Megatest

Changes On Branch 71a141e42eef4f86
Login

Changes In Branch v1.65-wip Through [71a141e42e] Excluding Merge-Ins

This is equivalent to a diff from 8a70b57bea to 71a141e42e

2019-10-21
14:00
Added defense against directories in the logs dir. check-in: 3eaa18cb5b user: mrwellan tags: v1.65
2019-10-02
00:05
still broke but differently check-in: 0869d10a93 user: matt tags: v1.65-wip
2019-10-01
22:54
Broke it with mutex deadlock check-in: 71a141e42e user: matt tags: v1.65-wip
22:35
Removed more globals from rmt:send-receive-orig check-in: 386b7b1848 user: matt tags: v1.65-wip
21:38
Merged in v1.65 check-in: a114850abb user: matt tags: v1.65-wip
10:40
Fix for Makefile for mtexec check-in: 8a70b57bea user: mrwellan tags: v1.65
09:52
Merge fix for periods in env vars from trunk check-in: 65e88e7bdc user: mrwellan tags: v1.65

Modified common.scm from [77eb320f92] to [91c6910f87].

277
278
279
280
281
282
283
284


285
286
287
288
289
290
291
  (server-url        (if *toppath* (server:check-if-running *toppath*))) ;; (server:check-if-running *toppath*) #f))
  (last-server-check 0)  ;; last time we checked to see if the server was alive
  (conndat           #f)
  (transport         *transport-type*)
  (server-timeout    (server:expiration-timeout))
  (force-server      #f)
  (ro-mode           #f)  
  (ro-mode-checked   #f)) ;; flag that indicates we have checked for ro-mode



;; launching and hosts
(defstruct host
  (reachable    #f)
  (last-update  0)
  (last-used    0)
  (last-cpuload 1))







|
>
>







277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
  (server-url        (if *toppath* (server:check-if-running *toppath*))) ;; (server:check-if-running *toppath*) #f))
  (last-server-check 0)  ;; last time we checked to see if the server was alive
  (conndat           #f)
  (transport         *transport-type*)
  (server-timeout    (server:expiration-timeout))
  (force-server      #f)
  (ro-mode           #f)  
  (ro-mode-checked   #f) ;; flag that indicates we have checked for ro-mode
  (ulex:conn         #f) ;; ulex db conn is not exactly a db connector, more like a network connector 
  )

;; launching and hosts
(defstruct host
  (reachable    #f)
  (last-update  0)
  (last-used    0)
  (last-cpuload 1))

Modified db.scm from [560d632862] to [5cbdd1ef19].

302
303
304
305
306
307
308






309
310
311
312
313
314
315
316
317
318
	       ;(mtdbmodtime  (if mtdbexists (common:lazy-sqlite-db-modification-time mtdbpath)   #f)) ; moving this before db:open-megatest-db is called. if wal mode is on -WAL and -shm file get created with causing the  tmpdbmodtime timestamp always greater than mtdbmodtime
	       ;(tmpdbmodtime (if dbfexists  (common:lazy-sqlite-db-modification-time tmpdbfname) #f)) 
					;if wal mode is on -WAL and -shm file get created when db:open-megatest-db is called. modtimedelta will always be < 10 so db in tmp not get synced
          ;(tmpdbmodtime (if dbfexists (db:get-last-update-time (car tmpdb)) #f))    
          ;(fmt (file-modification-time tmpdbfname))
	       (modtimedelta (and mtdbmodtime tmpdbmodtime (- mtdbmodtime tmpdbmodtime))))







          (when write-access
            (sqlite3:execute (car mtdb) "drop trigger if exists update_tests_trigger")
            (sqlite3:execute (car mtdb) "drop trigger if exists update_runs_trigger"))
          
         ;(print "mtdbmodtime " mtdbmodtime " tmpdbmodtime " tmpdbmodtime " mtdbpath " mtdbpath " " (conc *toppath* "/megatest.db"))
	        ;;(debug:print-info 13 *default-log-port* "db:open-db>> mtdbpath="mtdbpath" mtdbexists="mtdbexists" and write-access="write-access)
          (if (and dbexists (not write-access))
              (begin
                (set! *db-write-access* #f)
                (dbr:dbstruct-read-only-set! dbstruct #t)))







>
>
>
>
>
>
|
|
|







302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
	       ;(mtdbmodtime  (if mtdbexists (common:lazy-sqlite-db-modification-time mtdbpath)   #f)) ; moving this before db:open-megatest-db is called. if wal mode is on -WAL and -shm file get created with causing the  tmpdbmodtime timestamp always greater than mtdbmodtime
	       ;(tmpdbmodtime (if dbfexists  (common:lazy-sqlite-db-modification-time tmpdbfname) #f)) 
					;if wal mode is on -WAL and -shm file get created when db:open-megatest-db is called. modtimedelta will always be < 10 so db in tmp not get synced
          ;(tmpdbmodtime (if dbfexists (db:get-last-update-time (car tmpdb)) #f))    
          ;(fmt (file-modification-time tmpdbfname))
	       (modtimedelta (and mtdbmodtime tmpdbmodtime (- mtdbmodtime tmpdbmodtime))))

	  (handle-exceptions
	   exn
	   (let ((call-chain (get-call-chain))
		 (msg        ((condition-property-accessor 'exn 'message) exn)))
	     (debug:print 0 *default-log-port* "ERROR: attempted to drop triggers on MTRA/megatest.db but failed. Error is " msg)
	     (set! write-access #f)) ;; if we failed to drop the triggers then we probably don't have write access
	   (when write-access
		 (sqlite3:execute (car mtdb) "drop trigger if exists update_tests_trigger")
		 (sqlite3:execute (car mtdb) "drop trigger if exists update_runs_trigger")))
          
         ;(print "mtdbmodtime " mtdbmodtime " tmpdbmodtime " tmpdbmodtime " mtdbpath " mtdbpath " " (conc *toppath* "/megatest.db"))
	        ;;(debug:print-info 13 *default-log-port* "db:open-db>> mtdbpath="mtdbpath" mtdbexists="mtdbexists" and write-access="write-access)
          (if (and dbexists (not write-access))
              (begin
                (set! *db-write-access* #f)
                (dbr:dbstruct-read-only-set! dbstruct #t)))

Modified rmt.scm from [5054b48a41] to [5ad0893a6b].

23
24
25
26
27
28
29

30
31
32
33
34
35
36
(declare (unit rmt))
(declare (uses api))
(declare (uses http-transport))
(include "common_records.scm")
(declare (uses rmtmod))

(import rmtmod)


;;
;; THESE ARE ALL CALLED ON THE CLIENT SIDE!!!
;;

;; generate entries for ~/.megatestrc with the following
;;







>







23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
(declare (unit rmt))
(declare (uses api))
(declare (uses http-transport))
(include "common_records.scm")
(declare (uses rmtmod))

(import rmtmod)


;;
;; THESE ARE ALL CALLED ON THE CLIENT SIDE!!!
;;

;; generate entries for ~/.megatestrc with the following
;;
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
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
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
	      cinfo
	      (if (server:check-if-running areapath)
		  (client:setup areapath)
		  #f))))

(define *send-receive-mutex* (make-mutex)) ;; should have separate mutex per run-id
























;; RA => e.g. usage (rmt:send-receive 'get-var #f (list varname))
;;


(define (rmt:send-receive cmd rid params #!key (attemptnum 1)(area-dat #f)) ;; start attemptnum at 1 so the modulo below works as expected

  (common:telemetry-log (conc "rmt:"(->string cmd))
                        payload: `((rid . ,rid)
                                   (params . ,params)))
                          
  
  ;;DOT digraph megatest_state_status {
  ;;DOT   ranksep=0;
  ;;DOT   // rankdir=LR;
  ;;DOT   node [shape="box"];
  ;;DOT "rmt:send-receive" -> MUTEXLOCK;
  ;;DOT { edge [style=invis];"case 1" -> "case 2" -> "case 3" -> "case 4" -> "case 5" -> "case 6" -> "case 7" -> "case 8" -> "case 9" -> "case 10" -> "case 11"; }
  ;; do all the prep locked under the rmt-mutex
  (mutex-lock! *rmt-mutex*)
  
  ;; 1. check if server is started IFF cmd is a write OR if we are not on the homehost, store in runremote
  ;; 2. check the age of the connections. refresh the connection if it is older than timeout-20 seconds.
  ;; 3. do the query, if on homehost use local access
  ;;
  (let* ((start-time    (current-seconds)) ;; snapshot time so all use cases get same value
         (areapath      *toppath*);; TODO - resolve from dbstruct to be compatible with multiple areas
	 (runremote     (or area-dat
			    *runremote*))
	 (readonly-mode (rmtmod:calc-ro-mode runremote *toppath*)))

    ;; DOT INIT_RUNREMOTE; // leaving off - doesn't really add to the clarity
    ;; DOT MUTEXLOCK -> INIT_RUNREMOTE [label="no remote?"];
    ;; DOT INIT_RUNREMOTE -> MUTEXLOCK;
    ;; ensure we have a record for our connection for given area
    (if (not runremote)                   ;; can remove this one. should never get here.         
	(begin
	  (set! *runremote* (make-remote))
	  (set! runremote   *runremote*))) ;; new runremote will come from this on next iteration
    
    ;; DOT SET_HOMEHOST; // leaving off - doesn't really add to the clarity
    ;; DOT MUTEXLOCK -> SET_HOMEHOST [label="no homehost?"];
    ;; DOT SET_HOMEHOST -> MUTEXLOCK;
    ;; ensure we have a homehost record
    (if (not (pair? (remote-hh-dat runremote)))  ;; not on homehost
	(thread-sleep! 0.1) ;; since we shouldn't get here, delay a little
	(remote-hh-dat-set! runremote (common:get-homehost)))
    
    ;;(print "BB> readonly-mode is "readonly-mode" dbfile is "dbfile)
    (cond
     ;;DOT EXIT;
     ;;DOT MUTEXLOCK -> EXIT [label="> 15 attempts"]; {rank=same "case 1" "EXIT" }
     ;; give up if more than 15 attempts
     ((> attemptnum 15)
      (debug:print 0 *default-log-port* "ERROR: 15 tries to start/connect to server. Giving up.")
      (exit 1))

     ;;DOT CASE2 [label="local\nreadonly\nquery"];
     ;;DOT MUTEXLOCK -> CASE2; {rank=same "case 2" CASE2}
     ;;DOT CASE2 -> "rmt:open-qry-close-locally";
     ;; readonly mode, read request-  handle it - case 2
     ((and readonly-mode
           (member cmd api:read-only-queries)) 
      (mutex-unlock! *rmt-mutex*)
      (debug:print-info 12 *default-log-port* "rmt:send-receive, case 2")
      (rmt:open-qry-close-locally cmd 0 params)
      )

     ;;DOT CASE3 [label="write in\nread-only mode"];
     ;;DOT MUTEXLOCK -> CASE3 [label="readonly\nmode?"]; {rank=same "case 3" CASE3}
     ;;DOT CASE3 -> "#f";
     ;; readonly mode, write request.  Do nothing, return #f
     (readonly-mode (extras-readonly-mode *rmt-mutex* *default-log-port* cmd params))

     ;; This block was for pre-emptively resetting the connection if there had been no communication for some time.
     ;; I don't think it adds any value. If the server is not there, just fail and start a new connection.
     ;; also, the expire-time calculation might not be correct. We want, time-since-last-server-access > (server:get-timeout)
     ;;
     ;;DOT CASE4 [label="reset\nconnection"];
     ;;DOT MUTEXLOCK -> CASE4 [label="have connection,\nlast_access > expire_time"]; {rank=same "case 4" CASE4}
     ;;DOT CASE4 -> "rmt:send-receive";
     ;; reset the connection if it has been unused too long
     ((and runremote
           (remote-conndat runremote)
	   (> (current-seconds) ;; if it has been more than server-timeout seconds since last contact, close this connection and start a new on
	      (+ (http-transport:server-dat-get-last-access (remote-conndat runremote))
		 (remote-server-timeout runremote))))
      (debug:print-info 0 *default-log-port* "Connection to " (remote-server-url runremote) " expired due to no accesses, forcing new connection.")
      (http-transport:close-connections area-dat: runremote)
      (remote-conndat-set! runremote #f) ;; invalidate the connection, thus forcing a new connection.
      (mutex-unlock! *rmt-mutex*)
      (rmt:send-receive cmd rid params attemptnum: attemptnum))
     
     ;;DOT CASE5 [label="local\nread"];
     ;;DOT MUTEXLOCK -> CASE5 [label="server not required,\non homehost,\nread-only query"]; {rank=same "case 5" CASE5};
     ;;DOT CASE5 -> "rmt:open-qry-close-locally";

     ;; on homehost and this is a read
     ((and (not (remote-force-server runremote)) ;; honor forced use of server, i.e. server NOT required
	   (cdr (remote-hh-dat runremote))       ;; on homehost
           (member cmd api:read-only-queries))   ;; this is a read
      (mutex-unlock! *rmt-mutex*)
      (debug:print-info 12 *default-log-port* "rmt:send-receive, case  5")
      (rmt:open-qry-close-locally cmd 0 params))

     ;;DOT CASE6 [label="init\nremote"];
     ;;DOT MUTEXLOCK -> CASE6 [label="on homehost,\nwrite query,\nhave server,\ncan't reach it"]; {rank=same "case 6" CASE6};
     ;;DOT CASE6 -> "rmt:send-receive";
     ;; on homehost and this is a write, we already have a server, but server has died
     ((and (cdr (remote-hh-dat runremote))           ;; on homehost
           (not (member cmd api:read-only-queries))  ;; this is a write
           (remote-server-url runremote)             ;; have a server
           (not (server:ping (remote-server-url runremote))))  ;; server has died. NOTE: this is not a cheap call! Need better approach.
      (set! *runremote* (make-remote))
      (remote-force-server-set! runremote (common:force-server?))
      (mutex-unlock! *rmt-mutex*)
      (debug:print-info 12 *default-log-port* "rmt:send-receive, case  6")
      (rmt:send-receive cmd rid params attemptnum: attemptnum))

     ;;DOT CASE7 [label="homehost\nwrite"];
     ;;DOT MUTEXLOCK -> CASE7 [label="server not required,\non homehost,\na write,\nhave a server"]; {rank=same "case 7" CASE7};
     ;;DOT CASE7 -> "rmt:open-qry-close-locally";
     ;; on homehost and this is a write, we already have a server
     ((and (not (remote-force-server runremote))     ;; honor forced use of server, i.e. server NOT required
	   (cdr (remote-hh-dat runremote))           ;; on homehost
           (not (member cmd api:read-only-queries))  ;; this is a write
           (remote-server-url runremote))            ;; have a server
      (mutex-unlock! *rmt-mutex*)
      (debug:print-info 12 *default-log-port* "rmt:send-receive, case  4.1")
      (rmt:open-qry-close-locally cmd 0 params))

     ;;DOT CASE8 [label="force\nserver"];
     ;;DOT MUTEXLOCK -> CASE8 [label="server not required,\nhave homehost info,\nno connection yet,\nnot a read-only query"]; {rank=same "case 8" CASE8};
     ;;DOT CASE8 -> "rmt:open-qry-close-locally";
     ;;  on homehost, no server contact made and this is a write, passively start a server 
     ((and (not (remote-force-server runremote))     ;; honor forced use of server, i.e. server NOT required
	   (cdr (remote-hh-dat runremote))           ;; have homehost
           (not (remote-server-url runremote))       ;; no connection yet
	   (not (member cmd api:read-only-queries))) ;; not a read-only query
      (debug:print-info 12 *default-log-port* "rmt:send-receive, case  8")
      (let ((server-url  (server:check-if-running *toppath*))) ;; (server:read-dotserver->url *toppath*))) ;; (server:check-if-running *toppath*))) ;; Do NOT want to run server:check-if-running - very expensive to do for every write call
	(if server-url
	    (remote-server-url-set! runremote server-url) ;; the string can be consumed by the client setup if needed
	    (if (common:force-server?)
		(server:start-and-wait *toppath*)
		(server:kind-run *toppath*))))
      (remote-force-server-set! runremote (common:force-server?))
      (mutex-unlock! *rmt-mutex*)
      (debug:print-info 12 *default-log-port* "rmt:send-receive, case  8.1")
      (rmt:open-qry-close-locally cmd 0 params))

     ;;DOT CASE9 [label="force server\nnot on homehost"];
     ;;DOT MUTEXLOCK -> CASE9 [label="no connection\nand either require server\nor not on homehost"]; {rank=same "case 9" CASE9};
     ;;DOT CASE9 -> "start\nserver" -> "rmt:send-receive";
     ((or (and (remote-force-server runremote)              ;; we are forcing a server and don't yet have a connection to one
	       (not (remote-conndat runremote)))
	  (and (not (cdr (remote-hh-dat runremote)))        ;; not on a homehost 
	       (not (remote-conndat runremote))))           ;; and no connection
      (debug:print-info 12 *default-log-port* "rmt:send-receive, case 9, hh-dat: " (remote-hh-dat runremote) " conndat: " (remote-conndat runremote))
      (mutex-unlock! *rmt-mutex*)
      (if (not (server:check-if-running *toppath*)) ;; who knows, maybe one has started up?
	  (server:start-and-wait *toppath*))
      (remote-conndat-set! runremote (rmt:get-connection-info *toppath*)) ;; calls client:setup which calls client:setup-http
      (rmt:send-receive cmd rid params attemptnum: attemptnum)) ;; TODO: add back-off timeout as

     ;;DOT CASE10 [label="on homehost"];
     ;;DOT MUTEXLOCK -> CASE10 [label="server not required,\non homehost"]; {rank=same "case 10" CASE10};
     ;;DOT CASE10 -> "rmt:open-qry-close-locally";
     ;; all set up if get this far, dispatch the query
     ((and (not (remote-force-server runremote))
	   (cdr (remote-hh-dat runremote))) ;; we are on homehost
      (mutex-unlock! *rmt-mutex*)
      (debug:print-info 12 *default-log-port* "rmt:send-receive, case 10")
      (rmt:open-qry-close-locally cmd (if rid rid 0) params))

     ;;DOT CASE11 [label="send_receive"];
     ;;DOT MUTEXLOCK -> CASE11 [label="else"]; {rank=same "case 11" CASE11};
     ;;DOT CASE11 -> "rmt:send-receive" [label="call failed"];
     ;;DOT CASE11 -> "RESULT" [label="call succeeded"];
     ;; not on homehost, do server query
     (else (extras-case-11 *default-log-port* runremote cmd params attemptnum rid)))))
    ;;DOT }

;; bunch of small functions factored out of send-receive to make debug easier
;;

(define (extras-case-11 *default-log-port* runremote cmd params attemptnum rid)
  ;; (mutex-unlock! *rmt-mutex*)
  (debug:print-info 12 *default-log-port* "rmt:send-receive, case  9")







>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>


>
>
|

|




<
<
<
<
<
<

|






|
<
<
|

<
<
<
<
<
<
<
<
<
<
<
<







<
<


|


<
<
<



|
|
|


<
<
<

|





<
<
<






|


|
|

<
<
<





|
|
|

<
<
<







|
|
|

<
<
<





|
|
|

<
<
<





|
|



|
|

|
|
|

<
<
<




|
|
|
|
|
|

<
<
<



|
|
|

<
<
<
<

|
<







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
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
196
197
198
199
200



201
202
203
204
205
206
207




208
209

210
211
212
213
214
215
216
	      cinfo
	      (if (server:check-if-running areapath)
		  (client:setup areapath)
		  #f))))

(define *send-receive-mutex* (make-mutex)) ;; should have separate mutex per run-id



;; this entry point can decide based on cmd whether to dispatch to old api calls via remote or via ulex
;;
(define (rmt:send-receive cmd rid params #!key (attemptnum 1)(area-dat #f))
  (let* ((areapath      *toppath*);; TODO - resolve from dbstruct to be compatible with multiple areas
	 (runremote     (or area-dat
			    *runremote*)))
    ;; ensure we have a record for our connection for given area
    (if (not runremote)                   ;; can remove this one. should never get here.         
	(begin
	  (set! *runremote* (make-remote))
	  (set! runremote   *runremote*))) ;; new runremote will come from this on next iteration
    
    (if (member cmd '(blah))
      (begin
	(mutex-lock! *send-receive-mutex*)
	(if (not *runremote*)(set! *runremote* (make-remote)))
	(let ((ulex:conn (remote-ulex:conn *runremote*)))
	  (if (not ulex:conn)(remote-ulex:conn-set! *runremote* (rmtmod:setup-ulex *toppath*)))
	  (rmtmod:send-receive-ulex ulex:conn cmd rid params attemptnum area-dat)))
      (rmt:send-receive-orig *default-log-port* runremote *rmt-mutex* areapath *db-multi-sync-mutex* cmd rid params attemptnum: attemptnum area-dat: area-dat))))

;; RA => e.g. usage (rmt:send-receive 'get-var #f (list varname))
;;
;;  add multi-sync-mutex 
;;
(define (rmt:send-receive-orig log-port runremote rmt-mutex toppath multi-sync-mutex cmd rid params #!key (attemptnum 1)(area-dat #f)) ;; start attemptnum at 1 so the modulo below works as expected

  #;(common:telemetry-log (conc "rmt:"(->string cmd))
                        payload: `((rid . ,rid)
                                   (params . ,params)))
                          
  






  ;; do all the prep locked under the rmt-mutex
  (mutex-lock! rmt-mutex)
  
  ;; 1. check if server is started IFF cmd is a write OR if we are not on the homehost, store in runremote
  ;; 2. check the age of the connections. refresh the connection if it is older than timeout-20 seconds.
  ;; 3. do the query, if on homehost use local access
  ;;
  (let* ((start-time    (current-seconds)) ;; snapshot time so all use cases get same value



	 (readonly-mode (rmtmod:calc-ro-mode runremote toppath)))













    ;; ensure we have a homehost record
    (if (not (pair? (remote-hh-dat runremote)))  ;; not on homehost
	(thread-sleep! 0.1) ;; since we shouldn't get here, delay a little
	(remote-hh-dat-set! runremote (common:get-homehost)))
    
    ;;(print "BB> readonly-mode is "readonly-mode" dbfile is "dbfile)
    (cond


     ;; give up if more than 15 attempts
     ((> attemptnum 15)
      (debug:print 0 log-port "ERROR: 15 tries to start/connect to server. Giving up.")
      (exit 1))




     ;; readonly mode, read request-  handle it - case 2
     ((and readonly-mode
           (member cmd api:read-only-queries)) 
      (mutex-unlock! rmt-mutex)
      (debug:print-info 12 log-port "rmt:send-receive, case 2")
      (rmt:open-qry-close-locally log-port multi-sync-mutex cmd 0 params)
      )




     ;; readonly mode, write request.  Do nothing, return #f
     (readonly-mode (extras-readonly-mode rmt-mutex log-port cmd params))

     ;; This block was for pre-emptively resetting the connection if there had been no communication for some time.
     ;; I don't think it adds any value. If the server is not there, just fail and start a new connection.
     ;; also, the expire-time calculation might not be correct. We want, time-since-last-server-access > (server:get-timeout)
     ;;



     ;; reset the connection if it has been unused too long
     ((and runremote
           (remote-conndat runremote)
	   (> (current-seconds) ;; if it has been more than server-timeout seconds since last contact, close this connection and start a new on
	      (+ (http-transport:server-dat-get-last-access (remote-conndat runremote))
		 (remote-server-timeout runremote))))
      (debug:print-info 0 log-port "Connection to " (remote-server-url runremote) " expired due to no accesses, forcing new connection.")
      (http-transport:close-connections area-dat: runremote)
      (remote-conndat-set! runremote #f) ;; invalidate the connection, thus forcing a new connection.
      (mutex-unlock! rmt-mutex)
      (rmt:send-receive-orig log-port runremote rmt-mutex toppath multi-sync-mutex cmd rid params attemptnum: attemptnum))
     




     ;; on homehost and this is a read
     ((and (not (remote-force-server runremote)) ;; honor forced use of server, i.e. server NOT required
	   (cdr (remote-hh-dat runremote))       ;; on homehost
           (member cmd api:read-only-queries))   ;; this is a read
      (mutex-unlock! rmt-mutex)
      (debug:print-info 12 log-port "rmt:send-receive, case  5")
      (rmt:open-qry-close-locally log-port multi-sync-mutex cmd 0 params))




     ;; on homehost and this is a write, we already have a server, but server has died
     ((and (cdr (remote-hh-dat runremote))           ;; on homehost
           (not (member cmd api:read-only-queries))  ;; this is a write
           (remote-server-url runremote)             ;; have a server
           (not (server:ping (remote-server-url runremote))))  ;; server has died. NOTE: this is not a cheap call! Need better approach.
      (set! *runremote* (make-remote))
      (remote-force-server-set! runremote (common:force-server?))
      (mutex-unlock! rmt-mutex)
      (debug:print-info 12 log-port "rmt:send-receive, case  6")
      (rmt:send-receive-orig log-port runremote rmt-mutex toppath  multi-sync-mutex cmd rid params attemptnum: attemptnum))




     ;; on homehost and this is a write, we already have a server
     ((and (not (remote-force-server runremote))     ;; honor forced use of server, i.e. server NOT required
	   (cdr (remote-hh-dat runremote))           ;; on homehost
           (not (member cmd api:read-only-queries))  ;; this is a write
           (remote-server-url runremote))            ;; have a server
      (mutex-unlock! rmt-mutex)
      (debug:print-info 12 log-port "rmt:send-receive, case  4.1")
      (rmt:open-qry-close-locally  log-port multi-sync-mutex cmd 0 params))




     ;;  on homehost, no server contact made and this is a write, passively start a server 
     ((and (not (remote-force-server runremote))     ;; honor forced use of server, i.e. server NOT required
	   (cdr (remote-hh-dat runremote))           ;; have homehost
           (not (remote-server-url runremote))       ;; no connection yet
	   (not (member cmd api:read-only-queries))) ;; not a read-only query
      (debug:print-info 12 log-port "rmt:send-receive, case  8")
      (let ((server-url  (server:check-if-running toppath))) ;; (server:read-dotserver->url toppath))) ;; (server:check-if-running toppath))) ;; Do NOT want to run server:check-if-running - very expensive to do for every write call
	(if server-url
	    (remote-server-url-set! runremote server-url) ;; the string can be consumed by the client setup if needed
	    (if (common:force-server?)
		(server:start-and-wait toppath)
		(server:kind-run toppath))))
      (remote-force-server-set! runremote (common:force-server?))
      (mutex-unlock! rmt-mutex)
      (debug:print-info 12 log-port "rmt:send-receive, case  8.1")
      (rmt:open-qry-close-locally  log-port multi-sync-mutex cmd 0 params))




     ((or (and (remote-force-server runremote)              ;; we are forcing a server and don't yet have a connection to one
	       (not (remote-conndat runremote)))
	  (and (not (cdr (remote-hh-dat runremote)))        ;; not on a homehost 
	       (not (remote-conndat runremote))))           ;; and no connection
      (debug:print-info 12 log-port "rmt:send-receive, case 9, hh-dat: " (remote-hh-dat runremote) " conndat: " (remote-conndat runremote))
      (mutex-unlock! rmt-mutex)
      (if (not (server:check-if-running toppath)) ;; who knows, maybe one has started up?
	  (server:start-and-wait toppath))
      (remote-conndat-set! runremote (rmt:get-connection-info toppath)) ;; calls client:setup which calls client:setup-http
      (rmt:send-receive-orig log-port runremote rmt-mutex toppath multi-sync-mutex cmd rid params attemptnum: attemptnum)) ;; TODO: add back-off timeout as




     ;; all set up if get this far, dispatch the query
     ((and (not (remote-force-server runremote))
	   (cdr (remote-hh-dat runremote))) ;; we are on homehost
      (mutex-unlock! rmt-mutex)
      (debug:print-info 12 log-port "rmt:send-receive, case 10")
      (rmt:open-qry-close-locally  log-port multi-sync-mutex cmd (if rid rid 0) params))





     ;; not on homehost, do server query
     (else (extras-case-11 log-port runremote cmd params attemptnum rid)))))


;; bunch of small functions factored out of send-receive to make debug easier
;;

(define (extras-case-11 *default-log-port* runremote cmd params attemptnum rid)
  ;; (mutex-unlock! *rmt-mutex*)
  (debug:print-info 12 *default-log-port* "rmt:send-receive, case  9")
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
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
			     (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 #!key (remretries 5))
  (let* ((qry-is-write   (not (member cmd api:read-only-queries)))
	 (db-file-path   (db:dbfile-path)) ;;  0))
	 (dbstruct-local (db:setup #t))  ;; make-dbr:dbstruct path:  dbdir local: #t)))
	 (read-only      (not (file-write-access? db-file-path)))
	 (start          (current-milliseconds))
	 (resdat         (if (not (and read-only qry-is-write))
			     (let ((v (api:execute-requests dbstruct-local (vector (symbol->string cmd) params))))
			       (handle-exceptions ;; there has been a long history of receiving strange errors from values returned by the client when things go wrong..
				exn               ;;  This is an attempt to detect that situation and recover gracefully
				(begin
				  (debug:print0 *default-log-port* "ERROR: bad data from server " v " message: "  ((condition-property-accessor 'exn 'message) exn))
				  (vector #t '())) ;; should always get a vector but if something goes wrong return a dummy
				(if (and (vector? v)
					 (> (vector-length v) 1))
				    (let ((newvec (vector (vector-ref v 0)(vector-ref v 1))))
				      newvec)           ;; by copying the vector while inside the error handler we should force the detection of a corrupted record
				    (vector #t '()))))  ;; we could also check that the returned types are valid
			     (vector #t '())))
	 (success        (vector-ref resdat 0))
	 (res            (vector-ref resdat 1))
	 (duration       (- (current-milliseconds) start)))
    (if (and read-only qry-is-write)
        (debug:print 0 *default-log-port* "ERROR: attempt to write to read-only database ignored. cmd=" cmd))
    (if (not success)
	(if (> remretries 0)
	    (begin
	      (debug:print-error 0 *default-log-port* "local query failed. Trying again.")
	      (thread-sleep! (/ (random 5000) 1000)) ;; some random delay 
	      (rmt:open-qry-close-locally cmd run-id params remretries: (- remretries 1)))
	    (begin
	      (debug:print-error 0 *default-log-port* "too many retries in rmt:open-qry-close-locally, giving up")
	      #f))
	(begin
	  ;; (rmt:update-db-stats run-id cmd params duration)
	  ;; mark this run as dirty if this was a write, the watchdog is responsible for syncing it
	  (if qry-is-write
	      (let ((start-time (current-seconds)))
		(mutex-lock! *db-multi-sync-mutex*)
/		(set! *db-last-access* start-time)  ;; THIS IS PROBABLY USELESS? (we are on a client)
                (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))
	 (res  	   (handle-exceptions
		    exn
		    #f
		    (http-transport:client-api-send-receive run-id connection-info cmd params))))
    (if (and res (vector-ref res 0))







<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<
<







301
302
303
304
305
306
307











































308
309
310
311
312
313
314
			     (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:send-receive-no-auto-client-setup connection-info cmd run-id params)
  (let* ((run-id   (if run-id run-id 0))
	 (res  	   (handle-exceptions
		    exn
		    #f
		    (http-transport:client-api-send-receive run-id connection-info cmd params))))
    (if (and res (vector-ref res 0))
932
933
934
935
936
937
938

939
940



(define (rmt:test-get-archive-block-info archive-block-id)
  (rmt:send-receive 'test-get-archive-block-info #f (list archive-block-id)))

(set-functions rmt:send-receive                       remote-server-url-set!
	       http-transport:close-connections	      remote-conndat-set!
	       debug:print                            debug:print-info

	       remote-ro-mode                         remote-ro-mode-set!
	       remote-ro-mode-checked-set!            remote-ro-mode-checked)









>

|
>
>
861
862
863
864
865
866
867
868
869
870
871
872

(define (rmt:test-get-archive-block-info archive-block-id)
  (rmt:send-receive 'test-get-archive-block-info #f (list archive-block-id)))

(set-functions rmt:send-receive                       remote-server-url-set!
	       http-transport:close-connections	      remote-conndat-set!
	       debug:print                            debug:print-info
	       debug:print-error 
	       remote-ro-mode                         remote-ro-mode-set!
	       remote-ro-mode-checked-set!            remote-ro-mode-checked
	       db:dbfile-path                           db:setup
	       api:execute-requests                   api:read-only-queries)

Modified rmtmod.scm from [8f5d01960e] to [dce20d5361].

1
2
3
4
5
6
7
8
9
;;======================================================================
;; Copyright 2017, Matthew Welland.
;; 
;; This file is part of Megatest.
;; 
;;     Megatest is free software: you can redistribute it and/or modify
;;     it under the terms of the GNU General Public License as published by
;;     the Free Software Foundation, either version 3 of the License, or
;;     (at your option) any later version.

|







1
2
3
4
5
6
7
8
9
;;======================================================================
;; Copyright 2019, Matthew Welland.
;; 
;; This file is part of Megatest.
;; 
;;     Megatest is free software: you can redistribute it and/or modify
;;     it under the terms of the GNU General Public License as published by
;;     the Free Software Foundation, either version 3 of the License, or
;;     (at your option) any later version.
23
24
25
26
27
28
29

30
31
32
33
34
35
36
37
38
39
40
41
42





43
44
45
46

47
48
49



50
51
52
53

54
55


56
57
58
59



















































60
61
62
63
64
65
66

(module rmtmod
	*
	
(import scheme chicken data-structures extras)
(import (prefix sqlite3 sqlite3:) posix typed-records srfi-18)
(import commonmod)


;; Hack to make these functions visible to the refactored code, goal is to eliminate these over time.
(define (rmt:send-receive . params) #f)
(define (http-transport:close-connections . params) #f)
;; from remote defstruct in common.scm
(define (remote-conndat-set! . params) #f)
(define (remote-server-url-set! . params) #f)
(define (remote-ro-mode . params) #f)
(define (remote-ro-mode-set! . params) #f)
(define (remote-ro-mode-checked-set! . params) #f)
(define (remote-ro-mode-checked . params) #f)
(define (debug:print . params) #f)
(define (debug:print-info . params) #f)






(define (set-functions send-receive        rsus
		       close-connections   rcs
		       dbgp                dbgpinfo

		       ro-mode             ro-mode-set
		       ro-mode-checked-set ro-mode-checked
		       ) 



  (set! rmt:send-receive                 send-receive)
  (set! remote-server-url-set!           rsus)
  (set! http-transport:close-connections close-connections)
  (set! remote-conndat-set!              rcs)

  (set! debug:print                      dbgp)
  (set! debug:print-info                 dbgpinfo)


  (set! remote-ro-mode                   ro-mode)
  (set! remote-ro-mode-set!              ro-mode-set)
  (set! remote-ro-mode-checked-set!      ro-mode-checked-set)
  (set! remote-ro-mode-checked           ro-mode-checked))




















































(define (rmtmod:calc-ro-mode runremote *toppath*)
  (if (and runremote
	   (remote-ro-mode-checked runremote))
      (remote-ro-mode runremote)
      (let* ((dbfile  (conc *toppath* "/megatest.db"))
	     (ro-mode (not (file-write-access? dbfile)))) ;; TODO: use dbstruct or runremote to figure this out in future







>













>
>
>
>
>




>


|
>
>
>
|



>


>
>



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







23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
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

(module rmtmod
	*
	
(import scheme chicken data-structures extras)
(import (prefix sqlite3 sqlite3:) posix typed-records srfi-18)
(import commonmod)
(use (prefix ulex ulex:))

;; Hack to make these functions visible to the refactored code, goal is to eliminate these over time.
(define (rmt:send-receive . params) #f)
(define (http-transport:close-connections . params) #f)
;; from remote defstruct in common.scm
(define (remote-conndat-set! . params) #f)
(define (remote-server-url-set! . params) #f)
(define (remote-ro-mode . params) #f)
(define (remote-ro-mode-set! . params) #f)
(define (remote-ro-mode-checked-set! . params) #f)
(define (remote-ro-mode-checked . params) #f)
(define (debug:print . params) #f)
(define (debug:print-info . params) #f)
(define (debug:print-error . params) #f)
(define (db:dbfile-path . params) #f)
(define (db:setup . params) #f)
(define (api:execute-requests . params) #f)
(define (api:read-only-queries . params) #f)

(define (set-functions send-receive        rsus
		       close-connections   rcs
		       dbgp                dbgpinfo
		       dbgperr
		       ro-mode             ro-mode-set
		       ro-mode-checked-set ro-mode-checked

		       dbfile-path         dbsetup
		       exec-req            read-only-queries
		       ) 
  (set! rmt:send-receive            send-receive)
  (set! remote-server-url-set!           rsus)
  (set! http-transport:close-connections close-connections)
  (set! remote-conndat-set!              rcs)
  ;; print stuff
  (set! debug:print                      dbgp)
  (set! debug:print-info                 dbgpinfo)
  (set! debug:print-error                dbgperr)
  ;;
  (set! remote-ro-mode                   ro-mode)
  (set! remote-ro-mode-set!              ro-mode-set)
  (set! remote-ro-mode-checked-set!      ro-mode-checked-set)
  (set! remote-ro-mode-checked           ro-mode-checked)
  ;; db stuff for local db access
  (set! db:dbfile-path                   dbfile-path)
  (set! db:setup                         dbsetup)
  (set! apt:execute-requests             exec-req)
  (set! api:read-only-queries	         read-only-queries)
  )

(define (rmt:open-qry-close-locally log-port multi-sync-mutex cmd run-id params #!key (remretries 5))
  (let* ((qry-is-write   (not (member cmd api:read-only-queries)))
	 (db-file-path   (db:dbfile-path)) ;;  0))
	 (dbstruct-local (db:setup #t))  ;; make-dbr:dbstruct path:  dbdir local: #t)))
	 (read-only      (not (file-write-access? db-file-path)))
	 (start          (current-milliseconds))
	 (resdat         (if (not (and read-only qry-is-write))
			     (let ((v (api:execute-requests dbstruct-local (vector (symbol->string cmd) params))))
			       (handle-exceptions ;; there has been a long history of receiving strange errors from values returned by the client when things go wrong..
				exn               ;;  This is an attempt to detect that situation and recover gracefully
				(begin
				  (debug:print 0 log-port "ERROR: bad data from server " v " message: "  ((condition-property-accessor 'exn 'message) exn))
				  (vector #t '())) ;; should always get a vector but if something goes wrong return a dummy
				(if (and (vector? v)
					 (> (vector-length v) 1))
				    (let ((newvec (vector (vector-ref v 0)(vector-ref v 1))))
				      newvec)           ;; by copying the vector while inside the error handler we should force the detection of a corrupted record
				    (vector #t '()))))  ;; we could also check that the returned types are valid
			     (vector #t '())))
	 (success        (vector-ref resdat 0))
	 (res            (vector-ref resdat 1))
	 (duration       (- (current-milliseconds) start)))
    (if (and read-only qry-is-write)
        (debug:print 0 log-port "ERROR: attempt to write to read-only database ignored. cmd=" cmd))
    (if (not success)
	(if (> remretries 0)
	    (begin
	      (debug:print-error 0 log-port "local query failed. Trying again.")
	      (thread-sleep! (/ (random 5000) 1000)) ;; some random delay 
	      (rmt:open-qry-close-locally log-port multi-sync-mutex cmd run-id params remretries: (- remretries 1)))
	    (begin
	      (debug:print-error 0 log-port "too many retries in rmt:open-qry-close-locally, giving up")
	      #f))
	(begin
	  ;; (rmt:update-db-stats run-id cmd params duration)
	  ;; mark this run as dirty if this was a write, the watchdog is responsible for syncing it
	  #;(if qry-is-write
	      (let ((start-time (current-seconds)))
		(mutex-lock! multi-sync-mutex)
		(set! *db-last-access* start-time)  ;; THIS IS PROBABLY USELESS? (we are on a client)
                (mutex-unlock! multi-sync-mutex)))))
    res))



(define (rmtmod:calc-ro-mode runremote *toppath*)
  (if (and runremote
	   (remote-ro-mode-checked runremote))
      (remote-ro-mode runremote)
      (let* ((dbfile  (conc *toppath* "/megatest.db"))
	     (ro-mode (not (file-write-access? dbfile)))) ;; TODO: use dbstruct or runremote to figure this out in future
111
112
113
114
115
116
117
118



119












	(mutex-lock! *rmt-mutex*)
	(http-transport:close-connections area-dat: runremote)
	(set! *runremote* #f) ;; force starting over
	(mutex-unlock! *rmt-mutex*)
	(thread-sleep! wait-delay)
	(rmt:send-receive cmd rid params attemptnum: (+ attemptnum 1)))
      res)) ;; All good, return res




)




















>
>
>
|
>
>
>
>
>
>
>
>
>
>
>
>
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
	(mutex-lock! *rmt-mutex*)
	(http-transport:close-connections area-dat: runremote)
	(set! *runremote* #f) ;; force starting over
	(mutex-unlock! *rmt-mutex*)
	(thread-sleep! wait-delay)
	(rmt:send-receive cmd rid params attemptnum: (+ attemptnum 1)))
      res)) ;; All good, return res

;;======================================================================
;; ulex and steps stuff
;;======================================================================

(define (rmtmod:setup-ulex toppath)
  (ulex:make-area
   dbdir:   (conc toppath "/ulexdb")
   pktsdir: (conc toppath "/pkts") 
   ))



(define (rmtmod:send-receive-ulex ulex:conn cmd rid params attemptnum area-dat)
  #f)

)