Megatest

Check-in [c8adfd5902]
Login
Overview
Comment:Peek at lock - if stuff in flight, don't bother getting in the queue
Downloads: Tarball | ZIP archive | SQL archive
Timelines: family | ancestors | descendants | both | static-html
Files: files | file ages | folders
SHA1: c8adfd59027fab04164f1111e78cd75158e1f012
User & Date: matt on 2015-03-02 21:55:29
Other Links: branch diff | manifest | tags
Context
2015-03-03
20:50
Be more agressive about removing lockdb when things go poorly check-in: f2dc94dec8 user: matt tags: static-html
2015-03-02
21:55
Peek at lock - if stuff in flight, don't bother getting in the queue check-in: c8adfd5902 user: matt tags: static-html
20:17
moved functions around for test-summary.html generation check-in: 58d975e0c7 user: mrwellan tags: static-html
Changes

Modified lock-queue.scm from [121fc7b3b4] to [759ae6ed41].

79
80
81
82
83
84
85

86
87
88
89
90
91
92
93
	 (debug:print 0 "ERROR:  Failed to set lock state for test with id " test-id ", error: " ((condition-property-accessor 'exn 'message) exn) ", giving up.")
	 #f))
   (sqlite3:execute (lock-queue:db-dat-get-db dbdat) "UPDATE queue SET state=? WHERE test_id=?;"
		    newstate
		    test-id)))

(define (lock-queue:any-younger? dbdat mystart test-id #!key (remtries 10))

  (tasks:wait-on-journal (lock-queue:db-dat-get-path dbdat) 1200)
  (handle-exceptions
   exn
   (if (> remtries 0)
       (begin
	 (debug:print 0 "WARNING: exception on lock-queue:any-younger. Trying again in 30 seconds.")
	 (debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
	 (thread-sleep! 30)







>
|







79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
	 (debug:print 0 "ERROR:  Failed to set lock state for test with id " test-id ", error: " ((condition-property-accessor 'exn 'message) exn) ", giving up.")
	 #f))
   (sqlite3:execute (lock-queue:db-dat-get-db dbdat) "UPDATE queue SET state=? WHERE test_id=?;"
		    newstate
		    test-id)))

(define (lock-queue:any-younger? dbdat mystart test-id #!key (remtries 10))
  ;; no need to wait on journal on read only queries
  ;; (tasks:wait-on-journal (lock-queue:db-dat-get-path dbdat) 1200)
  (handle-exceptions
   exn
   (if (> remtries 0)
       (begin
	 (debug:print 0 "WARNING: exception on lock-queue:any-younger. Trying again in 30 seconds.")
	 (debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
	 (thread-sleep! 30)
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
218
219
220
221
222
223
224
225
226
227
228
229
230
231
;; returns #t if ok to proceed with task
;; otherwise waits
;;
(define (lock-queue:wait-turn fname test-id #!key (count 10)(waiting-msg #f))
  (let* ((dbdat   (lock-queue:open-db fname))
	 (mystart (current-seconds))
	 (db      (lock-queue:db-dat-get-db dbdat)))
    (tasks:wait-on-journal (lock-queue:db-dat-get-path dbdat) 1200 waiting-msg: "lock-queue:wait-turn; waiting on journal file")
    (handle-exceptions
     exn
     (begin
       (debug:print 0 "WARNING: Failed to find out if it is ok to skip the wait queue. Will try again in few seconds")
       (debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
       (print-call-chain (current-error-port))
       (thread-sleep! 10)
       (if (> count 0)
	   (begin
	     (sqlite3:finalize! db)
	     (lock-queue:wait-turn fname test-id count: (- count 1)))
	   (begin
	     (debug:print 0 "Giving up calls to lock-queue:wait-turn for test-id " test-id " at path " fname ", printing call chain")
	     (print-call-chain (current-error-port))
	     #f)))




     ;; (tasks:wait-on-journal (lock-queue:db-dat-get-path dbdat) 1200 waiting-msg: "lock-queue:wait-turn; waiting on journal file")
     (sqlite3:execute
      db
      "INSERT OR REPLACE INTO queue (test_id,start_time,state) VALUES (?,?,'waiting');"
      test-id mystart)
     ;; (thread-sleep! 1) ;; give other tests a chance to register
     (let ((result 
	    (let loop ((younger-waiting (lock-queue:any-younger? dbdat mystart test-id)))
	      (if younger-waiting
		  (begin
		    ;; no need for us to wait. mark in the lock queue db as skipping


		    (lock-queue:set-state dbdat test-id "skipping")
		    #f) ;; let the calling process know that nothing needs to be done
		  (if (lock-queue:get-lock dbdat test-id)
		      #t
		      (if (> (- (current-seconds) mystart) 36000) ;; waited too long, steal the lock
			  (lock-queue:steal-lock dbdat test-id)
			  (begin
			    (thread-sleep! 1)
			    (loop (lock-queue:any-younger? dbdat mystart test-id)))))))))
       (sqlite3:finalize! db)
       result))))
	  
            
;; (use trace)
;; (trace lock-queue:get-lock lock-queue:release-lock lock-queue:wait-turn lock-queue:any-younger? lock-queue:set-state)







|















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




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
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
;; returns #t if ok to proceed with task
;; otherwise waits
;;
(define (lock-queue:wait-turn fname test-id #!key (count 10)(waiting-msg #f))
  (let* ((dbdat   (lock-queue:open-db fname))
	 (mystart (current-seconds))
	 (db      (lock-queue:db-dat-get-db dbdat)))
    ;; (tasks:wait-on-journal (lock-queue:db-dat-get-path dbdat) 1200 waiting-msg: "lock-queue:wait-turn; waiting on journal file")
    (handle-exceptions
     exn
     (begin
       (debug:print 0 "WARNING: Failed to find out if it is ok to skip the wait queue. Will try again in few seconds")
       (debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
       (print-call-chain (current-error-port))
       (thread-sleep! 10)
       (if (> count 0)
	   (begin
	     (sqlite3:finalize! db)
	     (lock-queue:wait-turn fname test-id count: (- count 1)))
	   (begin
	     (debug:print 0 "Giving up calls to lock-queue:wait-turn for test-id " test-id " at path " fname ", printing call chain")
	     (print-call-chain (current-error-port))
	     #f)))
     ;; wait 10 seconds and then check to see if someone is already updating the html
     (thread-sleep! 10)
     (if (not (lock-queue:any-younger? dbdat mystart test-id)) ;; no processing in flight, must try to start processing
	 (begin
	   (tasks:wait-on-journal (lock-queue:db-dat-get-path dbdat) 1200 waiting-msg: "lock-queue:wait-turn; waiting on journal file")
	   (sqlite3:execute
	    db
	    "INSERT OR REPLACE INTO queue (test_id,start_time,state) VALUES (?,?,'waiting');"
	    test-id mystart)
	   ;; (thread-sleep! 1) ;; give other tests a chance to register
	   (let ((result 
		  (let loop ((younger-waiting (lock-queue:any-younger? dbdat mystart test-id)))
		    (if younger-waiting
			(begin
			  ;; no need for us to wait. mark in the lock queue db as skipping
			  ;; no point in marking anything in the queue - simply never register this
			  ;; test as it is *covered* by a previously started update to the html file
			  ;; (lock-queue:set-state dbdat test-id "skipping")
			  #f) ;; let the calling process know that nothing needs to be done
			(if (lock-queue:get-lock dbdat test-id)
			    #t
			    (if (> (- (current-seconds) mystart) 36000) ;; waited too long, steal the lock
				(lock-queue:steal-lock dbdat test-id)
				(begin
				  (thread-sleep! 1)
				  (loop (lock-queue:any-younger? dbdat mystart test-id)))))))))
	     (sqlite3:finalize! db)
	     result))))))
	  
            
;; (use trace)
;; (trace lock-queue:get-lock lock-queue:release-lock lock-queue:wait-turn lock-queue:any-younger? lock-queue:set-state)