Megatest

Check-in [6bffff1168]
Login
Overview
Comment:Fixed wrong settings control for server load gate. Fixed logic error on detecting defunct server starts
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | v1.64 | v1.6427
Files: files | file ages | folders
SHA1: 6bffff11684d284de8b2ea4a71fdaa93ec68e344
User & Date: mrwellan on 2017-07-26 16:34:53
Other Links: branch diff | manifest | tags
Context
2017-07-26
17:31
Fixed missing begin check-in: ec2dd2547c user: mrwellan tags: v1.64, v1.6427
16:34
Fixed wrong settings control for server load gate. Fixed logic error on detecting defunct server starts check-in: 6bffff1168 user: mrwellan tags: v1.64, v1.6427
2017-07-25
14:43
Fixed emergency loaded to not put out noise on file not exists check-in: 7d1e789cb3 user: mrwellan tags: v1.64
Changes

Modified http-transport.scm from [09510faceb] to [3650524f6a].

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
                                      (system (conc "sleep 1;mv -f " full-serv-fname " " new-serv-fname))
                                      (debug:print 0 *default-log-port* "INFO: cannot move " full-serv-fname " to " new-serv-fname))
                                  (exit)))))
    (if (and (not start-time-old) ;; last server start try was less than five seconds ago
	     (not server-starting))
	(begin
	  (cleanup-proc "NOT starting server, there is either a recently started server or a server in process of starting")
	  (exit))))
  ;; lets not even bother to start if there are already three or more server files ready to go
  (let* ((num-alive   (server:get-num-alive (server:get-list *toppath*))))
    (if (> num-alive 3)
        (cleanup-proc "ERROR: Aborting server start because there are already " num-alive " possible servers either running or starting up")))

  (let* ((th2 (make-thread (lambda ()
			     (debug:print-info 0 *default-log-port* "Server run thread started")
			     (http-transport:run 
			      (if (args:get-arg "-server")
				  (args:get-arg "-server")
				  "-")
			      )) "Server run"))
	 (th3 (make-thread (lambda ()
			     (debug:print-info 0 *default-log-port* "Server monitor thread started")
			     (http-transport:keep-running)
			   "Keep running"))))
    (thread-start! th2)
    (thread-sleep! 0.25) ;; give the server time to settle before starting the keep-running monitor.
    (thread-start! th3)
    (set! *didsomething* #t)
    (thread-join! th2)
    (exit)))

;; (define (http-transport:server-signal-handler signum)
;;   (signal-mask! signum)
;;   (handle-exceptions
;;    exn
;;    (debug:print 0 *default-log-port* " ... exiting ...")
;;    (let ((th1 (make-thread (lambda ()







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







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
                                      (system (conc "sleep 1;mv -f " full-serv-fname " " new-serv-fname))
                                      (debug:print 0 *default-log-port* "INFO: cannot move " full-serv-fname " to " new-serv-fname))
                                  (exit)))))
    (if (and (not start-time-old) ;; last server start try was less than five seconds ago
	     (not server-starting))
	(begin
	  (cleanup-proc "NOT starting server, there is either a recently started server or a server in process of starting")
	  (exit)))
    ;; lets not even bother to start if there are already three or more server files ready to go
    (let* ((num-alive   (server:get-num-alive (server:get-list *toppath*))))
      (if (> num-alive 3)
          (cleanup-proc (conc "ERROR: Aborting server start because there are already " num-alive " possible servers either running or starting up"))
          (exit)))
    (let* ((th2 (make-thread (lambda ()
                               (debug:print-info 0 *default-log-port* "Server run thread started")
                               (http-transport:run 
                                (if (args:get-arg "-server")
                                    (args:get-arg "-server")
                                    "-")
                                )) "Server run"))
           (th3 (make-thread (lambda ()
                               (debug:print-info 0 *default-log-port* "Server monitor thread started")
                               (http-transport:keep-running)
                               "Keep running"))))
      (thread-start! th2)
      (thread-sleep! 0.25) ;; give the server time to settle before starting the keep-running monitor.
      (thread-start! th3)
      (set! *didsomething* #t)
      (thread-join! th2)
      (exit))))

;; (define (http-transport:server-signal-handler signum)
;;   (signal-mask! signum)
;;   (handle-exceptions
;;    exn
;;    (debug:print 0 *default-log-port* " ... exiting ...")
;;    (let ((th1 (make-thread (lambda ()

Modified server.scm from [d21d7ab2e0] to [0b4350005b].

111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
	 (cmdln (conc (common:get-megatest-exe)
		      " -server " (or target-host "-") (if (equal? (configf:lookup *configdat* "server" "daemonize") "yes")
							   " -daemonize "
							   "")
		      ;; " -log " logfile
		      " -m testsuite:" testsuite)) ;; (conc " >> " logfile " 2>&1 &")))))
	 (log-rotate  (make-thread common:rotate-logs  "server run, rotate logs thread"))
         (load-limit  (configf:lookup-number *configdat* "server" "load-limit" default: 0.9)))
    ;; we want the remote server to start in *toppath* so push there
    (push-directory areapath)
    (debug:print 0 *default-log-port* "INFO: Trying to start server (" cmdln ") ...")
    (thread-start! log-rotate)
    
    ;; host.domain.tld match host?
    (if (and target-host 







|







111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
	 (cmdln (conc (common:get-megatest-exe)
		      " -server " (or target-host "-") (if (equal? (configf:lookup *configdat* "server" "daemonize") "yes")
							   " -daemonize "
							   "")
		      ;; " -log " logfile
		      " -m testsuite:" testsuite)) ;; (conc " >> " logfile " 2>&1 &")))))
	 (log-rotate  (make-thread common:rotate-logs  "server run, rotate logs thread"))
         (load-limit  (configf:lookup-number *configdat* "jobtools" "maxhomehostload" default: 3.0)))
    ;; we want the remote server to start in *toppath* so push there
    (push-directory areapath)
    (debug:print 0 *default-log-port* "INFO: Trying to start server (" cmdln ") ...")
    (thread-start! log-rotate)
    
    ;; host.domain.tld match host?
    (if (and target-host 
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
                       (common:low-noise-print 30 "sync-period"))
                  (debug:print-info 0 *default-log-port* "Increased sync period due to load: " sync-period))
	      ;; (if recently-synced (debug:print-info 0 *default-log-port* "Skipping sync due to recently-synced flag=" recently-synced))
	      ;; (debug:print-info 0 *default-log-port* "need-sync: " need-sync " sync-in-progress: " sync-in-progress " should-sync: " should-sync " will-sync: " will-sync)
	      (if will-sync (set! *db-sync-in-progress* #t))
	      (mutex-unlock! *db-multi-sync-mutex*)
	      (if will-sync

                  (let ((sync-start (current-milliseconds)))
		    (with-output-to-file start-file (lambda ()(print (current-process-id))))
		    
		    ;; put lock here
		    

                    (if (< sync-duration 3000) ;; NOTE: db:sync-to-megatest.db keeps track of time of last sync and syncs incrementally
                        (let ((res        (db:sync-to-megatest.db dbstruct no-sync-db: no-sync-db))) ;; did we sync any data? If so need to set the db touched flag to keep the server alive
                          (set! sync-duration (- (current-milliseconds) sync-start))
                          (if (> res 0) ;; some records were transferred, keep the db alive
                              (begin
                                (mutex-lock! *heartbeat-mutex*)
                                (set! *db-last-access* (current-seconds))
                                (mutex-unlock! *heartbeat-mutex*)
                                (debug:print-info 0 *default-log-port* "sync called, " res " records transferred."))
                              (debug:print-info 2 *default-log-port* "sync called but zero records transferred")))
                        ;; TODO: factor this next routine out into a function
                        (with-input-from-pipe ;; this should not block other threads but need to verify this
                         (conc "megatest -sync-to-megatest.db -m testsuite:" (common:get-area-name) ":" *toppath*)
                         (lambda ()
                           (let loop ((inl (read-line))
                                      (res #f))
                             (if (eof-object? inl)
                                 (begin
                                   (set! sync-duration (- (current-milliseconds) sync-start))
                                   (cond
                                    ((not res)
                                     (debug:print 0 *default-log-port* "ERROR: sync from /tmp db to megatest.db appears to have failed. Recommended that you stop your runs and run \"megatest -cleanup-db\""))
                                    ((> res 0)
                                     (mutex-lock! *heartbeat-mutex*)
                                     (set! *db-last-access* (current-seconds))
                                     (mutex-unlock! *heartbeat-mutex*))))
                                 (let ((num-synced (let ((matches (string-match "^Synced (\\d+).*$" inl)))
                                                     (if matches
                                                         (string->number (cadr matches))
                                                         #f))))
                                   (loop (read-line)
                                         (or num-synced res))))))))))
	      (if will-sync
		  (begin
		    (mutex-lock! *db-multi-sync-mutex*)
		    (set! *db-sync-in-progress* #f)
		    (set! *db-last-sync* start-time)
		    (with-output-to-file end-file (lambda ()(print (current-process-id))))








>
|




>
|








|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|







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
                       (common:low-noise-print 30 "sync-period"))
                  (debug:print-info 0 *default-log-port* "Increased sync period due to load: " sync-period))
	      ;; (if recently-synced (debug:print-info 0 *default-log-port* "Skipping sync due to recently-synced flag=" recently-synced))
	      ;; (debug:print-info 0 *default-log-port* "need-sync: " need-sync " sync-in-progress: " sync-in-progress " should-sync: " should-sync " will-sync: " will-sync)
	      (if will-sync (set! *db-sync-in-progress* #t))
	      (mutex-unlock! *db-multi-sync-mutex*)
	      (if will-sync
                  (let (;; (max-sync-duration  (configf:lookup-number *configdat* "server" "max-sync-duration")) ;; KEEPING THIS AVAILABLE BUT SHOULD NOT USE, I'M PRETTY SURE IT DOES NOT WORK!
                        (sync-start         (current-milliseconds)))
		    (with-output-to-file start-file (lambda ()(print (current-process-id))))
		    
		    ;; put lock here
		    
                    ;; (if (or (not max-sync-duration)
                    ;;        (< sync-duration max-sync-duration)) ;; NOTE: db:sync-to-megatest.db keeps track of time of last sync and syncs incrementally
                        (let ((res        (db:sync-to-megatest.db dbstruct no-sync-db: no-sync-db))) ;; did we sync any data? If so need to set the db touched flag to keep the server alive
                          (set! sync-duration (- (current-milliseconds) sync-start))
                          (if (> res 0) ;; some records were transferred, keep the db alive
                              (begin
                                (mutex-lock! *heartbeat-mutex*)
                                (set! *db-last-access* (current-seconds))
                                (mutex-unlock! *heartbeat-mutex*)
                                (debug:print-info 0 *default-log-port* "sync called, " res " records transferred."))
                              (debug:print-info 2 *default-log-port* "sync called but zero records transferred")))))
;;                         ;; TODO: factor this next routine out into a function
;;                         (with-input-from-pipe ;; this should not block other threads but need to verify this
;;                          (conc "megatest -sync-to-megatest.db -m testsuite:" (common:get-area-name) ":" *toppath*)
;;                          (lambda ()
;;                            (let loop ((inl (read-line))
;;                                       (res #f))
;;                              (if (eof-object? inl)
;;                                  (begin
;;                                    (set! sync-duration (- (current-milliseconds) sync-start))
;;                                    (cond
;;                                     ((not res)
;;                                      (debug:print 0 *default-log-port* "ERROR: sync from /tmp db to megatest.db appears to have failed. Recommended that you stop your runs and run \"megatest -cleanup-db\""))
;;                                     ((> res 0)
;;                                      (mutex-lock! *heartbeat-mutex*)
;;                                      (set! *db-last-access* (current-seconds))
;;                                      (mutex-unlock! *heartbeat-mutex*))))
;;                                  (let ((num-synced (let ((matches (string-match "^Synced (\\d+).*$" inl)))
;;                                                      (if matches
;;                                                          (string->number (cadr matches))
;;                                                          #f))))
;;                                    (loop (read-line)
;;                                          (or num-synced res))))))))))
	      (if will-sync
		  (begin
		    (mutex-lock! *db-multi-sync-mutex*)
		    (set! *db-sync-in-progress* #f)
		    (set! *db-last-sync* start-time)
		    (with-output-to-file end-file (lambda ()(print (current-process-id))))