Megatest

Diff
Login

Differences From Artifact [ef41e52830]:

To Artifact [0582204ff9]:


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
80
81
82
83
84
85
86
87
88

89


90
91
92









93
94







95
96
97
98
99
100

101
102
103

104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
	 (vector-set! record 1 count)
	 (if (and (> count 10)
		  (> queries-per-second 10))
	     (begin
	       (debug:print-info 1 "db write rate too high, starting a server, count=" count " start=" start " run-id=" run-id " queries-per-second=" queries-per-second)
	       #t)
	     #f))))











;; cmd is a symbol
;; vars is a json string encoding the parameters for the call
;;
(define (rmt:send-receive cmd rid params)
  ;; clean out old connections
  (mutex-lock! *db-multi-sync-mutex*)
  (let ((expire-time (- (current-seconds) 60)))
    (for-each 
     (lambda (run-id)
       (let ((connection (hash-table-ref/default *runremote* run-id #f)))
	 (if ;; (and connection 
		  (< (http-transport:server-dat-get-last-access connection) expire-time) ; )
	     (begin
	       (debug:print-info 0 "Discarding connection to server for run-id " run-id ", too long between accesses")
	       (hash-table-delete! *runremote* run-id)))))
     (hash-table-keys *runremote*)))
  (mutex-unlock! *db-multi-sync-mutex*)
  (let* ((run-id          (if rid rid 0))
	 (connection-info (let ((cinfo (hash-table-ref/default *runremote* run-id #f)))
			    (if cinfo
				cinfo
				;; NB// can cache the answer for server running for 10 seconds ...
				;;  ;; (and (not (rmt:write-frequency-over-limit? cmd run-id))
				(if (tasks:server-running-or-starting? (tasks:get-db) run-id)
				    (let ((res (client:setup run-id)))
				      (if res 
					  (hash-table-ref/default *runremote* run-id #f) ;; client:setup filled this in (hopefully)
					  #f))
				    #f))))
	 (jparams         (db:obj->string params)))
    (if connection-info

	(let ((res             (http-transport:client-api-send-receive run-id connection-info cmd jparams)))


	  (http-transport:server-dat-update-last-access connection-info)
	  (if res
	      (db:string->obj res)









	      (let ((new-connection-info (client:setup run-id)))
		(debug:print 0 "WARNING: Communication failed, trying call to http-transport:client-api-send-receive again.")







		(rmt:send-receive cmd run-id params))))
	(let ((max-avg-qry (string->number (or (configf:lookup *configdat* "server" "server-query-threshold") "-1"))))
	  (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*)
  (handle-exceptions
   exn
   (begin
     (debug:print 0 "WARNING: stats collection failed in update-db-stats")
     (debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
     (print "exn=" (condition->list exn))
     #f) ;; if this fails we don't care, it is just stats
   (let* ((cmd      (if (eq? rawcmd 'general-call) (car params) rawcmd))
	  (stat-vec (hash-table-ref/default *db-stats* cmd #f)))
     (if (not stat-vec)
	 (let ((newvec (vector 0 0)))
	   (hash-table-set! *db-stats* cmd newvec)
	   (set! stat-vec newvec)))
     (vector-set! stat-vec 0 (+ (vector-ref stat-vec 0) 1))
     (vector-set! stat-vec 1 (+ (vector-ref stat-vec 1) duration))))







>
>
>
>
>
>
>
>
>
>




|






|
|






|
<
<
<
<
<
<
<
<
<
<


>
|
>
>

|

>
>
>
>
>
>
>
>
>
|

>
>
>
>
>
>
>
|
<
<
<
<
|
>
|
<
|
>
|

|








|







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
80
81
82
83
84
85
86










87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114




115
116
117

118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
	 (vector-set! record 1 count)
	 (if (and (> count 10)
		  (> queries-per-second 10))
	     (begin
	       (debug:print-info 1 "db write rate too high, starting a server, count=" count " start=" start " run-id=" run-id " queries-per-second=" queries-per-second)
	       #t)
	     #f))))

(define (rmt:get-connection-info run-id)
  (let ((cinfo (hash-table-ref/default *runremote* run-id #f)))
    (if cinfo
	cinfo
	;; NB// can cache the answer for server running for 10 seconds ...
	;;  ;; (and (not (rmt:write-frequency-over-limit? cmd run-id))
	(if (tasks:server-running-or-starting? (db:delay-if-busy (tasks:open-db)) run-id)
	    (client:setup run-id)
	    #f))))

;; cmd is a symbol
;; vars is a json string encoding the parameters for the call
;;
(define (rmt:send-receive cmd rid params #!key (attemptnum 0))
  ;; clean out old connections
  (mutex-lock! *db-multi-sync-mutex*)
  (let ((expire-time (- (current-seconds) 60)))
    (for-each 
     (lambda (run-id)
       (let ((connection (hash-table-ref/default *runremote* run-id #f)))
	 (if (and connection 
		  (< (http-transport:server-dat-get-last-access connection) expire-time))
	     (begin
	       (debug:print-info 0 "Discarding connection to server for run-id " run-id ", too long between accesses")
	       (hash-table-delete! *runremote* run-id)))))
     (hash-table-keys *runremote*)))
  (mutex-unlock! *db-multi-sync-mutex*)
  (let* ((run-id          (if rid rid 0))
	 (connection-info (rmt:get-connection-info run-id))










	 (jparams         (db:obj->string params)))
    (if connection-info
	;; use the server if have connection info
	(let* ((dat     (http-transport:client-api-send-receive run-id connection-info cmd jparams))
	       (res     (if (and dat (vector? dat)) (vector-ref dat 1) #f))
	       (success (if (and dat (vector? dat)) (vector-ref dat 0) #f)))
	  (http-transport:server-dat-update-last-access connection-info)
	  (if success
	      (db:string->obj res)
	      ;; (if (< attemptnum 100)
	      ;;     (begin
	      ;;       (hash-table-delete! *runremote* run-id)
	      ;;       (thread-sleep! 0.5)
	      ;;       (rmt:send-receive cmd rid params attempnum: (+ attemptnum 1)))
	      ;;     (begin
	      ;;       (print-call-chain (current-error-port))
	      ;;       (debug:print 0 "ERROR: too many attempts to communicate have failed. Giving up. Kill your mtest processes and start over")
	      ;;       (exit 1)))))
	      (begin ;; let ((new-connection-info (client:setup run-id)))
		(debug:print 0 "WARNING: Communication failed, trying call to http-transport:client-api-send-receive again.")
		(hash-table-delete! *runremote* run-id) ;; don't keep using the same connection

		;; no longer killing the server in http-transport:client-api-send-receive
		;; may kill it here but what are the criteria?
		;; start with three calls then kill server
		(if (eq? attemptnum 3)(tasks:kill-server-run-id run-id))
		(thread-sleep! 2)
		(rmt:send-receive cmd run-id params attemptnum: (+ attemptnum 1)))))




	(if (and (< attemptnum 10)
		 (tasks:need-server run-id))
	    (begin

	      (tasks:start-and-wait-for-server (db:delay-if-busy (tasks:open-db)) run-id 10)
	      (rmt:send-receive cmd rid params (+ attemptnum 1)))
	    (rmt:open-qry-close-locally cmd run-id params)))))

(define (rmt:update-db-stats run-id rawcmd params duration)
  (mutex-lock! *db-stats-mutex*)
  (handle-exceptions
   exn
   (begin
     (debug:print 0 "WARNING: stats collection failed in update-db-stats")
     (debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
     (print "exn=" (condition->list exn))
     #f) ;; if this fails we don't care, it is just stats
   (let* ((cmd      (conc "run-id=" run-id " " (if (eq? rawcmd 'general-call) (car params) rawcmd)))
	  (stat-vec (hash-table-ref/default *db-stats* cmd #f)))
     (if (not stat-vec)
	 (let ((newvec (vector 0 0)))
	   (hash-table-set! *db-stats* cmd newvec)
	   (set! stat-vec newvec)))
     (vector-set! stat-vec 0 (+ (vector-ref stat-vec 0) 1))
     (vector-set! stat-vec 1 (+ (vector-ref stat-vec 1) duration))))
131
132
133
134
135
136
137
138
139



140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177

178
179
180
181
182
183
184
185
186
187


188
189
190
191
192
193
194
195
		(let ((cmd-dat (hash-table-ref *db-stats* cmd)))
		  (debug:print 18 (format #f fmtstr cmd (vector-ref cmd-dat 0) (vector-ref cmd-dat 1) (/ (vector-ref cmd-dat 1)(vector-ref cmd-dat 0))))))
	      (sort (hash-table-keys *db-stats*)
		    (lambda (a b)
		      (> (vector-ref (hash-table-ref *db-stats* a) 0)
			 (vector-ref (hash-table-ref *db-stats* b) 0)))))))

(define (rmt:get-max-query-average)
  (mutex-lock! *db-stats-mutex*)



  (let* ((cmds (hash-table-keys *db-stats*))
	 (res  (if (null? cmds)
		   (cons 'none 0)
		   (let loop ((cmd (car cmds))
			      (tal (cdr cmds))
			      (max-cmd (car cmds))
			      (res 0))
		     (let* ((cmd-dat (hash-table-ref *db-stats* cmd))
			    (tot     (vector-ref cmd-dat 0))
			    (curravg (/ (vector-ref cmd-dat 1) (vector-ref cmd-dat 0))) ;; count is never zero by construction
			    (currmax (max res curravg))
			    (newmax-cmd (if (> curravg res) cmd max-cmd)))
		       (if (null? tal)
			   (if (> tot 10)
			       (cons newmax-cmd currmax)
			       (cons 'none 0))
			   (loop (car tal)(cdr tal) newmax-cmd currmax)))))))
    (mutex-unlock! *db-stats-mutex*)
    res))
	  
(define (rmt:open-qry-close-locally cmd run-id params)
  (let* ((dbstruct-local (if *dbstruct-db*
			     *dbstruct-db*
			     (let* ((dbdir (conc    (configf:lookup *configdat* "setup" "linktree") "/.db"))
				    (db (make-dbr:dbstruct path:  dbdir local: #t)))
			       (set! *dbstruct-db* db)
			       db)))
	 (db-file-path   (db:dbfile-path 0)))
    ;; (read-only      (not (file-read-access? db-file-path)))
    (let* ((start         (current-milliseconds))
	   (res           (api:execute-requests dbstruct-local (symbol->string cmd) params))
	   (duration      (- (current-milliseconds) start)))
      (rmt:update-db-stats cmd params duration)
      ;; mark this run as dirty if this was a write
      (if (not (member cmd api:read-only-queries))
	  (let ((start-time (current-seconds)))
	    (mutex-lock! *db-multi-sync-mutex*)
	    (if (not (hash-table-ref/default *db-local-sync* run-id #f))

		(hash-table-set! *db-local-sync* run-id start-time)) ;; the oldest "write"
	    (mutex-unlock! *db-multi-sync-mutex*)))
      res)))

(define (rmt:send-receive-no-auto-client-setup connection-info cmd run-id params)
  (let* ((run-id   (if run-id run-id 0))
	 (jparams         (db:obj->string params)) ;; (rmt:dat->json-str params))
	 (res (http-transport:client-api-send-receive run-id connection-info cmd jparams)))
    (if res
	(db:string->obj res)


	res)))

;; Wrap json library for strings (why the ports crap in the first place?)
(define (rmt:dat->json-str dat)
  (with-output-to-string 
    (lambda ()
      (json-write dat))))








|

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















|




|
>
|





|
|
|
|
>
>
|







147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
		(let ((cmd-dat (hash-table-ref *db-stats* cmd)))
		  (debug:print 18 (format #f fmtstr cmd (vector-ref cmd-dat 0) (vector-ref cmd-dat 1) (/ (vector-ref cmd-dat 1)(vector-ref cmd-dat 0))))))
	      (sort (hash-table-keys *db-stats*)
		    (lambda (a b)
		      (> (vector-ref (hash-table-ref *db-stats* a) 0)
			 (vector-ref (hash-table-ref *db-stats* b) 0)))))))

(define (rmt:get-max-query-average run-id)
  (mutex-lock! *db-stats-mutex*)
  (let* ((runkey (conc "run-id=" run-id " "))
	 (cmds   (filter (lambda (x)
			   (substring-index runkey x))
			 (hash-table-keys *db-stats*)))
	 (res    (if (null? cmds)
		     (cons 'none 0)
		     (let loop ((cmd (car cmds))
				(tal (cdr cmds))
				(max-cmd (car cmds))
				(res 0))
		       (let* ((cmd-dat (hash-table-ref *db-stats* cmd))
			      (tot     (vector-ref cmd-dat 0))
			      (curravg (/ (vector-ref cmd-dat 1) (vector-ref cmd-dat 0))) ;; count is never zero by construction
			      (currmax (max res curravg))
			      (newmax-cmd (if (> curravg res) cmd max-cmd)))
			 (if (null? tal)
			     (if (> tot 10)
				 (cons newmax-cmd currmax)
				 (cons 'none 0))
			     (loop (car tal)(cdr tal) newmax-cmd currmax)))))))
    (mutex-unlock! *db-stats-mutex*)
    res))
	  
(define (rmt:open-qry-close-locally cmd run-id params)
  (let* ((dbstruct-local (if *dbstruct-db*
			     *dbstruct-db*
			     (let* ((dbdir (conc    (configf:lookup *configdat* "setup" "linktree") "/.db"))
				    (db (make-dbr:dbstruct path:  dbdir local: #t)))
			       (set! *dbstruct-db* db)
			       db)))
	 (db-file-path   (db:dbfile-path 0)))
    ;; (read-only      (not (file-read-access? db-file-path)))
    (let* ((start         (current-milliseconds))
	   (res           (api:execute-requests dbstruct-local (symbol->string cmd) params))
	   (duration      (- (current-milliseconds) start)))
      (rmt:update-db-stats run-id cmd params duration)
      ;; mark this run as dirty if this was a write
      (if (not (member cmd api:read-only-queries))
	  (let ((start-time (current-seconds)))
	    (mutex-lock! *db-multi-sync-mutex*)
	    ;; (if (not (hash-table-ref/default *db-local-sync* run-id #f))
	    ;; just set it every time. Is a write more expensive than a read and does it matter?
	    (hash-table-set! *db-local-sync* (or run-id 0) start-time) ;; the oldest "write"
	    (mutex-unlock! *db-multi-sync-mutex*)))
      res)))

(define (rmt:send-receive-no-auto-client-setup connection-info cmd run-id params)
  (let* ((run-id   (if run-id run-id 0))
	 (jparams  (db:obj->string params)) ;; (rmt:dat->json-str params))
	 (dat      (http-transport:client-api-send-receive run-id connection-info cmd jparams)))
    (if (and dat (vector-ref dat 0))
	(db:string->obj (vector-ref dat 1))
	(begin
	  (debug:print 0 "ERROR: rmt:send-receive-no-auto-client-setup failed, attempting to continue. Got " dat)
	  dat))))

;; Wrap json library for strings (why the ports crap in the first place?)
(define (rmt:dat->json-str dat)
  (with-output-to-string 
    (lambda ()
      (json-write dat))))

263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
  (rmt:send-receive 'get-test-id run-id (list run-id testname item-path)))

(define (rmt:get-test-info-by-id run-id test-id)
  (if (and (number? run-id)(number? test-id))
      (rmt:send-receive 'get-test-info-by-id run-id (list run-id test-id))
      (begin
	(debug:print 0 "WARNING: Bad data handed to rmt:get-test-info-by-id run-id=" run-id ", test-id=" test-id)
	(print-call-chain)
	#f)))

(define (rmt:test-get-rundir-from-test-id run-id test-id)
  (rmt:send-receive 'test-get-rundir-from-test-id run-id (list run-id test-id)))

(define (rmt:open-test-db-by-test-id run-id test-id #!key (work-area #f))
  (let* ((test-path (if (string? work-area)







|







285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
  (rmt:send-receive 'get-test-id run-id (list run-id testname item-path)))

(define (rmt:get-test-info-by-id run-id test-id)
  (if (and (number? run-id)(number? test-id))
      (rmt:send-receive 'get-test-info-by-id run-id (list run-id test-id))
      (begin
	(debug:print 0 "WARNING: Bad data handed to rmt:get-test-info-by-id run-id=" run-id ", test-id=" test-id)
	(print-call-chain (current-error-port))
	#f)))

(define (rmt:test-get-rundir-from-test-id run-id test-id)
  (rmt:send-receive 'test-get-rundir-from-test-id run-id (list run-id test-id)))

(define (rmt:open-test-db-by-test-id run-id test-id #!key (work-area #f))
  (let* ((test-path (if (string? work-area)
288
289
290
291
292
293
294
295
296
297


298

299
300
301


































302
303
304
305
306
307
308
309
310
311
  (rmt:send-receive 'set-tests-state-status run-id (list run-id testnames currstate currstatus newstate newstatus)))

(define (rmt:get-tests-for-run run-id testpatt states statuses offset limit not-in sort-by sort-order qryvals)
  (if (number? run-id)
      (rmt:send-receive 'get-tests-for-run run-id (list run-id testpatt states statuses offset limit not-in sort-by sort-order qryvals))
      (begin
	(debug:print "ERROR: rmt:get-tests-for-run called with bad run-id=" run-id)
	(print-call-chain)
	'())))



(define (rmt:get-tests-for-runs-mindata run-ids testpatt states status not-in)

  (let ((run-id-list (if run-ids
			 run-ids
			 (rmt:get-all-run-ids))))


































    (apply append (map (lambda (run-id)
			 (rmt:send-receive 'get-tests-for-run-mindata run-id (list run-ids testpatt states status not-in)))
		       run-id-list))))

(define (rmt:delete-test-records run-id test-id)
  (rmt:send-receive 'delete-test-records run-id (list run-id test-id)))

;; This is not needed as test steps are deleted on test delete call
;;
;; (define (rmt:delete-test-step-records run-id test-id)







|


>
>

>
|

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







310
311
312
313
314
315
316
317
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
  (rmt:send-receive 'set-tests-state-status run-id (list run-id testnames currstate currstatus newstate newstatus)))

(define (rmt:get-tests-for-run run-id testpatt states statuses offset limit not-in sort-by sort-order qryvals)
  (if (number? run-id)
      (rmt:send-receive 'get-tests-for-run run-id (list run-id testpatt states statuses offset limit not-in sort-by sort-order qryvals))
      (begin
	(debug:print "ERROR: rmt:get-tests-for-run called with bad run-id=" run-id)
	(print-call-chain (current-error-port))
	'())))

;; IDEA: Threadify these - they spend a lot of time waiting ...
;;
(define (rmt:get-tests-for-runs-mindata run-ids testpatt states status not-in)
  (let ((multi-run-mutex (make-mutex))
	(run-id-list (if run-ids
			 run-ids
			 (rmt:get-all-run-ids)))
	(result      '()))
    (if (null? run-id-list)
	'()
	(for-each 
	 (lambda (th)

	   (thread-join! th)) ;; I assume that joining completed threads just moves on
	 (let loop ((hed     (car run-id-list))
		    (tal     (cdr run-id-list))
		    (threads '()))
	   (let* ((newthread (make-thread
			      (lambda ()
				(let ((res (rmt:send-receive 'get-tests-for-run-mindata hed (list hed testpatt states status not-in))))
				  (if (list? res)
				      (begin
					(mutex-lock! multi-run-mutex)
					(set! result (append result res))
					(mutex-unlock! multi-run-mutex))
				      (debug:print 0 "ERROR: get-tests-for-run-mindata failed for run-id " hed ", testpatt " testpatt ", states " states ", status " status ", not-in " not-in))))
			      (conc "multi-run-thread for run-id " hed)))
		  (newthreads (cons newthread threads)))
	     (thread-start! newthread)
	     (thread-sleep! 0.5) ;; give that thread some time to start
	     (if (null? tal)
		 newthreads
		 (loop (car tal)(cdr tal) newthreads))))))
    result))

;; ;; IDEA: Threadify these - they spend a lot of time waiting ...
;; ;;
;; (define (rmt:get-tests-for-runs-mindata run-ids testpatt states status not-in)
;;   (let ((run-id-list (if run-ids
;; 			 run-ids
;; 			 (rmt:get-all-run-ids))))
;;     (apply append (map (lambda (run-id)
;; 			 (rmt:send-receive 'get-tests-for-run-mindata run-id (list run-ids testpatt states status not-in)))
;; 		       run-id-list))))

(define (rmt:delete-test-records run-id test-id)
  (rmt:send-receive 'delete-test-records run-id (list run-id test-id)))

;; This is not needed as test steps are deleted on test delete call
;;
;; (define (rmt:delete-test-step-records run-id test-id)