Megatest

Check-in [5526a18984]
Login
Overview
Comment:Fixed server kill, reduced noise levels
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | v1.60
Files: files | file ages | folders
SHA1: 5526a18984498a810bef7972971256a6a163c6c9
User & Date: mrwellan on 2014-10-09 17:09:08
Other Links: branch diff | manifest | tags
Context
2014-10-10
10:49
bumped version check-in: 69da7e37ab user: mrwellan tags: v1.60
2014-10-09
17:09
Fixed server kill, reduced noise levels check-in: 5526a18984 user: mrwellan tags: v1.60
00:03
Merged in refactor-sqlite3-calls. Can refactor in small steps from here on out check-in: c03075b49a user: matt tags: v1.60
Changes

Modified api.scm from [4a3da8fcef] to [859483d7bd].

48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
    testmeta-get-record))

;; These are called by the server on recipt of /api calls

(define (api:execute-requests dbstruct cmd params)
  (case (string->symbol cmd)
    ;; SERVERS
    ((start-server)                (apply server:kind-run params))
    ;; ((kill-server)
    ;;  (db:sync-tables (db:tbls *inmemdb*) *inmemdb* *db*)  ;; (db:sync-to *inmemdb* *db*)
    ;;  (let ((hostname (car  *runremote*))
    ;;        (port     (cadr *runremote*))
    ;;        (pid      (if (null? params) #f (car params)))
    ;;        (th1      (make-thread (lambda ()(thread-sleep! 3)(debug:print 0 "Server exiting!")(exit 0)) "Server exit thread")))
    ;;    (debug:print 0 "WARNING: Server on " hostname ":" port " going down by user request!")
    ;;    (debug:print-info 1 "current pid=" (current-process-id))
    ;;    (open-run-close tasks:server-deregister tasks:open-db 
    ;;     	       hostname
    ;;     	       port: port)
    ;;    (set! *server-run* #f)
    ;;    (thread-sleep! 3)
    ;;    (if pid 
    ;;        (process-signal pid signal/kill)
    ;;        (thread-start! th1))
    ;;    '(#t "exit process started")))

    ;; KEYS
    ((get-key-val-pairs)            (apply db:get-key-val-pairs dbstruct params))
    ((get-keys)                     (db:get-keys dbstruct))

    ;; TESTS
    ((test-toplevel-num-items)         (apply db:test-toplevel-num-items dbstruct params))







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







48
49
50
51
52
53
54
55











56





57
58
59
60
61
62
63
    testmeta-get-record))

;; These are called by the server on recipt of /api calls

(define (api:execute-requests dbstruct cmd params)
  (case (string->symbol cmd)
    ;; SERVERS
    ((start-server)                 (apply server:kind-run params))











    ((kill-server)                  (set! *server-run* #f))






    ;; KEYS
    ((get-key-val-pairs)            (apply db:get-key-val-pairs dbstruct params))
    ((get-keys)                     (db:get-keys dbstruct))

    ;; TESTS
    ((test-toplevel-num-items)         (apply db:test-toplevel-num-items dbstruct params))
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
    ((get-runs)                     (apply db:get-runs dbstruct params))
    ((get-all-run-ids)              (db:get-all-run-ids dbstruct))
    ((get-prev-run-ids)             (apply db:get-prev-run-ids dbstruct params))
    ((get-run-ids-matching-target)  (apply db:get-run-ids-matching-target dbstruct params))
    ((get-runs-by-patt)             (apply db:get-runs-by-patt dbstruct params))
    ((lock/unlock-run)              (apply db:lock/unlock-run dbstruct params))
    ((update-run-event_time)        (apply db:update-run-event_time dbstruct params))
    ((find-and-mark-incompete       (apply db:find-and-mark-incomplete dbstruct (car params) ovr-deadtime: (cadr params))))

    ;; STEPS
    ((teststep-set-status!)         (apply db:teststep-set-status! dbstruct params))

    ;; TEST DATA
    ((test-data-rollup)             (apply db:test-data-rollup dbstruct params))
    ((csv->test-data)               (apply db:csv->test-data dbstruct params))







|







95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
    ((get-runs)                     (apply db:get-runs dbstruct params))
    ((get-all-run-ids)              (db:get-all-run-ids dbstruct))
    ((get-prev-run-ids)             (apply db:get-prev-run-ids dbstruct params))
    ((get-run-ids-matching-target)  (apply db:get-run-ids-matching-target dbstruct params))
    ((get-runs-by-patt)             (apply db:get-runs-by-patt dbstruct params))
    ((lock/unlock-run)              (apply db:lock/unlock-run dbstruct params))
    ((update-run-event_time)        (apply db:update-run-event_time dbstruct params))
    ((find-and-mark-incomplete)     (apply db:find-and-mark-incomplete dbstruct params))

    ;; STEPS
    ((teststep-set-status!)         (apply db:teststep-set-status! dbstruct params))

    ;; TEST DATA
    ((test-data-rollup)             (apply db:test-data-rollup dbstruct params))
    ((csv->test-data)               (apply db:csv->test-data dbstruct params))

Modified db.scm from [8fe161caf5] to [7eb26e4c3a].

419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
	   ;; read the source table
	   (sqlite3:for-each-row
	    (lambda (a . b)
	      (set! fromdat (cons (apply vector a b) fromdat)))
	    fromdb
	    full-sel)

	   (debug:print 0 "INFO: found " (length fromdat) " records to sync")

	   ;; read the target table
	   (sqlite3:for-each-row
	    (lambda (a . b)
	      (hash-table-set! todat a (apply vector a b)))
	    todb
	    full-sel)







|







419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
	   ;; read the source table
	   (sqlite3:for-each-row
	    (lambda (a . b)
	      (set! fromdat (cons (apply vector a b) fromdat)))
	    fromdb
	    full-sel)

	   (debug:print-info 2 "found " (length fromdat) " records to sync")

	   ;; read the target table
	   (sqlite3:for-each-row
	    (lambda (a . b)
	      (hash-table-set! todat a (apply vector a b)))
	    todb
	    full-sel)
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
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
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
;;======================================================================

;;  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 run-id  #!key (ovr-deadtime #f))

  (let* ((incompleted '())
	 (oldlaunched '())
	 (toplevels   '())
	 (deadtime-str (configf:lookup *configdat* "setup" "deadtime"))
	 (deadtime     (if (and deadtime-str
				(string->number deadtime-str))
			   (string->number deadtime-str)
			   7200))) ;; two hours
    (if (number? ovr-deadtime)(set! deadtime ovr-deadtime))
    
    ;; in RUNNING or REMOTEHOSTSTART for more than 10 minutes
    ;;
    ;; THIS CANNOT WORK. The run_duration is not updated in the central db due to performance concerns.
    ;;                   The testdat.db file must be consulted.
    ;;
    ;; HOWEVER: this code in run:test seems to work fine
    ;;              (> (- (current-seconds)(+ (db:test-get-event_time testdat)
    ;;                     (db:test-get-run_duration testdat)))
    ;;                    600) 
    ;; (db:delay-if-busy)
    (sqlite3:for-each-row 
     (lambda (test-id run-dir uname testname item-path)
       (if (and (equal? uname "n/a")
		(equal? item-path "")) ;; this is a toplevel test
	   ;; what to do with toplevel? call rollup?
	   (begin
	     (set! toplevels   (cons (list test-id run-dir uname testname item-path run-id) toplevels))
	     (debug:print-info 0 "Found old toplevel test in RUNNING state, test-id=" test-id))
	   (set! incompleted (cons (list test-id run-dir uname testname item-path run-id) incompleted))))
     db
     "SELECT id,rundir,uname,testname,item_path FROM tests WHERE run_id=? AND (strftime('%s','now') - event_time) > 600 AND state IN ('RUNNING','REMOTEHOSTSTART');"
     run-id)
    
    ;; in LAUNCHED for more than one day. Could be long due to job queues TODO/BUG: Need override for this in config
    ;;
    ;; (db:delay-if-busy)
    (sqlite3:for-each-row
     (lambda (test-id run-dir uname testname item-path)
       (if (and (equal? uname "n/a")
		(equal? item-path "")) ;; this is a toplevel test
	   ;; what to do with toplevel? call rollup?
	   (set! toplevels   (cons (list test-id run-dir uname testname item-path run-id) toplevels))
	   (set! oldlaunched (cons (list test-id run-dir uname testname item-path run-id) oldlaunched))))
     db
     "SELECT id,rundir,uname,testname,item_path FROM tests WHERE run_id=? AND (strftime('%s','now') - event_time) > 86400 AND state IN ('LAUNCHED');"
     run-id)
    


    ;; These are defunct tests, do not do all the overhead of set-state-status. Force them to INCOMPLETE.
    ;;
    ;; (db:delay-if-busy)
    (let* ((min-incompleted (filter (lambda (x)
				      (let* ((testpath (cadr x))
					     (tdatpath (conc testpath "/testdat.db"))
					     (dbexists (file-exists? tdatpath)))
					(or (not dbexists) ;; if no file then something wrong - mark as incomplete
					    (> (- (current-seconds)(file-modification-time tdatpath)) 600)))) ;; no change in 10 minutes to testdat.db - she's dead Jim
				    incompleted))
	   (min-incompleted-ids (map car min-incompleted))
	   (all-ids             (append min-incompleted-ids (map car oldlaunched))))
      (if (> (length all-ids) 0)
	  (begin
	    (debug:print 0 "WARNING: Marking test(s); " (string-intersperse (map conc all-ids) ", ") " as INCOMPLETE")
	    (sqlite3:execute 
	     db
	     (conc "UPDATE tests SET state='INCOMPLETE' WHERE id IN (" 
		   (string-intersperse (map conc all-ids) ",")
		   ");")))))

    ;; Now do rollups for the toplevel tests
    ;;
    (for-each
     (lambda (toptest)
       (let ((test-name (list-ref toptest 3))
	     (run-id    (list-ref toptest 5)))
	 (cdb:top-test-set-per-pf-counts *runremote* run-id test-name)))
     toplevels)))
		     
;; Clean out old junk and vacuum the database
;;
;; Ultimately do something like this:
;;
;; 1. Look at test records either deleted or part of deleted run:







|
>
|











<
<
<















|
|
|














>
>



|
|
|
|
|
|
|
|














|
|
|







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
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
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
;;======================================================================

;;  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 dbstruct run-id ovr-deadtime)
  (let* ((db          (db:get-db dbstruct run-id))
	 (incompleted '())
	 (oldlaunched '())
	 (toplevels   '())
	 (deadtime-str (configf:lookup *configdat* "setup" "deadtime"))
	 (deadtime     (if (and deadtime-str
				(string->number deadtime-str))
			   (string->number deadtime-str)
			   7200))) ;; two hours
    (if (number? ovr-deadtime)(set! deadtime ovr-deadtime))
    
    ;; in RUNNING or REMOTEHOSTSTART for more than 10 minutes
    ;;



    ;; HOWEVER: this code in run:test seems to work fine
    ;;              (> (- (current-seconds)(+ (db:test-get-event_time testdat)
    ;;                     (db:test-get-run_duration testdat)))
    ;;                    600) 
    ;; (db:delay-if-busy)
    (sqlite3:for-each-row 
     (lambda (test-id run-dir uname testname item-path)
       (if (and (equal? uname "n/a")
		(equal? item-path "")) ;; this is a toplevel test
	   ;; what to do with toplevel? call rollup?
	   (begin
	     (set! toplevels   (cons (list test-id run-dir uname testname item-path run-id) toplevels))
	     (debug:print-info 0 "Found old toplevel test in RUNNING state, test-id=" test-id))
	   (set! incompleted (cons (list test-id run-dir uname testname item-path run-id) incompleted))))
     db
     "SELECT id,rundir,uname,testname,item_path 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
    ;;
    ;; (db:delay-if-busy)
    (sqlite3:for-each-row
     (lambda (test-id run-dir uname testname item-path)
       (if (and (equal? uname "n/a")
		(equal? item-path "")) ;; this is a toplevel test
	   ;; what to do with toplevel? call rollup?
	   (set! toplevels   (cons (list test-id run-dir uname testname item-path run-id) toplevels))
	   (set! oldlaunched (cons (list test-id run-dir uname testname item-path run-id) oldlaunched))))
     db
     "SELECT id,rundir,uname,testname,item_path FROM tests WHERE run_id=? AND (strftime('%s','now') - event_time) > 86400 AND state IN ('LAUNCHED');"
     run-id)
    
    (debug:print-info 18 "Found " (length oldlaunched) " old LAUNCHED items, " (length toplevels) " old LAUNCHED toplevel tests and " (length incompleted) " tests marked RUNNING but apparently dead.")

    ;; These are defunct tests, do not do all the overhead of set-state-status. Force them to INCOMPLETE.
    ;;
    ;; (db:delay-if-busy)
    (let* (;; (min-incompleted (filter (lambda (x)
	   ;;      		      (let* ((testpath (cadr x))
	   ;;      			     (tdatpath (conc testpath "/testdat.db"))
	   ;;      			     (dbexists (file-exists? tdatpath)))
	   ;;      			(or (not dbexists) ;; if no file then something wrong - mark as incomplete
	   ;;      			    (> (- (current-seconds)(file-modification-time tdatpath)) 600)))) ;; no change in 10 minutes to testdat.db - she's dead Jim
	   ;;      		    incompleted))
	   (min-incompleted-ids (map car incompleted)) ;; do 'em all
	   (all-ids             (append min-incompleted-ids (map car oldlaunched))))
      (if (> (length all-ids) 0)
	  (begin
	    (debug:print 0 "WARNING: Marking test(s); " (string-intersperse (map conc all-ids) ", ") " as INCOMPLETE")
	    (sqlite3:execute 
	     db
	     (conc "UPDATE tests SET state='INCOMPLETE' WHERE id IN (" 
		   (string-intersperse (map conc all-ids) ",")
		   ");")))))

    ;; Now do rollups for the toplevel tests
    ;;
    (for-each
     (lambda (toptest)
       (let ((test-name (list-ref toptest 3)))
;;	     (run-id    (list-ref toptest 5)))
	  (db:general-call db 'top-test-set-per-pf-counts (list test-name run-id test-name test-name test-name)))) ;; (list run-id test-name))))
     toplevels)))
		     
;; Clean out old junk and vacuum the database
;;
;; Ultimately do something like this:
;;
;; 1. Look at test records either deleted or part of deleted run:

Modified http-transport.scm from [7ee6223861] to [6063d55ba0].

373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
      (let ((start-time (current-milliseconds))
	    (sync-time  #f)
	    (rem-time   #f))

	(if *inmemdb* (db:sync-touched *inmemdb* *run-id* force-sync: #t))
	(set! sync-time  (- (current-milliseconds) start-time))
	(set! rem-time (quotient (- 4000 sync-time) 1000))
	(debug:print 0 "SYNC: time= " sync-time ", rem-time=" rem-time)

      ;;
      ;; set_running after our first pass through and start the db
      ;;
      (if (eq? server-state 'available)
	  (begin
	    (tasks:server-set-state! tdb server-id "dbprep")







|







373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
      (let ((start-time (current-milliseconds))
	    (sync-time  #f)
	    (rem-time   #f))

	(if *inmemdb* (db:sync-touched *inmemdb* *run-id* force-sync: #t))
	(set! sync-time  (- (current-milliseconds) start-time))
	(set! rem-time (quotient (- 4000 sync-time) 1000))
	(debug:print 2 "SYNC: time= " sync-time ", rem-time=" rem-time)

      ;;
      ;; set_running after our first pass through and start the db
      ;;
      (if (eq? server-state 'available)
	  (begin
	    (tasks:server-set-state! tdb server-id "dbprep")

Modified rmt.scm from [d4f16a53b3] to [7baaea28d7].

81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
		(rmt:send-receive cmd run-id params))))
	(let ((max-avg-qry (string->number (or (configf:lookup *configdat* "server" "server-query-threshold") "800"))))
	  (debug:print-info 4 "no server and read-only query, bypassing normal channel")
	  ;; (if (rmt:write-frequency-over-limit? cmd run-id)(server:kind-run run-id))
	  (let ((curr-max (rmt:get-max-query-average)))
	    (if (> (cdr curr-max) max-avg-qry)
		(begin
		  (debug:print-info 0 "Max average query, " (inexact->exact (round (cdr curr-max))) "ms (" (car curr-max) ") exceeds " max-avg-qry ", try starting server ...")
		  (server:kind-run run-id))))
	  (rmt:open-qry-close-locally cmd run-id params)))))

(define (rmt:update-db-stats rawcmd params duration)
  (mutex-lock! *db-stats-mutex*)
  (let* ((cmd      (if (eq? rawcmd 'general-call) (car params) rawcmd))
	 (stat-vec (hash-table-ref/default *db-stats* cmd #f)))







|







81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
		(rmt:send-receive cmd run-id params))))
	(let ((max-avg-qry (string->number (or (configf:lookup *configdat* "server" "server-query-threshold") "800"))))
	  (debug:print-info 4 "no server and read-only query, bypassing normal channel")
	  ;; (if (rmt:write-frequency-over-limit? cmd run-id)(server:kind-run run-id))
	  (let ((curr-max (rmt:get-max-query-average)))
	    (if (> (cdr curr-max) max-avg-qry)
		(begin
		  (debug:print-info 3 "Max average query, " (inexact->exact (round (cdr curr-max))) "ms (" (car curr-max) ") exceeds " max-avg-qry ", try starting server ...")
		  (server:kind-run run-id))))
	  (rmt:open-qry-close-locally cmd run-id params)))))

(define (rmt:update-db-stats rawcmd params duration)
  (mutex-lock! *db-stats-mutex*)
  (let* ((cmd      (if (eq? rawcmd 'general-call) (car params) rawcmd))
	 (stat-vec (hash-table-ref/default *db-stats* cmd #f)))
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

(define (rmt:update-run-event_time run-id)
  (rmt:send-receive 'update-run-event_time #f (list run-id)))

(define (rmt:get-runs-by-patt  keys runnamepatt targpatt offset limit)
  (rmt:send-receive 'get-runs-by-patt #f (list keys runnamepatt targpatt offset limit)))

(define (rmt:find-and-mark-incomplete run-id #!key (ovr-deadtime #f))
  (rmt:send-receive 'find-and-mark-incomplete #f (list run-id ovr-deadtime)))

;;======================================================================
;; M U L T I R U N   Q U E R I E S
;;======================================================================

;; Need to move this to multi-run section and make associated changes
(define (rmt:find-and-mark-incomplete-all-runs #!key (ovr-deadtime #f))
  (let ((run-ids (rmt:get-all-run-ids)))
    (for-each (lambda (run-id)
	       (rmt:find-and-mark-incomplete run-id ovr-deadtime: ovr-deadtime))
	     run-ids)))

;; get the previous record for when this test was run where all keys match but runname
;; returns #f if no such test found, returns a single test record if found
;; 
;; Run this at the client end since we have to connect to multiple run-id dbs
;;







|
|









|







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

(define (rmt:update-run-event_time run-id)
  (rmt:send-receive 'update-run-event_time #f (list run-id)))

(define (rmt:get-runs-by-patt  keys runnamepatt targpatt offset limit)
  (rmt:send-receive 'get-runs-by-patt #f (list keys runnamepatt targpatt offset limit)))

(define (rmt:find-and-mark-incomplete run-id ovr-deadtime)
  (rmt:send-receive 'find-and-mark-incomplete run-id (list run-id ovr-deadtime)))

;;======================================================================
;; M U L T I R U N   Q U E R I E S
;;======================================================================

;; Need to move this to multi-run section and make associated changes
(define (rmt:find-and-mark-incomplete-all-runs #!key (ovr-deadtime #f))
  (let ((run-ids (rmt:get-all-run-ids)))
    (for-each (lambda (run-id)
	       (rmt:find-and-mark-incomplete run-id ovr-deadtime))
	     run-ids)))

;; get the previous record for when this test was run where all keys match but runname
;; returns #f if no such test found, returns a single test record if found
;; 
;; Run this at the client end since we have to connect to multiple run-id dbs
;;