| ︙ | | | ︙ | |
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
|
(config-use-proxy (equal? (configf:lookup *configdat* "client" "use-http_proxy") "yes")))
(if (not config-use-proxy)
(determine-proxy (constantly #f)))
(debug:print-info 0 *default-log-port* "http-transport:try-start-server time=" (seconds->time-string (current-seconds)) " ipaddrsstr=" ipaddrstr " portnum=" portnum " config-hostname=" config-hostname)
(handle-exceptions
exn
(begin
(print-error-message exn)
(if (< portnum 64000)
(begin
(debug:print 0 *default-log-port* "WARNING: attempt to start server failed. Trying again ...")
(debug:print 0 *default-log-port* " message: " ((condition-property-accessor 'exn 'message) exn))
(debug:print 5 *default-log-port* "exn=" (condition->list exn))
(portlogger:open-run-close portlogger:set-failed portnum)
(debug:print 0 *default-log-port* "WARNING: failed to start on portnum: " portnum ", trying next port")
|
|
|
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
|
(config-use-proxy (equal? (configf:lookup *configdat* "client" "use-http_proxy") "yes")))
(if (not config-use-proxy)
(determine-proxy (constantly #f)))
(debug:print-info 0 *default-log-port* "http-transport:try-start-server time=" (seconds->time-string (current-seconds)) " ipaddrsstr=" ipaddrstr " portnum=" portnum " config-hostname=" config-hostname)
(handle-exceptions
exn
(begin
;; (print-error-message exn)
(if (< portnum 64000)
(begin
(debug:print 0 *default-log-port* "WARNING: attempt to start server failed. Trying again ...")
(debug:print 0 *default-log-port* " message: " ((condition-property-accessor 'exn 'message) exn))
(debug:print 5 *default-log-port* "exn=" (condition->list exn))
(portlogger:open-run-close portlogger:set-failed portnum)
(debug:print 0 *default-log-port* "WARNING: failed to start on portnum: " portnum ", trying next port")
|
| ︙ | | | ︙ | |
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
|
;; (remote-conndat-set! runremote #f))
;; Killing associated server to allow clean retry.")
;; (tasks:kill-server-run-id run-id) ;; better to kill the server in the logic that called this routine?
(mutex-unlock! *http-mutex*)
;; (signal (make-composite-condition
;; (make-property-condition 'commfail 'message "failed to connect to server")))
;; "communications failed"
(close-all-connections!)
(db:obj->string #f))
(with-input-from-request ;; was dat
fullurl
(list (cons 'key (or server-id "thekey"))
(cons 'cmd cmd)
(cons 'params sparams))
read-string))
|
|
>
|
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
|
;; (remote-conndat-set! runremote #f))
;; Killing associated server to allow clean retry.")
;; (tasks:kill-server-run-id run-id) ;; better to kill the server in the logic that called this routine?
(mutex-unlock! *http-mutex*)
;; (signal (make-composite-condition
;; (make-property-condition 'commfail 'message "failed to connect to server")))
;; "communications failed"
;; (close-all-connections!)
(close-connection! fullurl)
(db:obj->string #f))
(with-input-from-request ;; was dat
fullurl
(list (cons 'key (or server-id "thekey"))
(cons 'cmd cmd)
(cons 'params sparams))
read-string))
|
| ︙ | | | ︙ | |
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
|
(signal (make-composite-condition
(make-property-condition
'timeout
'message "nmsg-transport:client-api-send-receive-raw timed out talking to server")))))))
;; careful closing of connections stored in *runremote*
;;
(define (http-transport:close-connections #!key (area-dat #f))
(let* ((runremote (or area-dat *runremote*))
(server-dat (if runremote
(remote-conndat runremote)
#f))) ;; (hash-table-ref/default *runremote* run-id #f)))
(if (vector? server-dat)
(let ((api-dat (http-transport:server-dat-get-api-uri server-dat)))
(handle-exceptions
exn
(begin
(print-call-chain *default-log-port*)
(debug:print-error 0 *default-log-port* " closing connection failed with error: " ((condition-property-accessor 'exn 'message) exn) ", exn=" exn))
(close-connection! api-dat)
(close-idle-connections!)
|
|
|
>
>
|
>
|
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
371
372
373
374
|
(signal (make-composite-condition
(make-property-condition
'timeout
'message "nmsg-transport:client-api-send-receive-raw timed out talking to server")))))))
;; careful closing of connections stored in *runremote*
;;
(define (http-transport:close-connections runremote)
(let* (;; (runremote (or area-dat *runremote*))
(server-dat (if runremote
(remote-conndat runremote)
#f))) ;; (hash-table-ref/default *runremote* run-id #f)))
(if (vector? server-dat)
(let ((api-dat (http-transport:server-dat-get-api-uri server-dat)))
(handle-exceptions
exn
(begin
(print-call-chain *default-log-port*)
(debug:print-error 0 *default-log-port* " closing connection failed with error: " ((condition-property-accessor 'exn 'message) exn) ", exn=" exn))
(if (args:any-defined? "-server" "-execute" "-run")
(debug:print-info 0 *default-log-port* "Closing connections to "api-dat))
(close-connection! api-dat)
(close-connection! (http-transport:server-dat-make-url server-dat))
(remote-conndat-set! runremote #f)
#t))
#f)))
(define (make-http-transport:server-dat)(make-vector 6))
(define (http-transport:server-dat-get-iface vec) (vector-ref vec 0))
(define (http-transport:server-dat-get-port vec) (vector-ref vec 1))
|
| ︙ | | | ︙ | |
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
|
(print-call-chain (current-error-port))
(debug:print-error 0 *default-log-port* "call to http-transport:server-dat-update-last-access with non-vector!!"))))
;;
;; connect
;;
(define (http-transport:client-connect iface port server-id)
(debug:print-info 0 *default-log-port* "Connecting to client at "iface":"port", with server-id "server-id)
(let* ((api-url (conc "http://" iface ":" port "/api"))
(api-uri (uri-reference (conc "http://" iface ":" port "/api")))
(api-req (make-request method: 'POST uri: api-uri))
(server-dat (vector iface port api-uri api-url api-req (current-seconds) server-id)))
server-dat))
|
|
|
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
|
(print-call-chain (current-error-port))
(debug:print-error 0 *default-log-port* "call to http-transport:server-dat-update-last-access with non-vector!!"))))
;;
;; connect
;;
(define (http-transport:client-connect iface port server-id)
(debug:print-info 2 *default-log-port* "Connecting to server at "iface":"port", id "server-id)
(let* ((api-url (conc "http://" iface ":" port "/api"))
(api-uri (uri-reference (conc "http://" iface ":" port "/api")))
(api-req (make-request method: 'POST uri: api-uri))
(server-dat (vector iface port api-uri api-url api-req (current-seconds) server-id)))
server-dat))
|
| ︙ | | | ︙ | |
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
|
(debug:print-info 0 *default-log-port* "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))
(let* ((servinfodir (conc *toppath*"/.servinfo"))
(ipaddr (car sdat))
(port (cadr sdat))
(servinf (conc servinfodir"/"ipaddr":"port)))
(set! servinfofile servinf)
(if (not (file-exists? servinfodir))
(create-directory servinfodir #t))
(with-output-to-file servinf
(lambda ()
(let* ((serv-id (server:mk-signature)))
(set! *server-id* serv-id)
(print "SERVER STARTED: "ipaddr":"port" AT "(current-seconds)" server-id: "serv-id" pid: "(current-process-id))
(print "started: "(seconds->year-week/day-time (current-seconds))))))
(set! *on-exit-procs* (cons
(lambda ()
(delete-file* servinf))
*on-exit-procs*))
;; put data about this server into a simple flat file host.port
(debug:print-info 0 *default-log-port* "Received server alive signature")
#;(common:save-pkt `((action . alive)
(T . server)
(pid . ,(current-process-id))
(ipaddr . ,(car sdat))
(port . ,(cadr sdat)))
*configdat* #t)
(let* ((ipaddr (car sdat))
(port (cadr sdat))
(servinf (conc *toppath*"/.servinfo/"ipaddr":"port)))
(debug:print-error 0 *default-log-port* "transport appears to have died, exiting server")
;; (delete-file* servinf) ;; handled by on-exit, can be removed
#;(common:save-pkt `((action . died)
(T . server)
(pid . ,(current-process-id))
(ipaddr . ,(car sdat))
(port . ,(cadr sdat))
(msg . "Transport died?"))
*configdat* #t)
(exit))
(loop start-time
(equal? sdat last-sdat)
sdat)))))))
(iface (car server-info))
(port (cadr server-info))
(last-access 0)
(server-timeout (server:expiration-timeout))
|
|
<
<
<
<
<
<
>
|
|
|
<
<
<
<
|
<
<
<
|
>
|
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
|
(debug:print-info 0 *default-log-port* "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))
(let* ((servinfodir (server:get-servinfo-dir *toppath*)) ;; (conc *toppath*"/.servinfo"))
(ipaddr (car sdat))
(port (cadr sdat))
(servinf (conc servinfodir"/"ipaddr":"port)))
(set! servinfofile servinf)
(if (not (file-exists? servinfodir))
(create-directory servinfodir #t))
(with-output-to-file servinf
(lambda ()
(let* ((serv-id (server:mk-signature)))
(set! *server-id* serv-id)
(print "SERVER STARTED: "ipaddr":"port" AT "(current-seconds)" server-id: "serv-id" pid: "(current-process-id))
(print "started: "(seconds->year-week/day-time (current-seconds))))))
(set! *on-exit-procs* (cons
(lambda ()
(delete-file* servinf))
*on-exit-procs*))
;; put data about this server into a simple flat file host.port
(debug:print-info 0 *default-log-port* "Received server alive signature")
sdat)
(begin
(debug:print-info 0 *default-log-port* "Still waiting, last-sdat=" last-sdat)
(sleep 4)
(if (> (- (current-seconds) start-time) 120) ;; been waiting for two minutes
(if sdat
(let* ((ipaddr (car sdat))
(port (cadr sdat))
(servinf (conc (server:get-servinfo-dir *toppath*)"/"ipaddr":"port)))
(debug:print-error 0 *default-log-port* "transport appears to have died, exiting server")
(exit))
(exit)
)
(loop start-time
(equal? sdat last-sdat)
sdat)))))))
(iface (car server-info))
(port (cadr server-info))
(last-access 0)
(server-timeout (server:expiration-timeout))
|
| ︙ | | | ︙ | |
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
|
(set! server-going #t)
(debug:print 0 *default-log-port* "SERVER: running, megatest version: " (common:get-full-version))) ;; NOTE: the server is NOT yet marked as running in the log. We do that in the keep-running routine.
(if (and no-sync-db
(common:low-noise-print 10 "sync-all")) ;; cheesy way to reduce frequency of running sync :)
(begin
(if (common:low-noise-print 120 "sync-all-print")
(debug:print 0 *default-log-port* "keep-running calling db:all-db-sync at " (time->string (seconds->local-time) "%H:%M:%S")))
(db:all-db-sync *dbstruct-dbs*))))
;; when things go wrong we don't want to be doing the various queries too often
;; so we strive to run this stuff only every four seconds or so.
(let* ((sync-time (- (current-milliseconds) start-time))
(rem-time (quotient (- 4000 sync-time) 1000)))
(if (and (<= rem-time 4)
(> rem-time 0))
|
|
>
|
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
|
(set! server-going #t)
(debug:print 0 *default-log-port* "SERVER: running, megatest version: " (common:get-full-version))) ;; NOTE: the server is NOT yet marked as running in the log. We do that in the keep-running routine.
(if (and no-sync-db
(common:low-noise-print 10 "sync-all")) ;; cheesy way to reduce frequency of running sync :)
(begin
(if (common:low-noise-print 120 "sync-all-print")
(debug:print 0 *default-log-port* "keep-running calling db:all-db-sync at " (time->string (seconds->local-time) "%H:%M:%S")))
(db:all-db-sync *dbstruct-dbs*)
)))
;; when things go wrong we don't want to be doing the various queries too often
;; so we strive to run this stuff only every four seconds or so.
(let* ((sync-time (- (current-milliseconds) start-time))
(rem-time (quotient (- 4000 sync-time) 1000)))
(if (and (<= rem-time 4)
(> rem-time 0))
|
| ︙ | | | ︙ | |
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
|
(current-seconds)))
(if (common:low-noise-print 120 "server continuing")
(debug:print-info 0 *default-log-port* "Server continuing, seconds since last db access: " (- (current-seconds) last-access))
(let ((curr-time (current-seconds)))
(handle-exceptions
exn
(debug:print 0 *default-log-port* "ERROR: Failed to change timestamp on info file " servinfofile ". Are you out of space on that disk? exn=" exn)
(if (and (< (- (current-seconds) server-start-time) 600) ;; run for ten minutes for experiment, 3600 thereafter
(not *server-overloaded*)
(file-exists? servinfofile))
(change-file-times servinfofile curr-time curr-time)))
(if (or (common:low-noise-print 120 "start new server")
(> *api-process-request-count* 50)) ;; if this server is kind of busy start up another
(begin
(debug:print-info 0 *default-log-port* "Server is busy, parallel-api-count "*api-process-request-count*", start another if possible...")
(server:kind-run *toppath*)
(if (> *api-process-request-count* 100)
(begin
(debug:print-info 0 *default-log-port* "Server is overloaded at parallel-api-count="*api-process-request-count*", removing "servinfofile)
(delete-file* servinfofile)))))))
(loop 0 server-state bad-sync-count (current-milliseconds)))
(else
(debug:print-info 0 *default-log-port* "Server timed out. seconds since last db access: " (- (current-seconds) last-access))
(http-transport:server-shutdown port)))))))
(define (http-transport:server-shutdown port)
|
|
|
|
|
|
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
|
(current-seconds)))
(if (common:low-noise-print 120 "server continuing")
(debug:print-info 0 *default-log-port* "Server continuing, seconds since last db access: " (- (current-seconds) last-access))
(let ((curr-time (current-seconds)))
(handle-exceptions
exn
(debug:print 0 *default-log-port* "ERROR: Failed to change timestamp on info file " servinfofile ". Are you out of space on that disk? exn=" exn)
(if (and ;; (< (- (current-seconds) server-start-time) 600) ;; run for ten minutes for experiment, 3600 thereafter
(not *server-overloaded*)
(file-exists? servinfofile))
(change-file-times servinfofile curr-time curr-time)))
(if (and (common:low-noise-print 120 "start new server")
(> *api-process-request-count* 50)) ;; if this server is kind of busy start up another
(begin
(debug:print-info 0 *default-log-port* "Server is busy, api-count "*api-process-request-count*", start another if possible...")
(server:kind-run *toppath*)
(if (> *api-process-request-count* 100)
(begin
(debug:print-info 0 *default-log-port* "Server is overloaded at api-count=" *api-process-request-count*", removing "servinfofile)
(delete-file* servinfofile)))))))
(loop 0 server-state bad-sync-count (current-milliseconds)))
(else
(debug:print-info 0 *default-log-port* "Server timed out. seconds since last db access: " (- (current-seconds) last-access))
(http-transport:server-shutdown port)))))))
(define (http-transport:server-shutdown port)
|
| ︙ | | | ︙ | |
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
|
(exit)))
;; all routes though here end in exit ...
;;
;; start_server?
;;
(define (http-transport:launch)
;; check that a server start is in progress, pause or exit if so
(let* ((tmp-area (common:get-db-tmp-area))
(server-start (conc tmp-area "/.server-start"))
(server-started (conc tmp-area "/.server-started"))
(start-time (common:lazy-modification-time server-start))
(started-time (common:lazy-modification-time server-started))
(server-starting (< start-time started-time)) ;; if start-time is less than started-time then a server is still starting
(start-time-old (> (- (current-seconds) start-time) 5))
(cleanup-proc (lambda (msg)
(let* ((serv-fname (conc "server-" (current-process-id) "-" (get-host-name) ".log"))
(full-serv-fname (conc *toppath* "/logs/" serv-fname))
(new-serv-fname (conc *toppath* "/logs/" "defunct-" serv-fname)))
(debug:print 0 *default-log-port* msg)
(if (common:file-exists? full-serv-fname)
(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)
(begin
(cleanup-proc (conc "ERROR: Aborting server start because there are already " num-alive " possible servers either running or starting up"))
(exit))))
#;(common:save-pkt `((action . start)
(T . server)
(pid . ,(current-process-id)))
*configdat* #t)
(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))))
|
|
|
|
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
|
<
<
<
<
<
<
|
|
|
|
<
<
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
|
(exit)))
;; all routes though here end in exit ...
;;
;; start_server?
;;
(define (http-transport:launch)
;; check the .servinfo directory, are there other servers running on this
;; or another host?
(let* ((server-start-is-ok (server:minimal-check *toppath*)))
(if (not server-start-is-ok)
(begin
(debug:print 0 *default-log-port* "ERROR: server start not ok, exiting now.")
(exit 1))))
;; check that a server start is in progress, pause or exit if so
(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 ()
|
| ︙ | | | ︙ | |