Overview
Comment: | Split find and mark incomplete to iterate over runs, fixed logic bug in select statements for find and mark. |
---|---|
Downloads: | Tarball | ZIP archive | SQL archive |
Timelines: | family | ancestors | descendants | both | v1.55 |
Files: | files | file ages | folders |
SHA1: |
7bb000c05a4ba70ed5dd8ea2714420c0 |
User & Date: | matt on 2013-10-22 23:10:55 |
Other Links: | branch diff | manifest | tags |
Context
2013-10-22
| ||
23:51 | Upped the timeout on marking INCOMPLETE in the fullrun test check-in: 5b7a69fb1d user: matt tags: v1.55 | |
23:10 | Split find and mark incomplete to iterate over runs, fixed logic bug in select statements for find and mark. check-in: 7bb000c05a user: matt tags: v1.55 | |
2013-10-21
| ||
20:25 | Added ability to control deadtime from config. Set default deadtime to 1800 seconds check-in: aba5b97c7d user: mrwellan tags: v1.55 | |
Changes
Modified db.scm from [a74f9a32c4] to [0e0b9ca4ad].
︙ | ︙ | |||
462 463 464 465 466 467 468 | ((< mver megatest-version) (db:set-var db "MEGATEST_VERSION" megatest-version)))))) ;;====================================================================== ;; M A I N T E N A N C E ;;====================================================================== | > > > > > > | > > > > > | > > | | | | | > | > > | | | | | > > | > | 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 | ((< mver megatest-version) (db:set-var db "MEGATEST_VERSION" megatest-version)))))) ;;====================================================================== ;; M A I N T E N A N C E ;;====================================================================== ;; select end_time-now from ;; (select testname,item_path,event_time+run_duration as ;; end_time,strftime('%s','now') as now from tests where state in ;; ('RUNNING','REMOTEHOSTSTART','LAUNCED')); (define (db:find-and-mark-incomplete db #!key (ovr-deadtime #f)) (let* ((incompleted '()) (deadtime-str (configf:lookup *configdat* "setup" "deadtime")) (deadtime (if (and deadtime-str (string->number deadtime-str)) (string->number deadtime-str) 7200)) ;; two hours (run-ids (db:get-run-ids db))) ;; iterate over runs to divy up the calls (if (number? ovr-deadtime)(set! deadtime ovr-deadtime)) (for-each (lambda (run-id) ;; in RUNNING or REMOTEHOSTSTART for more than 10 minutes ;; (sqlite3:for-each-row (lambda (test-id) (set! incompleted (cons test-id incompleted))) db "SELECT id FROM tests WHERE run_id=? AND (strftime('%s','now') - event_time - run_duration) > ? AND state IN ('RUNNING','REMOTEHOSTSTART');" run-id deadtime) ;; in LAUNCHED for more than one day. Could be long due to job queues TODO/BUG: Need override for this in config ;; (sqlite3:for-each-row (lambda (test-id) (set! incompleted (cons test-id incompleted))) db "SELECT id FROM tests WHERE run_id=? AND (strftime('%s','now') - event_time - run_duration) > ? AND state IN ('LAUNCHED');" run-id (* 60 60 24))) run-ids) ;; These are defunct tests, do not do all the overhead of set-state-status. Force them to INCOMPLETE. ;; (if (> (length incompleted) 0) (begin (debug:print 0 "WARNING: Marking test(s); " (string-intersperse (map conc incompleted) ", ") " as INCOMPLETE") (sqlite3:execute db (conc "UPDATE tests SET state='INCOMPLETE' WHERE id IN (" (string-intersperse (map conc incompleted) ",") |
︙ | ︙ | |||
875 876 877 878 879 880 881 882 883 884 885 886 887 888 | "unlocked" "locked")))) ;; semi-failsafe (sqlite3:execute db "UPDATE runs SET state=? WHERE id=?;" newlockval run-id) (sqlite3:execute db "INSERT INTO access_log (user,accessed,args) VALUES(?,strftime('%s','now'),?);" user (conc newlockval " " run-id)) (debug:print-info 1 "" newlockval " run number " run-id))) ;;====================================================================== ;; K E Y S ;;====================================================================== ;; get key val pairs for a given run-id ;; ( (FIELDNAME1 keyval1) (FIELDNAME2 keyval2) ... ) (define (db:get-key-val-pairs db run-id) | > > > > > > > > | 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 | "unlocked" "locked")))) ;; semi-failsafe (sqlite3:execute db "UPDATE runs SET state=? WHERE id=?;" newlockval run-id) (sqlite3:execute db "INSERT INTO access_log (user,accessed,args) VALUES(?,strftime('%s','now'),?);" user (conc newlockval " " run-id)) (debug:print-info 1 "" newlockval " run number " run-id))) (define (db:get-run-ids db) (let ((res '())) (sqlite3:for-each-row (lambda (id) (set! res (cons id res))) db "SELECT id FROM runs;"))) ;;====================================================================== ;; K E Y S ;;====================================================================== ;; get key val pairs for a given run-id ;; ( (FIELDNAME1 keyval1) (FIELDNAME2 keyval2) ... ) (define (db:get-key-val-pairs db run-id) |
︙ | ︙ |
Modified runs.scm from [31ee3d42e6] to [e64acf4e6f].
︙ | ︙ | |||
721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 | )))))))) ;; 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 all-tests-registry) ;; 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)) (tests-info (mt:get-tests-for-run run-id #f '() '())) ;; qryvals: "id,testname,item_path")) (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))) ;; length of the register queue ahead (reglen (if (number? reglen-in) reglen-in 1)) | > > > > > | | | | 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 | )))))))) ;; 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 all-tests-registry) ;; 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)) ;; Do mark-and-find clean up of db before starting runing of quue ;; (cdb:remote-run db:find-and-mark-incomplete #f) (let ((run-info (cdb:remote-run db:get-run-info #f run-id)) (tests-info (mt:get-tests-for-run run-id #f '() '())) ;; qryvals: "id,testname,item_path")) (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))) ;; length of the register queue ahead (reglen (if (number? reglen-in) reglen-in 1)) (last-time-incomplete (current-seconds))) ;; Initialize the test-registery hash with tests that already have a record ;; convert state to symbol and use that as the hash value (for-each (lambda (trec) (let ((id (db:test-get-id trec)) (tn (db:test-get-testname trec)) (ip (db:test-get-item-path trec)) (st (db:test-get-state trec))) (if (not (equal? st "DELETED")) (hash-table-set! test-registry (runs:make-full-test-name tn ip) (string->symbol st))))) tests-info) (set! max-retries (if (and max-retries (string->number max-retries))(string->number max-retries) 100)) (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)) ;; 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)) (cdb:remote-run db:find-and-mark-incomplete #f))) ;; (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)) |
︙ | ︙ |