Overview
Comment: | Made remove-runs more conservative. It attempts to kill runs before removing |
---|---|
Downloads: | Tarball | ZIP archive | SQL archive |
Timelines: | family | ancestors | descendants | both | dev |
Files: | files | file ages | folders |
SHA1: |
5647188f19793d009323b3a247d759fe |
User & Date: | mrwellan on 2013-06-05 18:20:03 |
Other Links: | branch diff | manifest | tags |
Context
2013-06-05
| ||
20:50 | Added -run-wait feature. check-in: d134500c45 user: mrwellan tags: dev | |
18:20 | Made remove-runs more conservative. It attempts to kill runs before removing check-in: 5647188f19 user: mrwellan tags: dev | |
2013-06-04
| ||
18:08 | Merged v1.54 branch back into development check-in: 0121a1b669 user: mrwellan tags: dev | |
Changes
Modified common.scm from [48fe1a19fd] to [788afc4d5c].
︙ | ︙ | |||
68 69 70 71 72 73 74 | (define *test-info* (make-hash-table)) ;; cache the test info records, update the state, status, run_duration etc. from testdat.db ;; Awful. Please FIXME (define *env-vars-by-run-id* (make-hash-table)) (define *current-run-name* #f) (define (common:clear-caches) | < | 68 69 70 71 72 73 74 75 76 77 78 79 80 81 | (define *test-info* (make-hash-table)) ;; cache the test info records, update the state, status, run_duration etc. from testdat.db ;; Awful. Please FIXME (define *env-vars-by-run-id* (make-hash-table)) (define *current-run-name* #f) (define (common:clear-caches) (set! *keys* (make-hash-table)) (set! *keyvals* (make-hash-table)) (set! *toptest-paths* (make-hash-table)) (set! *test-paths* (make-hash-table)) (set! *test-ids* (make-hash-table)) (set! *test-info* (make-hash-table)) (set! *env-vars-by-run-id* (make-hash-table)) |
︙ | ︙ |
Modified db.scm from [8c210cc811] to [484861755e].
︙ | ︙ | |||
723 724 725 726 727 728 729 | db qry run-id))) keys) (debug:print-info 11 "db:get-key-val-pairs END keys: " keys " run-id: " run-id) (reverse res))) ;; get key vals for a given run-id (define (db:get-key-vals db run-id) | < < < | | | | | | | | | | | | | | < < < < < | | < | | 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 | db qry run-id))) keys) (debug:print-info 11 "db:get-key-val-pairs END keys: " keys " run-id: " run-id) (reverse res))) ;; get key vals for a given run-id (define (db:get-key-vals db run-id) (let* ((keys (db:get-keys db)) (res '())) (debug:print-info 11 "db:get-key-vals START keys: " keys " run-id: " run-id) (for-each (lambda (key) (let ((qry (conc "SELECT " key " FROM runs WHERE id=?;"))) ;; (debug:print 0 "qry: " qry) (sqlite3:for-each-row (lambda (key-val) (set! res (cons key-val res))) db qry run-id))) keys) (debug:print-info 11 "db:get-key-vals END keys: " keys " run-id: " run-id) (reverse res))) ;; The target is keyval1/keyval2..., cached in *target* as it is used often (define (db:get-target db run-id) (let* ((keyvals (db:get-key-vals db run-id)) (thekey (string-intersperse (map (lambda (x)(if x x "-na-")) keyvals) "/"))) thekey)) ;;====================================================================== ;; T E S T S ;;====================================================================== ;; states and statuses are lists, turn them into ("PASS","FAIL"...) and use NOT IN ;; i.e. these lists define what to NOT show. |
︙ | ︙ |
Modified runs.scm from [3e021e4c4c] to [ccf97b9702].
︙ | ︙ | |||
546 547 548 549 550 551 552 | ;; 'set-state-status ;; ;; NB// should pass in keys? ;; (define (runs:operate-on action target runnamepatt testpatt #!key (state #f)(status #f)(new-state-status #f)) (common:clear-caches) ;; clear all caches (let* ((db #f) | | | | > > > > | | > > > > > | | | | | | | | < | | > > > > > > > > > > > > > > > > > > > > > | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | > > | | | < < < < | | | | | | | 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 | ;; 'set-state-status ;; ;; NB// should pass in keys? ;; (define (runs:operate-on action target runnamepatt testpatt #!key (state #f)(status #f)(new-state-status #f)) (common:clear-caches) ;; clear all caches (let* ((db #f) (keys (cdb:remote-run db:get-keys db)) (rundat (cdb:remote-run runs:get-runs-by-patt db keys runnamepatt target)) (header (vector-ref rundat 0)) (runs (vector-ref rundat 1)) (states (if state (string-split state ",") '())) (statuses (if status (string-split status ",") '())) (state-status (if (string? new-state-status) (string-split new-state-status ",") '(#f #f)))) (debug:print-info 4 "runs:operate-on => Header: " header " action: " action " new-state-status: " new-state-status) (if (> 2 (length state-status)) (begin (debug:print 0 "ERROR: the parameter to -set-state-status is a comma delimited string. E.g. COMPLETED,FAIL") (exit))) (for-each (lambda (run) (let ((runkey (string-intersperse (map (lambda (k) (db:get-value-by-header run header k)) keys) "/")) (dirs-to-remove (make-hash-table))) (let* ((run-id (db:get-value-by-header run header "id")) (run-state (db:get-value-by-header run header "state")) (tests (if (not (equal? run-state "locked")) (cdb:remote-run db:get-tests-for-run db run-id testpatt states statuses not-in: #f sort-by: (case action ((remove-runs) 'rundir) (else 'event_time))) '())) (lasttpath "/does/not/exist/I/hope")) (debug:print-info 4 "runs:operate-on run=" run ", header=" header) (if (not (null? tests)) (begin (case action ((remove-runs) (debug:print 1 "Removing tests for run: " runkey " " (db:get-value-by-header run header "runname"))) ((set-state-status) (debug:print 1 "Modifying state and staus for tests for run: " runkey " " (db:get-value-by-header run header "runname"))) ((print-run) (debug:print 1 "Printing info for run " runkey ", run=" run ", tests=" tests ", header=" header) action) (else (debug:print-info 0 "action not recognised " action))) (let ((sorted-tests (sort tests (lambda (a b)(let ((dira (db:test-get-rundir a)) (dirb (db:test-get-rundir b))) (if (and (string? dira)(string? dirb)) (> (string-length dira)(string-length dirb)) #f))))) (test-retry-time (make-hash-table)) (allow-run-time 25)) ;; seconds to allow for killing tests before just brutally killing 'em (let loop ((test (car sorted-tests)) (tal (cdr sorted-tests))) (let* ((test-id (db:test-get-id test)) (new-test-dat (cdb:remote-run db:get-test-info-by-id #f test-id)) (item-path (db:test-get-item-path new-test-dat)) (test-name (db:test-get-testname new-test-dat)) (run-dir (db:test-get-rundir new-test-dat)) ;; run dir is from the link tree (real-dir (if (file-exists? run-dir) (resolve-pathname run-dir) #f)) (test-state (db:test-get-state new-test-dat)) (test-fulln (db:test-get-fullname new-test-dat))) (case action ((remove-runs) (debug:print-info 0 "test-state: " test-state) (if (member test-state (list "RUNNING" "LAUNCHED" "REMOTEHOSTSTART" "KILLREQ")) (begin (if (not (hash-table-ref/default test-retry-time test-fulln #f)) (hash-table-set! test-retry-time test-fulln (current-seconds))) (if (> (- (current-seconds)(hash-table-ref test-retry-time test-fulln)) allow-run-time) ;; This test is not in a correct state for cleaning up. Let's try some graceful shutdown steps first ;; Set the test to "KILLREQ" and wait five seconds then try again. Repeat up to five times then give ;; up and blow it away. (begin (debug:print 0 "WARNING: could not gracefully remove test " test-fulln ", tried to kill it to no avail. Forcing state to FAILEDKILL and continuing") (cdb:remote-run db:test-set-state-status-by-id db (db:test-get-id test) "FAILEDKILL" "n/a" #f) (thread-sleep! 1)) (begin (cdb:remote-run db:test-set-state-status-by-id db (db:test-get-id test) "KILLREQ" "n/a" #f) (thread-sleep! 1))) ;; NOTE: This is suboptimal as the testdata will be used later and the state/status may have changed ... (if (null? tal) (loop new-test-dat tal) (loop (car tal)(append tal (list new-test-dat))))) (begin (cdb:remote-run db:delete-test-records db #f (db:test-get-id test)) (debug:print-info 1 "Attempting to remove " (if real-dir (conc " dir " real-dir " and ") "") " link " run-dir) (if (and real-dir (> (string-length real-dir) 5) (file-exists? real-dir)) ;; bad heuristic but should prevent /tmp /home etc. (begin ;; let* ((realpath (resolve-pathname run-dir))) (debug:print-info 1 "Recursively removing " real-dir) (if (file-exists? real-dir) (if (> (system (conc "rm -rf " real-dir)) 0) (debug:print 0 "ERROR: There was a problem removing " real-dir " with rm -f")) (debug:print 0 "WARNING: test dir " real-dir " appears to not exist or is not readable"))) (if real-dir (debug:print 0 "WARNING: directory " real-dir " does not exist") (debug:print 0 "WARNING: no real directory corrosponding to link " run-dir ", nothing done"))) (if (symbolic-link? run-dir) (begin (debug:print-info 1 "Removing symlink " run-dir) (handle-exceptions exn (debug:print 0 "ERROR: Failed to remove symlink " run-dir ((condition-property-accessor 'exn 'message) exn) ", attempting to continue") (delete-file run-dir))) (if (directory? run-dir) (if (> (directory-fold (lambda (f x)(+ 1 x)) 0 run-dir) 0) (debug:print 0 "WARNING: refusing to remove " run-dir " as it is not empty") (handle-exceptions exn (debug:print 0 "ERROR: Failed to remove directory " run-dir ((condition-property-accessor 'exn 'message) exn) ", attempting to continue") (delete-directory run-dir))) (if run-dir (debug:print 0 "WARNING: not removing " run-dir " as it either doesn't exist or is not a symlink") (debug:print 0 "NOTE: the run dir for this test is undefined. Test may have already been deleted.")) )) (if (not (null? tal)) (loop (car tal)(cdr tal)))))) ((set-state-status) (debug:print-info 2 "new state " (car state-status) ", new status " (cadr state-status)) (cdb:remote-run db:test-set-state-status-by-id db (db:test-get-id test) (car state-status)(cadr state-status) #f))))) ))) ;; remove the run if zero tests remain (if (eq? action 'remove-runs) (let ((remtests (cdb:remote-run db:get-tests-for-run db (db:get-value-by-header run header "id") #f '("DELETED") '("n/a") not-in: #t))) (if (null? remtests) ;; no more tests remaining (let* ((dparts (string-split lasttpath "/")) (runpath (conc "/" (string-intersperse (take dparts (- (length dparts) 1)) "/")))) (debug:print 1 "Removing run: " runkey " " (db:get-value-by-header run header "runname") " and related record") (cdb:remote-run db:delete-run db run-id) ;; This is a pretty good place to purge old DELETED tests (cdb:remote-run db:delete-tests-for-run db run-id) (cdb:remote-run db:delete-old-deleted-test-records db) (cdb:remote-run db:set-var db "DELETED_TESTS" (current-seconds)) ;; need to figure out the path to the run dir and remove it if empty ;; (if (null? (glob (conc runpath "/*"))) ;; (begin ;; (debug:print 1 "Removing run dir " runpath) ;; (system (conc "rmdir -p " runpath)))) ))))) )) |
︙ | ︙ | |||
693 694 695 696 697 698 699 | (target (or (args:get-arg "-reqtarg") (args:get-arg "-target")))) (if (not (setup-for-run)) (begin (debug:print 0 "Failed to setup, exiting") (exit 1))) ;; (if (args:get-arg "-server") | | | 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 | (target (or (args:get-arg "-reqtarg") (args:get-arg "-target")))) (if (not (setup-for-run)) (begin (debug:print 0 "Failed to setup, exiting") (exit 1))) ;; (if (args:get-arg "-server") ;; (cdb:remote-run server:start db (args:get-arg "-server"))) (set! keys (keys:config-get-fields *configdat*)) ;; have enough to process -target or -reqtarg here (if (args:get-arg "-reqtarg") (let* ((runconfigf (conc *toppath* "/runconfigs.config")) ;; DO NOT EVALUATE ALL (runconfig (read-config runconfigf #f #t environ-patt: #f))) (if (hash-table-ref/default runconfig (args:get-arg "-reqtarg") #f) (keys:target-set-args keys (args:get-arg "-reqtarg") args:arg-hash) |
︙ | ︙ | |||
725 726 727 728 729 730 731 | ;;====================================================================== ;; Lock/unlock runs ;;====================================================================== (define (runs:handle-locking target keys runname lock unlock user) (let* ((db #f) | | | | 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 | ;;====================================================================== ;; Lock/unlock runs ;;====================================================================== (define (runs:handle-locking target keys runname lock unlock user) (let* ((db #f) (rundat (cdb:remote-run runs:get-runs-by-patt db keys runname target)) (header (vector-ref rundat 0)) (runs (vector-ref rundat 1))) (for-each (lambda (run) (let ((run-id (db:get-value-by-header run header "id"))) (if (or lock (and unlock (begin (print "Do you really wish to unlock run " run-id "?\n y/n: ") (equal? "y" (read-line))))) (cdb:remote-run db:lock/unlock-run db run-id lock unlock user) (debug:print-info 0 "Skipping lock/unlock on " run-id)))) runs))) ;;====================================================================== ;; Rollup runs ;;====================================================================== ;; Update the test_meta table for this test |
︙ | ︙ | |||
771 772 773 774 775 776 777 | (for-each (lambda (test-name) (let* ((test-path (conc *toppath* "/tests/" test-name)) (test-configf (conc test-path "/testconfig")) (testexists (and (file-exists? test-configf)(file-read-access? test-configf))) ;; read configs with tricks turned off (i.e. no system) (test-conf (if testexists (read-config test-configf #f #f)(make-hash-table)))) | | | | | | | | | 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 | (for-each (lambda (test-name) (let* ((test-path (conc *toppath* "/tests/" test-name)) (test-configf (conc test-path "/testconfig")) (testexists (and (file-exists? test-configf)(file-read-access? test-configf))) ;; read configs with tricks turned off (i.e. no system) (test-conf (if testexists (read-config test-configf #f #f)(make-hash-table)))) ;; use the cdb:remote-run instead of passing in db (runs:update-test_meta test-name test-conf))) test-names))) ;; This could probably be refactored into one complex query ... (define (runs:rollup-run keys runname user keyvals) (debug:print 4 "runs:rollup-run, keys: " keys " :runname " runname " user: " user) (let* ((db #f) (new-run-id (cdb:remote-run db:register-run #f keyvals runname "new" "n/a" user)) (prev-tests (cdb:remote-run test:get-matching-previous-test-run-records db new-run-id "%" "%")) (curr-tests (cdb:remote-run db:get-tests-for-run db new-run-id "%/%" '() '())) (curr-tests-hash (make-hash-table))) (cdb:remote-run db:update-run-event_time db new-run-id) ;; index the already saved tests by testname and itemdat in curr-tests-hash (for-each (lambda (testdat) (let* ((testname (db:test-get-testname testdat)) (item-path (db:test-get-item-path testdat)) (full-name (conc testname "/" item-path))) (hash-table-set! curr-tests-hash full-name testdat))) curr-tests) ;; NOPE: Non-optimal approach. Try this instead. ;; 1. tests are received in a list, most recent first ;; 2. replace the rollup test with the new *always* (for-each (lambda (testdat) (let* ((testname (db:test-get-testname testdat)) (item-path (db:test-get-item-path testdat)) (full-name (conc testname "/" item-path)) (prev-test-dat (hash-table-ref/default curr-tests-hash full-name #f)) (test-steps (cdb:remote-run db:get-steps-for-test db (db:test-get-id testdat))) (new-test-record #f)) ;; replace these with insert ... select (apply sqlite3:execute db (conc "INSERT OR REPLACE INTO tests (run_id,testname,state,status,event_time,host,cpuload,diskfree,uname,rundir,item_path,run_duration,final_logf,comment) " "VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?,?);") new-run-id (cddr (vector->list testdat))) (set! new-testdat (car (cdb:remote-run db:get-tests-for-run db new-run-id (conc testname "/" item-path) '() '()))) (hash-table-set! curr-tests-hash full-name new-testdat) ;; this could be confusing, which record should go into the lookup table? ;; Now duplicate the test steps (debug:print 4 "Copying records in test_steps from test_id=" (db:test-get-id testdat) " to " (db:test-get-id new-testdat)) (cdb:remote-run (lambda () (sqlite3:execute db (conc "INSERT OR REPLACE INTO test_steps (test_id,stepname,state,status,event_time,comment) " "SELECT " (db:test-get-id new-testdat) ",stepname,state,status,event_time,comment FROM test_steps WHERE test_id=?;") (db:test-get-id testdat)) ;; Now duplicate the test data |
︙ | ︙ |
Modified tests.scm from [b1e14ea02f] to [8111ef0037].
︙ | ︙ | |||
537 538 539 540 541 542 543 | ;;====================================================================== ;; test steps ;;====================================================================== ;; teststep-set-status! used to be here (define (test-get-kill-request test-id) ;; run-id test-name itemdat) | < | > | | 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 | ;;====================================================================== ;; test steps ;;====================================================================== ;; teststep-set-status! used to be here (define (test-get-kill-request test-id) ;; run-id test-name itemdat) (let* ((testdat (cdb:get-test-info-by-id *runremote* test-id))) ;; run-id test-name item-path))) (and testdat (equal? (test:get-state testdat) "KILLREQ")))) (define (test:tdb-get-rundat-count tdb) (if tdb (let ((res 0)) (sqlite3:for-each-row (lambda (count) (set! res count)) |
︙ | ︙ |