33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
(import debugprint)
(import tcp-transportmod)
(use srfi-69
srfi-18
posix
matchable
s11n)
;; allow these queries through without starting a server
;;
(define api:read-only-queries
'(get-key-val-pairs
get-var
get-keys
|
|
>
|
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
|
(import debugprint)
(import tcp-transportmod)
(use srfi-69
srfi-18
posix
matchable
s11n
typed-records)
;; allow these queries through without starting a server
;;
(define api:read-only-queries
'(get-key-val-pairs
get-var
get-keys
|
93
94
95
96
97
98
99
100
101
102
103
104
105
106
|
read-test-data
read-test-data-varpatt
login
tasks-get-last
testmeta-get-record
have-incompletes?
get-changed-record-ids
get-run-record-ids
get-not-completed-cnt))
(define api:write-queries
'(
get-keys-write ;; dummy "write" query to force server start
|
>
>
>
|
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
|
read-test-data
read-test-data-varpatt
login
tasks-get-last
testmeta-get-record
have-incompletes?
get-changed-record-ids
get-all-runids
get-changed-record-test-ids
get-changed-record-run-ids
get-run-record-ids
get-not-completed-cnt))
(define api:write-queries
'(
get-keys-write ;; dummy "write" query to force server start
|
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
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
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
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
|
;; TASKS
tasks-add
tasks-set-state-given-param-key
))
(define *db-write-mutexes* (make-hash-table))
(define *server-signature* #f)
;; ;; These are called by the server on recipt of /api calls
;; ;; - keep it simple, only return the actual result of the call, i.e. no meta info here
;; ;;
;; ;; - returns #( flag result )
;; ;;
;; (define (api:execute-requests dbstruct dat)
;; (if (> *api-process-request-count* 50)
;; (begin
;; (if (common:low-noise-print 30 "too many threads")
;; (debug:print 0 *default-log-port* "WARNING: "*api-process-request-count*" threads, potential overload, adding 0.5 sec delay."))
;; ;; (thread-sleep! 0.5) ;; take a nap - no, the napping is moved to the clients via tt:backoff-incr
;; ))
;; (cond
;; ((not (vector? dat)) ;; it is an error to not receive a vector
;; (vector #f (vector #f "remote must be called with a vector")))
;; (else
;; (let* ((cmd-in (vector-ref dat 0))
;; (cmd (if (symbol? cmd-in)
;; cmd-in
;; (string->symbol cmd-in)))
;; (params (vector-ref dat 1))
;; (run-id (if (null? params)
;; 0
;; (car params)))
;; (write-mutex (if (hash-table-exists? *db-write-mutexes* run-id)
;; (hash-table-ref *db-write-mutexes* run-id)
;; (let* ((newmutex (make-mutex)))
;; (hash-table-set! *db-write-mutexes* run-id newmutex)
;; newmutex)))
;; (start-t (current-milliseconds))
;; (readonly-mode (dbr:dbstruct-read-only dbstruct))
;; (readonly-command (member cmd api:read-only-queries))
;; (writecmd-in-readonly-mode (and readonly-mode (not readonly-command))))
;; (if (not readonly-command)
;; (mutex-lock! write-mutex))
;; (let* ((tmppath (dbr:dbstruct-tmppath dbstruct))
;; (clean-run-id (cond
;; ((number? run-id) run-id)
;; ((equal? run-id #f) "main")
;; (else "other")))
;; (crumbfile (dbfile:wait-for-qif tmppath clean-run-id (cons cmd params)))
;; (res
;; (if writecmd-in-readonly-mode
;; (conc "attempt to run write command "cmd" on a read-only database")
;; (api:dispatch-request dbstruct cmd run-id params))))
;; (delete-file* crumbfile)
;; (if (not readonly-command)
;; (mutex-unlock! write-mutex))
;;
;; ;; save all stats
;; (let ((delta-t (- (current-milliseconds)
;; start-t))
;; (modified-cmd (if (eq? cmd 'general-call)
;; (string->symbol (conc "general-call-" (car params)))
;; cmd)))
;; (hash-table-set! *db-api-call-time* modified-cmd
;; (cons delta-t (hash-table-ref/default *db-api-call-time* modified-cmd '()))))
;; (if writecmd-in-readonly-mode
;; (begin
;; #;(common:telemetry-log (conc "api-out:"(->string cmd))
;; payload: `((params . ,params)
;; (ok-res . #t)))
;; (vector #f res))
;; (begin
;; #;(common:telemetry-log (conc "api-out:"(->string cmd))
;; payload: `((params . ,params)
;; (ok-res . #f)))
;; (vector #t res))))))))
(define (api:register-thread th-in)
(set! *api-threads* (cons (cons th-in (current-seconds)) *api-threads*)))
(define (api:unregister-thread th-in)
(set! *api-threads* (filter (lambda (thdat)
(not (eq? th-in (car thdat))))
*api-threads*)))
(define (api:remove-dead-or-terminated)
(set! *api-threads* (filter (lambda (thdat)
(not (member (thread-state (car thdat)) '(terminated dead))))
*api-threads*)))
(define (api:get-count-threads-alive)
(length *api-threads*))
(api:register-thread (current-thread))
(let* (;; (indat (deserialize))
(newcount (+ *api-process-request-count* 1))
(numthreads (api:get-count-threads-alive))
(delay-wait (if (> newcount 10)
(- newcount 10)
0))
(normal-proc (lambda (cmd run-id params)
(case cmd
((ping) *server-signature*)
(else
(api:dispatch-request dbstruct cmd run-id params))))))
(set! *api-process-request-count* newcount)
(set! *db-last-access* (current-seconds))
(if (not (eq? newcount numthreads))
(begin
(api:remove-dead-or-terminated)
(let ((threads-now (api:get-count-threads-alive)))
(debug:print 0 *default-log-port* "WARNING: newcount="newcount", numthreads="numthreads", remaining="threads-now)
(set! newcount threads-now))))
(match indat
((cmd run-id params meta)
(let* ((db-ok (let* ((dbfname (dbmod:run-id->dbfname run-id))
(ok (equal? dbfname (dbr:dbstruct-dbfname dbstruct))))
(case cmd
((ping) #t) ;; we are fine
(else
(if (not ok)(debug:print 0 *default-log-port* "ERROR: "cmd", run-id "run-id", not correct for dbfname "(dbr:dbstruct-dbfname dbstruct)))
(assert ok "FATAL: database file and run-id not aligned.")))))
(ttdat *server-info*)
(server-state (tt-state ttdat))
(status (cond
((> newcount 3) 'busy)
;; ((> newcount 5) 'loaded) ;; this gets transmitted to the client which calls tt:backoff-incr to slow stuff down.
(else 'ok)))
(errmsg (case status
((busy) (conc "Server overloaded, "newcount" threads in flight"))
((loaded) (conc "Server loaded, "newcount" threads in flight"))
(else #f)))
(result (case status
((busy)
(if (eq? cmd 'ping)
(normal-proc cmd run-id params)
;; newcount must be greater than 5 for busy
(* 1 (- newcount 3)) ;; was 15
)) ;; (- newcount 29)) ;; call back in as many seconds
((loaded)
;; (if (eq? (rmt:transport-mode) 'tcp)
;; (thread-sleep! 0.5))
(normal-proc cmd run-id params))
(else
(normal-proc cmd run-id params))))
(meta (case cmd
((ping) `((sstate . ,server-state)))
(else `((wait . ,delay-wait)))))
(payload (list status errmsg result meta)))
(set! *api-process-request-count* (- *api-process-request-count* 1))
;; (serialize payload)
(api:unregister-thread (current-thread))
payload))
(else
(assert #f "FATAL: failed to deserialize indat "indat))))))
((start-server) (apply server:kind-run params))
((kill-server) (set! *server-run* #f))
;; TESTS
;;((test-set-state-status-by-id) (apply mt:test-set-state-status-by-id dbstruct params))
;;BB - commented out above because it was calling below, eventually, incorrectly (dbstruct passed to mt:test-set-state-status-by-id, which previosly did more, but now only passes thru to db:set-state-status-and-roll-up-items.
((test-set-state-status-by-id)
|
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
|
|
>
>
>
>
>
>
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
|
|
<
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
>
|
|
|
|
|
<
|
|
|
>
|
>
>
>
>
>
>
>
>
|
|
>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
<
|
<
>
|
|
|
>
>
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
|
|
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
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
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
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
|
;; TASKS
tasks-add
tasks-set-state-given-param-key
))
(define *db-write-mutexes* (make-hash-table))
(define *server-signature* #f)
(define *api-threads* '())
(define (api:register-thread th-in command)
(set! *api-threads* (cons (list th-in (current-seconds) command) *api-threads*)))
(define (api:get-thread-command th-in)
(let ((thread-data (assoc th-in *api-threads*)))
(if thread-data
(third thread-data) ; Assuming the command is the third element in the list
#f))) ; Return #f if the thread is not found
(define (api:unregister-thread th-in)
(set! *api-threads* (filter (lambda (thdat)
(not (eq? th-in (car thdat))))
*api-threads*)))
(define (api:remove-dead-or-terminated)
(set! *api-threads* (filter (lambda (thdat)
(not (member (thread-state (car thdat)) '(terminated dead))))
*api-threads*)))
(define (api:get-count-threads-alive)
(length *api-threads*))
(define (api:get-threads)
(map (lambda (thdat)
(let ((thread (first thdat))
(timestamp (second thdat))
(command (third thdat)))
(format "\nThread: ~a, age: ~a, Command: ~a" thread (- (current-seconds) timestamp) command)))
*api-threads*))
(define *api:last-stats-print* 0)
(define *api-print-db-stats-mutex* (make-mutex))
(define (api:print-db-stats)
(debug:print-info 0 *default-log-port* "Started periodic db stats printer")
(let loop ()
(mutex-lock! *api-print-db-stats-mutex*)
(if (> (- (current-seconds) *api:last-stats-print*) 15)
(begin
(rmt:print-db-stats)
(set! *api:last-stats-print* (current-seconds))))
(mutex-unlock! *api-print-db-stats-mutex*)
(thread-sleep! 5)
(loop)))
;; indat is (cmd run-id params meta)
;;
;; WARNING: Do not print anything in the lambda of this function as it
;; reads/writes to current in/out port
;;
(define (api:tcp-dispatch-request-make-handler dbstruct) ;; cmd run-id params)
(assert *toppath* "FATAL: api:tcp-dispatch-request-make-handler called but *toppath* not set.")
(if (not *server-signature*)
(set! *server-signature* (tt:mk-signature *toppath*)))
(lambda (indat)
(api:register-thread (current-thread) (car indat))
(let* ((result
(let* ((numthreads (api:get-count-threads-alive))
(delay-wait (if (> numthreads 10)
(- numthreads 10)
0))
(normal-proc (lambda (cmd run-id params)
(case cmd
((ping) *server-signature*)
(else
(api:dispatch-request dbstruct cmd run-id params))))))
(set! *api-process-request-count* numthreads)
(set! *db-last-access* (current-seconds))
;; (if (not (eq? numthreads numthreads))
;; (begin
;; (api:remove-dead-or-terminated)
;; (let ((threads-now (api:get-count-threads-alive)))
;; (debug:print 0 *default-log-port* "WARNING: numthreads="numthreads", numthreads="numthreads", remaining="threads-now)
;; (set! numthreads threads-now))))
(match indat
((cmd run-id params meta)
(let* ((start-t (current-milliseconds))
(db-ok (let* ((dbfname (dbmod:run-id->dbfname run-id))
(ok (equal? dbfname (dbr:dbstruct-dbfname dbstruct))))
(case cmd
((ping) #t) ;; we are fine
(else
(assert ok "FATAL: database file and run-id not aligned.")))))
(ttdat *server-info*)
(server-state (tt-state ttdat))
(maxthreads 20) ;; make this a parameter?
(status (cond
((> numthreads maxthreads)
(let* ((testsuite (common:get-testsuite-name))
(mtexe (common:find-local-megatest))
(proc (lambda ()
;; we are overloaded, try to start another server
(debug:print 0 *default-log-port* "Too many threads running, starting another server")
(tt:server-process-run *toppath* testsuite mtexe run-id))))
(set! *server-start-requests* (cons proc *server-start-requests*)))
;; 'busy
'loaded ;; not ideal since the client will not backoff
)
(else 'ok)))
(errmsg (case status
((busy) (conc "Server overloaded, "numthreads" threads in flight, current cmd: " cmd "\n current threads: " (api:get-threads)))
((loaded) (conc "Server loaded, "numthreads" threads in flight"))
(else #f)))
(result (case status
((busy)
(if (eq? cmd 'ping)
(normal-proc cmd run-id params)
;; numthreads must be greater than 5 for busy
(* 0.1 (- numthreads maxthreads)) ;; was 15
)) ;; (- numthreads 29)) ;; call back in as many seconds
((loaded)
;; (if (eq? (rmt:transport-mode) 'tcp)
;; (thread-sleep! 0.5))
(normal-proc cmd run-id params))
(else
(normal-proc cmd run-id params))))
(meta (case cmd
((ping) `((sstate . ,server-state)(sload . ,numthreads)))
(else `((wait . ,delay-wait)))))
(payload (list status errmsg result meta)))
;; (cmd run-id params meta)
(db:add-stats cmd run-id params (- (current-milliseconds) start-t))
payload))
(else
(assert #f "FATAL: failed to deserialize indat "indat))))))
;; (set! *api-process-request-count* (- *api-process-request-count* 1))
;; (serialize payload)
(api:unregister-thread (current-thread))
result)))
(define *api-halt-writes* #f)
(define (api:dispatch-request dbstruct cmd run-id params)
(if (not *no-sync-db*)
(db:open-no-sync-db))
(let* ((start-time (current-milliseconds)))
(if (member cmd api:write-queries)
(let loop ()
(if *api-halt-writes*
(begin
(thread-sleep! 0.2)
(if (< (- (current-milliseconds) start-time)
5000) ;; hope it don't take more than five seconds to sync
(loop-time)
#;(debug:print 0 *default-log-port* "ERROR: writes halted for more than 5 seconds, sync might be taking too long"))))))
(db:add-stats 'api-write-blocking-for-sync run-id params (- (current-milliseconds) start-time)))
(case cmd
;;===============================================
;; READ/WRITE QUERIES
;;===============================================
((get-keys-write) (db:get-keys dbstruct)) ;; force a dummy "write" query to force server; for debug in -repl
;; SERVERS
((start-server) (apply tt:server-process-run params))
((kill-server) (set! *server-run* #f))
;; TESTS
;;((test-set-state-status-by-id) (apply mt:test-set-state-status-by-id dbstruct params))
;;BB - commented out above because it was calling below, eventually, incorrectly (dbstruct passed to mt:test-set-state-status-by-id, which previosly did more, but now only passes thru to db:set-state-status-and-roll-up-items.
((test-set-state-status-by-id)
|
398
399
400
401
402
403
404
405
406
407
408
409
410
411
|
((tasks-get-last) (apply tasks:get-last dbstruct params))
;; NO SYNC DB
((no-sync-set) (apply db:no-sync-set *no-sync-db* params))
((no-sync-get/default) (apply db:no-sync-get/default *no-sync-db* params))
((no-sync-del!) (apply db:no-sync-del! *no-sync-db* params))
((no-sync-get-lock) (apply db:no-sync-get-lock *no-sync-db* params))
;; ARCHIVES
;; ((archive-get-allocations)
((archive-register-disk) (apply db:archive-register-disk dbstruct params))
((archive-register-block-name)(apply db:archive-register-block-name dbstruct params))
;; ((archive-allocate-testsuite/area-to-block)(apply db:archive-allocate-testsuite/area-to-block dbstruct block-id testsuite-name areakey))
|
>
>
>
>
>
>
|
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
|
((tasks-get-last) (apply tasks:get-last dbstruct params))
;; NO SYNC DB
((no-sync-set) (apply db:no-sync-set *no-sync-db* params))
((no-sync-get/default) (apply db:no-sync-get/default *no-sync-db* params))
((no-sync-del!) (apply db:no-sync-del! *no-sync-db* params))
((no-sync-get-lock) (apply db:no-sync-get-lock *no-sync-db* params))
;; NO SYNC DB PROCESSES
((register-process) (apply dbfile:register-process *no-sync-db* params))
((set-process-done) (apply dbfile:set-process-done *no-sync-db* params))
((set-process-status) (apply dbfile:set-process-status *no-sync-db* params))
((get-process-options) (apply dbfile:get-process-options *no-sync-db* params))
;; ARCHIVES
;; ((archive-get-allocations)
((archive-register-disk) (apply db:archive-register-disk dbstruct params))
((archive-register-block-name)(apply db:archive-register-block-name dbstruct params))
;; ((archive-allocate-testsuite/area-to-block)(apply db:archive-allocate-testsuite/area-to-block dbstruct block-id testsuite-name areakey))
|
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
|
((general-call) (let ((stmtname (car params))
(run-id (cadr params))
(realparams (cddr params)))
(db:general-call dbstruct run-id stmtname realparams)))
((sdb-qry) (apply sdb:qry params))
((ping) (current-process-id))
((get-changed-record-ids) (apply db:get-changed-record-ids dbstruct params))
((get-run-record-ids) (apply db:get-run-record-ids dbstruct params))
;; TESTMETA
((testmeta-get-record) (apply db:testmeta-get-record dbstruct params))
;; TASKS
((find-task-queue-records) (apply tasks:find-task-queue-records dbstruct params))
(else
(debug:print 0 *default-log-port* "ERROR: bad api call " cmd)
(conc "ERROR: BAD api call " cmd))))
;; http-server send-response
;; api:process-request
;; db:*
;;
;; NB// Runs on the server as part of the server loop
;;
(define (api:process-request dbstruct $) ;; the $ is the request vars proc
(debug:print 4 *default-log-port* "server-id:" *server-id*)
(let* ((cmd ($ 'cmd))
(paramsj ($ 'params))
(key ($ 'key))
(params (db:string->obj paramsj transport: 'http))) ;; incoming data from the POST (or is it a GET?)
(debug:print 4 *default-log-port* "cmd:" cmd " with params " params "key " key)
(if (equal? key *server-id*)
(begin
(set! *api-process-request-count* (+ *api-process-request-count* 1))
(let* ((resdat (api:execute-requests dbstruct (vector cmd params))) ;; process the request, resdat = #( flag result )
(success (vector-ref resdat 0))
(res (vector-ref resdat 1))) ;; (vector flag payload), get the payload, ignore the flag (why?)
(debug:print 4 *default-log-port* "res:" res)
(if (not success)
(debug:print 0 *default-log-port* "ERROR: success flag is #f for " cmd " with params " params))
(if (> *api-process-request-count* *max-api-process-requests*)
(set! *max-api-process-requests* *api-process-request-count*))
(set! *api-process-request-count* (- *api-process-request-count* 1))
;; This can be here but needs controls to ensure it doesn't run more than every 4 seconds
;; (rmt:dat->json-str
;; (if (or (string? res)
;; (list? res)
;; (number? res)
;; (boolean? res))
;; res
;; (list "ERROR, not string, list, number or boolean" 1 cmd params res)))))
(db:obj->string res transport: 'http)))
(begin
(debug:print 0 *default-log-port* "Server refused to process request. Server id mismatch. recived " key " expected: " *server-id* ".\nOther arguments recived: cmd=" cmd " params = " params)
(db:obj->string (conc "Server refused to process request server-id mismatch: " key ", " *server-id*) transport: 'http)))))
|
>
>
|
>
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
|
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
|
((general-call) (let ((stmtname (car params))
(run-id (cadr params))
(realparams (cddr params)))
(db:general-call dbstruct run-id stmtname realparams)))
((sdb-qry) (apply sdb:qry params))
((ping) (current-process-id))
((get-changed-record-ids) (apply db:get-changed-record-ids dbstruct params))
((get-changed-record-test-ids) (apply db:get-changed-record-test-ids dbstruct params))
((get-changed-record-run-ids) (apply db:get-changed-record-run-ids dbstruct params))
((get-run-record-ids) (apply db:get-run-record-ids dbstruct params))
((get-all-runids) (apply db:get-all-runids dbstruct))
;; TESTMETA
((testmeta-get-record) (apply db:testmeta-get-record dbstruct params))
;; TASKS
((find-task-queue-records) (apply tasks:find-task-queue-records dbstruct params))
(else
(debug:print 0 *default-log-port* "ERROR: bad api call " cmd)
(conc "ERROR: BAD api call " cmd))))
|