Megatest

Changes On Branch 0e1320be3508a1ed
Login

Changes In Branch v1.65-side2 Through [0e1320be35] Excluding Merge-Ins

This is equivalent to a diff from 2769e4b7c9 to 0e1320be35

2021-03-01
17:42
Manually patched in the new view check-in: f5206150ee user: mrwellan tags: v1.6569-new-view
2021-01-26
14:00
Fix for the > crash. Maybe... Leaf check-in: 5a05fc04ff user: matt tags: v1.6569-gt-crash-fix
2021-01-25
12:03
rebased lazy-queue rollup check-in: 07ab120544 user: matt tags: v1.65-lazyqueue-items-rollup
2021-01-15
22:46
begin diet check-in: badd71f3b3 user: matt tags: v1.6569-diet
21:34
eval-string-in-environment if was disabled, re-enabled check-in: 9564772564 user: matt tags: v1.6569-reenable-eval-if
2021-01-08
11:42
enable custom value for max delay between archive time and test last update time Leaf check-in: 86a3d1148e user: pjhatwal tags: v1.6569-refactor
2020-11-25
12:00
Fixed issues in server gating code Leaf check-in: 063273e8cb user: mrwellan tags: v1.6569-server-gate-fix
2020-11-24
22:27
Added support for resetting run - allows to reload tests-paths to add tests to a run part way though. Just run megatest -clean-cache -runname $MT_RUNNAME Leaf check-in: 213021e02d user: mrwellan tags: v1.6596-reload-tests-paths
2020-10-13
16:46
Changed version from 69 to 76. No other changes. Will compile with chicken 13 check-in: 87ca35010f user: mmgraham tags: v1.65, v1.6576
2020-10-12
16:58
Merged minor change to v1.65 check-in: 60a665385a user: mrwellan tags: v1.65-side2
16:49
Reduced message from failed to info. Reverted a delay which seems to help pass full stack ext-tests. Leaf check-in: 9e35b1252c user: mrwellan tags: v1.65-minor-patch
13:06
Added in the query-rest stuff. check-in: 0e1320be35 user: mrwellan tags: v1.65-side2
10:19
For running local do not use ssh to run pstree check-in: 55f3c8af7f user: mrwellan tags: v1.65-side2
10:18
Safe vector access in rmt. check-in: 58bb6d997a user: mrwellan tags: v1.65-side2
2020-10-11
22:46
Patched forward adjutant code. check-in: f936717bfa user: matt tags: v1.65-adjutant-again
2020-10-05
22:49
Do not exit on failure to create directory - race conditons on NFS cause false fail scenarios - just keep going and cross your fingers... (cherrypicked from v1.6572) check-in: 05b253a452 user: matt tags: v1.65-sidework
22:46
run duration testdat check-in: 4a0b43f3c6 user: matt tags: v1.65-test-rundat2
2020-09-28
23:41
Moving call throttle attempt to v1.6564 ==5.71/0.8/WARN/1201== Leaf check-in: 919bc699d8 user: mrwellan tags: v1.65-rmt-call-throttle
2020-09-21
15:36
merged in 1.65-test-rundat branch ==/FAIL/orion,mars/== check-in: cfd25d66e9 user: mmgraham tags: v1.6571, v1.65-failed-testdat
07:00
Added get-testsuite-name all over launch:setup and still not set when needed! This did NOT work. Closed-Leaf check-in: 2efe8ad422 user: mrwellan tags: v1.65-get-testsuitename
2020-09-19
04:21
Start moving test_rundat to no-sync db. ==/20/2/WARN/1203/mars/== check-in: abfabdb839 user: matt tags: v1.65-test-rundat
2020-09-18
17:30
added check for file existence before file delete ==/14/1.9/WARN/orion,mars/== NOTE: This is the last v1.65 before the split off. I.e code from before this point IS in the far future v1.65 branch. Code from this point to that branch might NOT be in the branch. check-in: 2769e4b7c9 user: mmgraham tags: v1.65, v1.6569
12:27
cherry picked 2 fixes, changed version to 1.6569 ==/7.2/2.0/PASS/1201/mars/== check-in: d145d0eb02 user: mmgraham tags: v1.65

Modified api.scm from [4fa67bb6bd] to [cc4c2bfc8f].

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
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







-
+



-
+





-
+










+
+







    ((not (vector? dat))                    ;; it is an error to not receive a vector
     (vector #f (vector #f "remote must be called with a vector")))
    ((> *api-process-request-count* 20) ;; 20)
     (debug:print 0 *default-log-port* "WARNING: api:execute-requests received an overloaded message.")
     (set! *server-overloaded* #t)
     (vector #f (vector #f 'overloaded))) ;; the inner vector is what gets returned. nope, don't know why. please refactor!
    (else  
     (let* ((cmd-in            (vector-ref dat 0))
     (let* ((cmd-in            (common:safe-vector-ref dat 0 'nocmd))
            (cmd               (if (symbol? cmd-in)
				   cmd-in
				   (string->symbol cmd-in)))
            (params            (vector-ref dat 1))
            (params            (common:safe-vector-ref dat 1 '()))
            (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)))
            (foo               (begin
                                 (common:telemetry-log (conc "api-in:"(->string cmd))
                                 #;(common:telemetry-log (conc "api-in:"(->string cmd))
                                                       payload: `((params . ,params)))
                                 
                                 #t))
            (res    
             (if writecmd-in-readonly-mode
                 (conc "attempt to run write command "cmd" on a read-only database")
                 (case cmd
                   ;;===============================================
                   ;; READ/WRITE QUERIES
                   ;;===============================================

		   ((nocmd)                         '(#f "All broken!"))

                   ((get-keys-write)                        (db:get-keys dbstruct)) ;; force a dummy "write" query to force server; for debug in -repl
                   
                   ;; SERVERS
                   ((start-server)                    (apply server:kind-run params))
                   ((kill-server)                     (set! *server-run* #f))

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
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







-
+




-
+
















-
-
+
+







       ;; save all stats
       (let ((delta-t (- (current-milliseconds)
			 start-t)))
	 (hash-table-set! *db-api-call-time* cmd
			  (cons delta-t (hash-table-ref/default *db-api-call-time* cmd '()))))
       (if writecmd-in-readonly-mode
           (begin
             (common:telemetry-log (conc "api-out:"(->string cmd))
             #;(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))
             #;(common:telemetry-log (conc "api-out:"(->string cmd))
                                   payload: `((params . ,params)
                                              (ok-res . #f)))
             (vector #t res))))))))

;; 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
  (set! *api-process-request-count* (+ *api-process-request-count* 1))
  (let* ((cmd     ($ 'cmd))
	 (paramsj ($ 'params))
	 (params  (db:string->obj paramsj transport: 'http)) ;; incoming data from the POST (or is it a GET?)
	 (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?)
	 (success (common:safe-vector-ref resdat 0 #f))
	 (res     (common:safe-vector-ref resdat 1 #f))) ;; (vector flag payload), get the payload, ignore the flag (why?)
    (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

Modified common.scm from [33c7316880] to [2732dee33e].

486
487
488
489
490
491
492









493

494
495
496
497
498
499
500
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501

502
503
504
505
506
507
508
509







+
+
+
+
+
+
+
+
+
-
+







        ;;     copy <file>.hrs.gz <file>.days.gz
        (when (>= (age-wks daysfile) 1)
          (copy daysfile wksfile)
          (copy hrsfile daysfile))
        #t)
      #f))
  
(define (common:safe-vector-ref vec indx default)
  (if (vector? vec)
      (handle-exceptions
	  exn
	(begin
	  (debug:print-info 0 *default-log-port* "remote data issue: exn=" exn)
	  default)
	(vector-ref vec indx))
      default))
        

        
;; Rotate logs, logic: 
;;                 if > 500k and older than 1 week:
;;                     remove previous compressed log and compress this log
;; WARNING: This proc operates assuming that it is in the directory above the
;;          logs directory you wish to log-rotate.
;;

Modified launch.scm from [d0067277fa] to [d79c56de3e].

769
770
771
772
773
774
775


776

777
778
779
780
781
782
783
769
770
771
772
773
774
775
776
777

778
779
780
781
782
783
784
785







+
+
-
+







                 (item-path (vector-ref running-test 11)))
			       	(debug:print 0 *default-log-port* "test " test-name "/" item-path " not completed")
              (if (not (null? tal))
				  (loop (car tal) (cdr tal)))))))))))        
 
(define (launch:is-test-alive host pid)
  (if (and host pid (not (equal? host "n/a")))
      (let* ((is-local (equal? host (get-host-name)))
	     (ssh-cmd   (if is-local " " (conc "ssh " host " ")))
      (let* ((cmd (conc "ssh " host " pstree -A " pid))
	     (cmd (conc ssh-cmd "pstree -A " pid))
	     (output (with-input-from-pipe cmd read-lines)))
	(debug:print 2 *default-log-port* "Running " cmd " received " output)
	(if (eq? (length output) 0)
	   #f
	   #t))
      #t))
 

Modified rmt.scm from [39d97c528a] to [23bba59d7b].

52
53
54
55
56
57
58



















59
60
61
62
63
64
65


66

67
68
69
70
71
72
73
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







+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+







+
+
-
+







	      cinfo
	      (if (server:check-if-running areapath)
		  (client:setup areapath)
		  #f))))

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

(define *rmt-query-last-call-time* 0)
(define *rmt-query-last-rest-time* 0) ;; last time there was at least a 1/2 second rest - giving other processes access to the db

;; NOTE: This query rest algorythm will not adapt to long query times. REDESIGN NEEDED. TODO. FIXME.
;;
(define (rmt:query-rest)
  (let* ((now (current-milliseconds)))
    (cond
     ((> (- now *rmt-query-last-call-time*) 500)  ;; it's been a while since last query - no need to rest
      (set! *rmt-query-last-rest-time*  now)
      (set! *rmt-query-last-call-time*  now))
     ((> (- now *rmt-query-last-rest-time*) 5000) ;; no natural rests have happened
      (debug:print 0 *default-log-port* "query rest needed. blocking for 1/2 second.")
      (thread-sleep! 0.5) ;; force a rest of a half second
      (set! *rmt-query-last-rest-time* now)
      (set! *rmt-query-last-call-time* now))
     (else ;; sufficient rests have occurred, just record the last query time
      (set! *rmt-query-last-call-time* now)))))

;; 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)))
  (if (not (equal? (configf:lookup *configdat* "setup" "query-rest") "no"))
      (rmt:query-rest))
                          
  
  (if (> attemptnum 2)
      (debug:print 0 *default-log-port* "INFO: attemptnum in rmt:send-receive is " attemptnum))
    
  (cond
   ((> attemptnum 2) (thread-sleep! 0.05))
   ((> attemptnum 10) (thread-sleep! 0.5))
   ((> attemptnum 20) (thread-sleep! 1)))
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
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







-
-
+
+


















-
+







				  (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))
	 (success        (common:safe-vector-ref resdat 0 #f)) ;; (vector-ref resdat 0))
	 (res            (common:safe-vector-ref resdat 1 #f)) ;; (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)
		(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