Overview
Comment: | Refactor process on runs launch loop |
---|---|
Downloads: | Tarball | ZIP archive | SQL archive |
Timelines: | family | ancestors | descendants | both | refactor |
Files: | files | file ages | folders |
SHA1: |
8de570b5170519fb4328c384e60dd2e7 |
User & Date: | mrwellan on 2013-04-30 21:31:32 |
Other Links: | branch diff | manifest | tags |
Context
2013-04-30
| ||
22:21 | Mostly working reworked loop, keeping for prosterity but will be replaced check-in: 60c6d1adb3 user: matt tags: refactor | |
21:31 | Refactor process on runs launch loop check-in: 8de570b517 user: mrwellan tags: refactor | |
20:55 | Capturing illustration of run launch loop check-in: 051ff9ebd0 user: matt tags: refactor | |
Changes
Modified runs.scm from [71ed251c12] to [ad0152f814].
︙ | ︙ | |||
318 319 320 321 322 323 324 325 326 327 328 329 330 331 | (if (not (vector? t)) (conc t) (conc (db:test-get-testname t) ":" (db:test-get-state t) "/" (db:test-get-status t)))) lst)) (define (runs:make-full-test-name testname itempath) (if (equal? itempath "") testname (conc testname "/" itempath))) ;; 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 keyvallst flags test-patts) ;; 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 ", keyvallst: " keyvallst " flags: " (hash-table->alist flags)) (let ((sorted-test-names (tests:sort-by-priority-and-waiton test-records)) | > > > > > > > > > > > > > > > > > > > > > > > > > > | | | > > | > > | | | < < < | | > > > | | | < | < | | | | | | | < | > | > | < | | | | | > > > | | | | > > > | | | | | > > > | | | 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 | (if (not (vector? t)) (conc t) (conc (db:test-get-testname t) ":" (db:test-get-state t) "/" (db:test-get-status t)))) lst)) (define (runs:make-full-test-name testname itempath) (if (equal? itempath "") testname (conc testname "/" itempath))) (define (runs:queue-next-hed tal reg n) (if (> (length reg) n) (car reg) (if (null? tal) (car reg) (car tal)))) (define (runs:queue-next-tal tal reg n) (if (> (length reg) n) ;; rotate (let ((nexttal (if (null? tal) reg (cdr tal)))) (if (null? nexttal) reg nexttal)) (if (null? reg) '() (cdr reg)))) (define (runs:queue-next-reg tal reg n) (if (> (length reg) n) (cdr reg) reg)) ;; 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 keyvallst flags test-patts) ;; 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 ", keyvallst: " keyvallst " flags: " (hash-table->alist flags)) (let ((sorted-test-names (tests:sort-by-priority-and-waiton test-records)) (test-registry (make-hash-table)) (registry-mutex (make-mutex)) (num-retries 0) (max-retries (config-lookup *configdat* "setup" "maxretries")) (max-concurrent-jobs (let ((mcj (config-lookup *configdat* "setup" "max_concurrent_jobs"))) (if (and mcj (string->number mcj)) (string->number mcj) 1))) (reglen 10)) ;; length of the register queue ahead (set! max-retries (if (and max-retries (string->number max-retries))(string->number max-retries) 100)) (if (not (null? sorted-test-names)) (let loop ((hed (car sorted-test-names)) (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)) ;; (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)) (testmode (let ((m (config-lookup tconfig "requirements" "mode"))) (if m (string->symbol m) 'normal))) (waitons (tests:testqueue-get-waitons test-record)) (priority (tests:testqueue-get-priority test-record)) (itemdat (tests:testqueue-get-itemdat test-record)) ;; itemdat can be a string, list or #f (items (tests:testqueue-get-items test-record)) (item-path (item-list->path itemdat)) (newtal (append tal (list hed)))) ;; (if (> (length reg) 10) ;; (begin ;; (set! tal (cons hed tal)) ;; (set! hed (car reg)) ;; (set! reg (cdr reg)) ;; (set! newtal tal))) (debug:print 6 "test-name: " test-name "\n hed: " hed "\n itemdat: " itemdat "\n items: " items "\n item-path: " item-path "\n waitons: " waitons "\n num-retries: " num-retries "\n tal: " tal "\n reruns: " reruns) ;; check for hed in waitons => this would be circular, remove it and issue an ;; error (if (member test-name waitons) (begin (debug:print 0 "ERROR: test " test-name " has listed itself as a waiton, please correct this!") (set! waiton (filter (lambda (x)(not (equal? x hed))) waitons)))) (cond ;; OUTER COND ((not items) ;; when false the test is ok to be handed off to launch (but not before) (let* ((run-limits-info (runs:can-run-more-tests test-record 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 (db: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 8 "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)) ;; 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 (not (null? tal)) (loop (runs:queue-next-hed tal reg reglen) (runs:queue-next-tal tal reg reglen) (runs:queue-next-reg tal reg reglen) reruns))) ;; Registry has been started for this test but has not yet completed ;; this should be rare, the case where there are only a couple of tests and the db is slow ;; delay a short while and continue ;; ((eq? (hash-table-ref/default test-registry (runs:make-full-test-name test-name item-path) #f) 'start) ;; (thread-sleep! 0.01) ;; (loop (car newtal)(cdr newtal) reruns)) ;; count number of 'done, if more than 100 then skip on through. ((not (hash-table-ref/default test-registry (runs:make-full-test-name test-name item-path) #f)) ;; ) ;; too many changes required. Implement later. (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?) (loop (car newtal) (runs:queue-next-tal tal reg reglen) (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)) ((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)) ((and have-resources (or (null? prereqs-not-met) (and (eq? testmode 'toplevel) (null? non-completed)))) (run:test run-id runname keyvallst test-record flags #f) (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 (not (null? tal)) (loop (runs:queue-next-hed tal reg reglen) (runs:queue-next-tal tal reg reglen) (runs:queue-next-reg tal reg reglen) reruns))) (else ;; must be we have unmet prerequisites (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 (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) (runs:queue-next-tal tal reg reglen) (runs:queue-next-reg tal reg reglen) (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?) ;; (thread-sleep! (+ 0.01 *global-delta*)) (loop hed tal reg reruns))))))))) ;; END OF INNER COND ;; 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 (if (and (debug:debug-mode 1) ;; (>= *verbosity* 1) (> (length items) 0) (> (length (car items)) 0)) (pp items)) (for-each (lambda (my-itemdat) (let* ((new-test-record (let ((newrec (make-tests:testqueue))) (vector-copy! test-record newrec) newrec)) (my-item-path (item-list->path my-itemdat))) (if (tests:match test-patts hed my-item-path) ;; (patt-list-match my-item-path item-patts) ;; yes, we want to process this item, NOTE: Should not need this check here! (let ((newtestname (runs:make-full-test-name hed my-item-path))) ;; test names are unique on testname/item-path (tests:testqueue-set-items! new-test-record #f) (tests:testqueue-set-itemdat! new-test-record my-itemdat) (tests:testqueue-set-item_path! new-test-record my-item-path) (hash-table-set! test-records newtestname new-test-record) (set! tal (cons newtestname tal)))))) ;; since these are itemized create new test names testname/itempath items) (if (not (null? tal)) (begin (debug:print-info 4 "End of items list, looping with next after short delay") ;; (thread-sleep! (+ 0.01 *global-delta*)) (loop (runs:queue-next-hed tal reg reglen) (runs:queue-next-tal tal reg reglen) (runs:queue-next-reg tal reg reglen) reruns)))) ;; if items is a proc then need to run items:get-items-from-config, get the list and loop ;; - but only do that if resources exist to kick off the job ((or (procedure? items)(eq? items 'have-procedure)) (let ((can-run-more (runs:can-run-more-tests test-record max-concurrent-jobs))) (if (and (list? can-run-more) (car can-run-more)) (let* ((prereqs-not-met (db: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 8 "can-run-more: " can-run-more "\n testname: " hed "\n prereqs-not-met: " (runs:pretty-string prereqs-not-met) "\n non-completed: " (runs:pretty-string non-completed) "\n fails: " (runs:pretty-string fails) "\n testmode: " testmode "\n num-retries: " num-retries "\n (eq? testmode 'toplevel): " (eq? testmode 'toplevel) "\n (null? non-completed): " (null? non-completed) |
︙ | ︙ | |||
554 555 556 557 558 559 560 | (setenv "MT_RUNNAME" runname) (set-megatest-env-vars run-id) ;; these may be needed by the launching process (let ((items-list (items:get-items-from-config tconfig))) (if (list? items-list) (begin (tests:testqueue-set-items! test-record items-list) ;; (thread-sleep! *global-delta*) | | | > > > | | > > > | | | 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 | (setenv "MT_RUNNAME" runname) (set-megatest-env-vars run-id) ;; these may be needed by the launching process (let ((items-list (items:get-items-from-config tconfig))) (if (list? items-list) (begin (tests:testqueue-set-items! test-record items-list) ;; (thread-sleep! *global-delta*) (loop hed tal reg reruns)) (begin (debug:print 0 "ERROR: The proc from reading the setup did not yield a list - please report this") (exit 1)))))) ((null? fails) (debug:print-info 4 "fails is null, moving on in the queue but keeping " hed " for now") ;; only increment num-retries when there are no tests runing (if (eq? 0 (list-ref can-run-more 1)) (begin ;; TRY (if (> num-retries 100) ;; first 100 retries are low time cost ;; TRY (thread-sleep! (+ 2 *global-delta*)) ;; TRY (thread-sleep! (+ 0.01 *global-delta*))) (set! num-retries (+ num-retries 1)))) (if (> num-retries max-retries) (if (not (null? tal)) (loop (runs:queue-next-hed tal reg reglen) (runs:queue-next-tal tal reg reglen) (runs:queue-next-reg tal reg reglen) reruns)) (loop (car newtal)(cdr newtal) reg reruns))) ;; an issue with prereqs not yet met? ((and (not (null? fails))(eq? testmode 'normal)) (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") (if (not (null? tal)) (begin ;; (thread-sleep! *global-delta*) (loop (runs:queue-next-hed tal reg reglen) (runs:queue-next-tal tal reg reglen) (runs:queue-next-reg tal reg reglen) (cons hed reruns))))) (else (debug:print 8 "ERROR: No handler for this condition.") ;; TRY (thread-sleep! (+ 1 *global-delta*)) (loop (car newtal)(cdr newtal) reg reruns)))) ;; END OF IF CAN RUN MORE ;; if can't run more just loop with next possible test (begin (debug:print-info 4 "processing the case with a lambda for items or 'have-procedure. Moving through the queue without dropping " hed) ;; (thread-sleep! (+ 2 *global-delta*)) (loop (car newtal)(cdr newtal) reg reruns))))) ;; END OF (or (procedure? items)(eq? items 'have-procedure)) ;; this case should not happen, added to help catch any bugs ((and (list? items) itemdat) (debug:print 0 "ERROR: Should not have a list of items in a test and the itemspath set - please report this") (exit 1)) |
︙ | ︙ | |||
608 609 610 611 612 613 614 | ;; (thread-sleep! (+ 1 *global-delta*)) (if (not (null? newlst)) ;; since reruns have been tacked on to newlst create new reruns from junked (loop (car newlst)(cdr newlst) reg (delete-duplicates junked))))) ((not (null? tal)) (debug:print-info 4 "I'm pretty sure I shouldn't get here.")) (else | | | 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 | ;; (thread-sleep! (+ 1 *global-delta*)) (if (not (null? newlst)) ;; since reruns have been tacked on to newlst create new reruns from junked (loop (car newlst)(cdr newlst) reg (delete-duplicates junked))))) ((not (null? tal)) (debug:print-info 4 "I'm pretty sure I shouldn't get here.")) (else (debug:print-info 4 "Exiting loop with...\n hed=" hed "\n tal=" tal "\n reruns=" reruns)) )))) ;; LET* ((test-record ;; we get here on "drop through" - loop for next test in queue ;; FIXME!!!! THIS SHOULD NOT REQUIRE AN EXIT!!!!!!! (debug:print-info 1 "All tests launched") (thread-sleep! 0.5) |
︙ | ︙ |
Added tests/fdktestqa/testqa/Makefile version [c0a9a7e634].
> > > | 1 2 3 | all : megatest -runtests % -target a/b :runname c |
Modified tests/fdktestqa/testqa/tests/bigrun/testconfig from [54bef43c00] to [cc46c7aa7f].
1 2 3 4 5 6 7 8 9 10 11 | # Add additional steps here. Format is "stepname script" [ezsteps] step1 step1.sh # Test requirements are specified here [requirements] # waiton setup priority 0 # Iteration for your tests are controlled by the items section [items] | | | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 | # Add additional steps here. Format is "stepname script" [ezsteps] step1 step1.sh # Test requirements are specified here [requirements] # waiton setup priority 0 # Iteration for your tests are controlled by the items section [items] NUMBER #{scheme (string-intersperse (map number->string (sort (let loop ((a 0)(res '()))(if (< a 30)(loop (+ a 1)(cons a res)) res)) >)) " ")} # test_meta is a section for storing additional data on your test [test_meta] author matt owner matt description An example test tags tagone,tagtwo |
︙ | ︙ |