| ︙ | | | ︙ | |
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
|
headers: '((content-type text/plain))))
(else (continue))))))))
(http-transport:try-start-server run-id ipaddrstr start-port server-id)))
;; This is recursively run by http-transport:run until sucessful
;;
(define (http-transport:try-start-server run-id ipaddrstr portnum server-id)
(let ((config-hostname (configf:lookup *configdat* "server" "hostname")))
(debug:print-info 0 "http-transport:try-start-server run-id=" run-id " ipaddrsstr=" ipaddrstr " portnum=" portnum " server-id=" server-id " config-hostname=" config-hostname)
(handle-exceptions
exn
(begin
(print-error-message exn)
(if (< portnum 64000)
(begin
(debug:print 0 "WARNING: attempt to start server failed. Trying again ...")
(debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
(debug:print 0 "exn=" (condition->list exn))
(portlogger:open-run-close portlogger:set-failed portnum)
(debug:print 0 "WARNING: failed to start on portnum: " portnum ", trying next port")
(thread-sleep! 0.1)
;; get_next_port goes here
(http-transport:try-start-server run-id
ipaddrstr
(portlogger:open-run-close portlogger:find-port)
server-id))
(begin
(tasks:wait-on-busy-monitor.db)
(tasks:server-force-clean-run-record (tasks:get-db) run-id ipaddrstr portnum " http-transport:try-start-server")
(tasks:wait-on-busy-monitor.db)
(tasks:get-db)
server-id
ipaddrstr portnum)
(debug:print 0 "INFO: Trying to start server on " ipaddrstr ":" portnum)
;; This starts the spiffy server
;; NEED WAY TO SET IP TO #f TO BIND ALL
;; (start-server bind-address: ipaddrstr port: portnum)
(if config-hostname ;; this is a hint to bind directly
(start-server port: portnum bind-address: (if (equal? config-hostname "-")
ipaddrstr
config-hostname))
(start-server port: portnum))
;; (portlogger:open-run-close portlogger:set-port portnum "released")
(tasks:wait-on-busy-monitor.db)
(tasks:server-force-clean-run-record (tasks:get-db) run-id ipaddrstr portnum " http-transport:try-start-server")
|
|
>
<
|
<
|
<
|
|
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
|
headers: '((content-type text/plain))))
(else (continue))))))))
(http-transport:try-start-server run-id ipaddrstr start-port server-id)))
;; This is recursively run by http-transport:run until sucessful
;;
(define (http-transport:try-start-server run-id ipaddrstr portnum server-id)
(let ((config-hostname (configf:lookup *configdat* "server" "hostname"))
(tdbdat (tasks:open-db)))
(debug:print-info 0 "http-transport:try-start-server run-id=" run-id " ipaddrsstr=" ipaddrstr " portnum=" portnum " server-id=" server-id " config-hostname=" config-hostname)
(handle-exceptions
exn
(begin
(print-error-message exn)
(if (< portnum 64000)
(begin
(debug:print 0 "WARNING: attempt to start server failed. Trying again ...")
(debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
(debug:print 0 "exn=" (condition->list exn))
(portlogger:open-run-close portlogger:set-failed portnum)
(debug:print 0 "WARNING: failed to start on portnum: " portnum ", trying next port")
(thread-sleep! 0.1)
;; get_next_port goes here
(http-transport:try-start-server run-id
ipaddrstr
(portlogger:open-run-close portlogger:find-port)
server-id))
(begin
(tasks:server-force-clean-run-record (db:delay-if-busy tdbdat) run-id ipaddrstr portnum " http-transport:try-start-server")
(print "ERROR: Tried and tried but could not start the server"))))
;; any error in following steps will result in a retry
(set! *server-info* (list ipaddrstr portnum))
(tasks:server-set-interface-port
(db:delay-if-busy tdbdat)
server-id
ipaddrstr portnum)
(debug:print 0 "INFO: Trying to start server on " ipaddrstr ":" portnum)
;; This starts the spiffy server
;; NEED WAY TO SET IP TO #f TO BIND ALL
;; (start-server bind-address: ipaddrstr port: portnum)
(if config-hostname ;; this is a hint to bind directly
(start-server port: portnum bind-address: (if (equal? config-hostname "-")
ipaddrstr
config-hostname))
(start-server port: portnum))
;; (portlogger:open-run-close portlogger:set-port portnum "released")
(tasks:server-force-clean-run-record (db:delay-if-busy tdbdat) run-id ipaddrstr portnum " http-transport:try-start-server")
(debug:print 1 "INFO: server has been stopped"))))
;;======================================================================
;; S E R V E R U T I L I T I E S
;;======================================================================
;;======================================================================
|
| ︙ | | | ︙ | |
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
|
exn
(debug:print 0 "WARNING: closing connections failed. Server at " fullurl " almost certainly dead")
(close-all-connections!))
(debug:print 0 "WARNING: Failed to communicate with server, trying again, numretries left: " numretries)
(http-transport:client-api-send-receive run-id serverdat cmd params numretries: (- numretries 1)))
(begin
(mutex-unlock! *http-mutex*)
(tasks:wait-on-busy-monitor.db)
(debug:print 0 "WARNING: failure in with-input-from-request. Killing associated server to allow clean retry.")
(debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
(hash-table-delete! *runremote* run-id)
(tasks:kill-server-run-id run-id)
#f)
(with-input-from-request ;; was dat
fullurl
(list (cons 'key "thekey")
|
<
|
|
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
|
exn
(debug:print 0 "WARNING: closing connections failed. Server at " fullurl " almost certainly dead")
(close-all-connections!))
(debug:print 0 "WARNING: Failed to communicate with server, trying again, numretries left: " numretries)
(http-transport:client-api-send-receive run-id serverdat cmd params numretries: (- numretries 1)))
(begin
(mutex-unlock! *http-mutex*)
(tasks:kill-server-run-id run-id)
#f))
(begin
(debug:print-info 11 "fullurl=" fullurl ", cmd=" cmd ", params=" params ", run-id=" run-id "\n")
;; set up the http-client here
(max-retry-attempts 1)
;; consider all requests indempotent
(retry-request? (lambda (request)
#f))
;; send the data and get the response
;; extract the needed info from the http data and
;; process and return it.
(let* ((send-recieve (lambda ()
(mutex-lock! *http-mutex*)
;; (condition-case (with-input-from-request "http://localhost"; #f read-lines)
;; ((exn http client-error) e (print e)))
(set! res (handle-exceptions
exn
(begin
(debug:print 0 "WARNING: failure in with-input-from-request to " fullrul ". Killing associated server to allow clean retry.")
(debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
(hash-table-delete! *runremote* run-id)
(tasks:kill-server-run-id run-id)
#f)
(with-input-from-request ;; was dat
fullurl
(list (cons 'key "thekey")
|
| ︙ | | | ︙ | |
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
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
423
424
425
|
;; run http-transport:keep-running in a parallel thread to monitor that the db is being
;; used and to shutdown after sometime if it is not.
;;
(define (http-transport:keep-running server-id run-id)
;; if none running or if > 20 seconds since
;; server last used then start shutdown
;; This thread waits for the server to come alive
(let* ((server-info (let loop ((start-time (current-seconds))
(let ((tdb (tasks:open-db)))
(debug:print 0 "ERROR: transport appears to have died, exiting server " server-id " for run " run-id)
(tasks:wait-on-busy-monitor.db)
(tasks:server-delete-record tdb server-id "failed to start, never received server alive signature")
(sqlite3:finalize! tdb)
(tdb (tasks:open-db))
(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:wait-on-busy-monitor.db)
(tasks:server-set-state! tdb server-id "dbprep")
(tasks:server-set-state! tdb server-id "running")))
(if (and (<= rem-time 4)
(> rem-time 0))
(thread-sleep! rem-time)
(thread-sleep! 4))) ;; fallback for if the math is changed ...
(if (< count 1) ;; 3x3 = 9 secs aprox
|
>
|
|
<
|
<
<
|
<
|
|
|
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
|
;; run http-transport:keep-running in a parallel thread to monitor that the db is being
;; used and to shutdown after sometime if it is not.
;;
(define (http-transport:keep-running server-id run-id)
;; if none running or if > 20 seconds since
;; server last used then start shutdown
;; This thread waits for the server to come alive
(let* ((tdbdat (tasks:open-db))
(server-info (let loop ((start-time (current-seconds))
(changed #t)
(last-sdat "not this"))
(let ((sdat #f))
(thread-sleep! 0.01)
(debug:print-info 0 "Waiting for server alive signature")
(mutex-lock! *heartbeat-mutex*)
(set! sdat *server-info*)
(mutex-unlock! *heartbeat-mutex*)
(if (and sdat
(not changed)
(> (- (current-seconds) start-time) 2))
sdat
(begin
(debug:print-info 0 "Still waiting, last-sdat=" last-sdat)
(sleep 4)
(if (> (- (current-seconds) start-time) 120) ;; been waiting for two minutes
(begin
(debug:print 0 "ERROR: transport appears to have died, exiting server " server-id " for run " run-id)
(tasks:server-delete-record (db:delay-if-busy tdbdat) server-id "failed to start, never received server alive signature")
(exit))
(loop start-time
(equal? sdat last-sdat)
sdat)))))))
(iface (car server-info))
(port (cadr server-info))
(last-access 0)
(server-timeout (let ((tmo (configf:lookup *configdat* "server" "timeout")))
(if (and (string? tmo)
(string->number tmo))
(* 60 60 (string->number tmo))
;; (* 3 24 60 60) ;; default to three days
(* 60 1) ;; default to one minute
;; (* 60 60 25) ;; default to 25 hours
))))
(let loop ((count 0)
(server-state 'available))
;; Use this opportunity to sync the inmemdb to db
(let ((start-time (current-milliseconds))
(sync-time #f)
(rem-time #f))
;; inmemdb is a dbstruct
(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! (db:delay-if-busy tdbdat) server-id "dbprep")
(thread-sleep! 5) ;; give some margin for queries to complete before switching from file based access to server based access
(set! *inmemdb* (db:setup run-id))
(tasks:server-set-state! (db:delay-if-busy tdbdat) server-id "running")))
(if (and (<= rem-time 4)
(> rem-time 0))
(thread-sleep! rem-time)
(thread-sleep! 4))) ;; fallback for if the math is changed ...
(if (< count 1) ;; 3x3 = 9 secs aprox
|
| ︙ | | | ︙ | |
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
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
560
561
562
563
564
565
566
|
;; (tasks:server-set-state! tdb server-id "running"))
;;
(loop 0 server-state))
(begin
(debug:print-info 0 "Starting to shutdown the server.")
;; need to delete only *my* server entry (future use)
(set! *time-to-exit* #t)
(tasks:wait-on-busy-monitor.db) ;; wait here in addition to just before the shutting-down
(tasks:server-set-state! tdb server-id "shutting-down")
(portlogger:open-run-close portlogger:set-port port "released")
(thread-sleep! 5)
(debug:print-info 0 "Max cached queries was " *max-cache-size*)
(debug:print-info 0 "Number of cached writes " *number-of-writes*)
(debug:print-info 0 "Average cached write time "
(if (eq? *number-of-writes* 0)
"n/a (no writes)"
(/ *writes-total-delay*
*number-of-writes*))
" ms")
(debug:print-info 0 "Number non-cached queries " *number-non-write-queries*)
(debug:print-info 0 "Average non-cached time "
(if (eq? *number-non-write-queries* 0)
"n/a (no queries)"
(/ *total-non-write-delay*
*number-non-write-queries*))
" ms")
(debug:print-info 0 "Server shutdown complete. Exiting")
(tasks:wait-on-busy-monitor.db)
(tasks:server-delete-record tdb server-id " http-transport:keep-running")
(set! *run-id* run-id)
(if (args:get-arg "-daemonize")
(begin
(daemon:ize)
(if *alt-log-file* ;; we should re-connect to this port, I think daemon:ize disrupts it
(begin
(current-error-port *alt-log-file*)
(current-output-port *alt-log-file*)))))
(if (server:check-if-running run-id)
(begin
(debug:print 0 "INFO: Server for run-id " run-id " already running")
(exit 0)))
(tasks:wait-on-busy-monitor.db)
(let loop ((server-id (tasks:server-lock-slot (tasks:get-db) run-id))
(remtries 4))
(if (not server-id)
(if (> remtries 0)
(begin
(thread-sleep! 2)
(tasks:wait-on-busy-monitor.db)
(loop (tasks:server-lock-slot (tasks:get-db) run-id)
(- remtries 1)))
(begin
;; since we didn't get the server lock we are going to clean up and bail out
(debug:print-info 2 "INFO: server pid=" (current-process-id) ", hostname=" (get-host-name) " not starting due to other candidates ahead in start queue")
(tasks:wait-on-busy-monitor.db)
(tasks:server-delete-records-for-this-pid (tasks:get-db) " http-transport:launch")
))
(let* ((th2 (make-thread (lambda ()
(debug:print-info 0 "Server run thread started")
(http-transport:run
(if (args:get-arg "-server")
(args:get-arg "-server")
"-")
run-id
server-id)) "Server run"))
(th3 (make-thread (lambda ()
(debug:print-info 0 "Server monitor thread started")
(http-transport:keep-running server-id run-id))
"Keep running")))
;; Database connection
;; don't start the db here
;; (set! *inmemdb* (db:setup run-id))
(thread-start! th2)
(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 " ... exiting ...")
(let ((th1 (make-thread (lambda ()
|
<
|
<
|
>
|
|
|
|
|
|
|
|
|
|
|
|
<
|
|
|
|
|
|
<
|
|
|
|
|
<
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
461
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
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
|
;; (tasks:server-set-state! tdb server-id "running"))
;;
(loop 0 server-state))
(begin
(debug:print-info 0 "Starting to shutdown the server.")
;; need to delete only *my* server entry (future use)
(set! *time-to-exit* #t)
(if *inmemdb* (db:sync-touched *inmemdb* *run-id* force-sync: #t))
;;
;; start_shutdown
;;
(tasks:server-set-state! (db:delay-if-busy tdbdat) server-id "shutting-down")
(portlogger:open-run-close portlogger:set-port port "released")
(thread-sleep! 5)
(debug:print-info 0 "Max cached queries was " *max-cache-size*)
(debug:print-info 0 "Number of cached writes " *number-of-writes*)
(debug:print-info 0 "Average cached write time "
(if (eq? *number-of-writes* 0)
"n/a (no writes)"
(/ *writes-total-delay*
*number-of-writes*))
" ms")
(debug:print-info 0 "Number non-cached queries " *number-non-write-queries*)
(debug:print-info 0 "Average non-cached time "
(if (eq? *number-non-write-queries* 0)
"n/a (no queries)"
(/ *total-non-write-delay*
*number-non-write-queries*))
" ms")
(debug:print-info 0 "Server shutdown complete. Exiting")
(tasks:server-delete-record (db:delay-if-busy tdbdat) server-id " http-transport:keep-running")
(exit))))))
;; all routes though here end in exit ...
;;
;; start_server?
;;
(define (http-transport:launch run-id)
(let* ((tdbdat (tasks:open-db)))
(set! *run-id* run-id)
(if (args:get-arg "-daemonize")
(begin
(daemon:ize)
(if *alt-log-file* ;; we should re-connect to this port, I think daemon:ize disrupts it
(begin
(current-error-port *alt-log-file*)
(current-output-port *alt-log-file*)))))
(if (server:check-if-running run-id)
(begin
(debug:print 0 "INFO: Server for run-id " run-id " already running")
(exit 0)))
(let loop ((server-id (tasks:server-lock-slot (db:delay-if-busy tdbdat) run-id))
(remtries 4))
(if (not server-id)
(if (> remtries 0)
(begin
(thread-sleep! 2)
(loop (tasks:server-lock-slot (db:delay-if-busy tdbdat) run-id)
(- remtries 1)))
(begin
;; since we didn't get the server lock we are going to clean up and bail out
(debug:print-info 2 "INFO: server pid=" (current-process-id) ", hostname=" (get-host-name) " not starting due to other candidates ahead in start queue")
(tasks:server-delete-records-for-this-pid (db:delay-if-busy tdbdat) " http-transport:launch")
))
(let* ((th2 (make-thread (lambda ()
(debug:print-info 0 "Server run thread started")
(http-transport:run
(if (args:get-arg "-server")
(args:get-arg "-server")
"-")
run-id
server-id)) "Server run"))
(th3 (make-thread (lambda ()
(debug:print-info 0 "Server monitor thread started")
(http-transport:keep-running server-id run-id))
"Keep running")))
;; Database connection
;; don't start the db here
;; (set! *inmemdb* (db:setup run-id))
(thread-start! th2)
(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 " ... exiting ...")
(let ((th1 (make-thread (lambda ()
|
| ︙ | | | ︙ | |