once.
      ;;
      (if (> (current-seconds)(+ last-time-incomplete 900))
          (begin
            (set! last-time-incomplete (current-seconds))
            ;; (rmt:find-and-mark-incomplete-all-runs)
	    ))

      ;; (print "Top of loop, hed=" hed ", tal=" tal " ,reruns=" reruns)
      (let* ((test-record (hash-table-ref test-records hed))
	     (test-name   (tests:testqueue-get-testname test-record))
	     (tconfig     (tests:testqueue-get-testconfig test-record))
	     (jobgroup    (config-lookup tconfig "test_meta" "jobgroup"))
	     (testmode    (let ((m (config-lookup tconfig "requirements" "mode")))