Megatest

Diff
Login

Differences From Artifact [13e6c119c2]:

To Artifact [1ca55ce1cc]:


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

;;                       0           1              2              3
(defstruct launch:einf (pid #t)(exit-status #t)(exit-code #t)(rollup-status 0))

;; return (conc status ": " comment) from the final section so that
;;   the comment can be set in the step record in launch.scm
;;
(define (launch:load-logpro-dat run-id test-id stepname)
  (let ((cname (conc stepname ".dat")))
    (if (file-exists? cname)
	(let* ((dat  (read-config cname #f #f))
	       (csvr (db:logpro-dat->csv dat stepname))
	       (csvt (let-values (((fmt-cell fmt-record fmt-csv) (make-format ",")))
		       (fmt-csv (map list->csv-record csvr))))
	       (status (configf:lookup dat "final" "exit-status"))
	       (msg     (configf:lookup dat "final" "message")))
          (if csvt  ;; this if blocked stack dump caused by .dat file from logpro being 0-byte.  fixed by upgrading logpro
              (rmt:csv->test-data run-id test-id csvt)
	      (debug:print 0 *default-log-port* "ERROR: no csvdat exists for run-id: " run-id " test-id: " test-id " stepname: " stepname ", check that logpro version is 1.15 or newer"))
	  ;;  (BB> "Error: run-id/test-id/stepname="run-id"/"test-id"/"stepname" => bad csvr="csvr)
	  ;;  )
	  (cond
	   ((equal? status "PASS") "PASS") ;; skip the message part if status is pass
	   (status (conc (configf:lookup dat "final" "exit-status") ": " (if msg msg "no message")))
	   (else #f)))
	#f)))

(define (launch:runstep ezstep run-id test-id exit-info m tal testconfig)
  (let* ((stepname       (car ezstep))  ;; do stuff to run the step
	 (stepinfo       (cadr ezstep))
	 (stepparts      (string-match (regexp "^(\\{([^\\}]*)\\}\\s*|)(.*)$") stepinfo))
	 (stepparms      (list-ref stepparts 2)) ;; for future use, {VAR=1,2,3}, run step for each 
	 (stepcmd        (list-ref stepparts 3))
	 (script         "") ; "#!/bin/bash\n") ;; yep, we depend on bin/bash FIXME!!!\
	 (logpro-file    (conc stepname ".logpro"))







|









|









|







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

;;                       0           1              2              3
(defstruct launch:einf (pid #t)(exit-status #t)(exit-code #t)(rollup-status 0))

;; return (conc status ": " comment) from the final section so that
;;   the comment can be set in the step record in launch.scm
;;
(define (launch:load-logpro-dat area-dat run-id test-id stepname)
  (let ((cname (conc stepname ".dat")))
    (if (file-exists? cname)
	(let* ((dat  (read-config cname #f #f))
	       (csvr (db:logpro-dat->csv dat stepname))
	       (csvt (let-values (((fmt-cell fmt-record fmt-csv) (make-format ",")))
		       (fmt-csv (map list->csv-record csvr))))
	       (status (configf:lookup dat "final" "exit-status"))
	       (msg     (configf:lookup dat "final" "message")))
          (if csvt  ;; this if blocked stack dump caused by .dat file from logpro being 0-byte.  fixed by upgrading logpro
              (rmt:csv->test-data area-dat run-id test-id csvt)
	      (debug:print 0 *default-log-port* "ERROR: no csvdat exists for run-id: " run-id " test-id: " test-id " stepname: " stepname ", check that logpro version is 1.15 or newer"))
	  ;;  (BB> "Error: run-id/test-id/stepname="run-id"/"test-id"/"stepname" => bad csvr="csvr)
	  ;;  )
	  (cond
	   ((equal? status "PASS") "PASS") ;; skip the message part if status is pass
	   (status (conc (configf:lookup dat "final" "exit-status") ": " (if msg msg "no message")))
	   (else #f)))
	#f)))

(define (launch:runstep area-dat ezstep run-id test-id exit-info m tal testconfig)
  (let* ((stepname       (car ezstep))  ;; do stuff to run the step
	 (stepinfo       (cadr ezstep))
	 (stepparts      (string-match (regexp "^(\\{([^\\}]*)\\}\\s*|)(.*)$") stepinfo))
	 (stepparms      (list-ref stepparts 2)) ;; for future use, {VAR=1,2,3}, run step for each 
	 (stepcmd        (list-ref stepparts 3))
	 (script         "") ; "#!/bin/bash\n") ;; yep, we depend on bin/bash FIXME!!!\
	 (logpro-file    (conc stepname ".logpro"))
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
    ;;   (if (and prevstep (file-exists? prev-env))
    ;;       (set! script (conc script "source " prev-env))))
    
    ;; call the command using mt_ezstep
    ;; (set! script (conc "mt_ezstep " stepname " " (if prevstep prevstep "x") " " stepcmd))
    
    (debug:print 4 *default-log-port* "script: " script)
    (rmt:teststep-set-status! run-id test-id stepname "start" "-" #f #f)
    ;; now launch the actual process
    (call-with-environment-variables 
     (list (cons "PATH" (conc (get-environment-variable "PATH") ":.")))
     (lambda () ;; (process-run "/bin/bash" "-c" "exec ls -l /tmp/foobar > /tmp/delme-more.log 2>&1")
       (let* ((cmd (conc stepcmd " > " stepname ".log 2>&1")) ;; >outfile 2>&1 
	      (pid (process-run "/bin/bash" (list "-c" cmd))))

         (with-output-to-file "Makefile.ezsteps"
           (lambda ()
             (print stepname ".log :")
             (print "\t" cmd)
             (if (file-exists? (conc stepname ".logpro"))
                 (print "\tlogpro " stepname ".logpro " stepname ".html < " stepname ".log"))
             (print)
             (print stepname " : " stepname ".log")
             (print))
           #:append)

	 (rmt:test-set-top-process-pid run-id test-id pid)
	 (let processloop ((i 0))
	   (let-values (((pid-val exit-status exit-code)(process-wait pid #t)))
		       (mutex-lock! m)
		       (launch:einf-pid-set!         exit-info pid)         ;; (vector-set! exit-info 0 pid)
		       (launch:einf-exit-status-set! exit-info exit-status) ;; (vector-set! exit-info 1 exit-status)
		       (launch:einf-exit-code-set!   exit-info exit-code)   ;; (vector-set! exit-info 2 exit-code)
		       (mutex-unlock! m)







|


















|







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
    ;;   (if (and prevstep (file-exists? prev-env))
    ;;       (set! script (conc script "source " prev-env))))
    
    ;; call the command using mt_ezstep
    ;; (set! script (conc "mt_ezstep " stepname " " (if prevstep prevstep "x") " " stepcmd))
    
    (debug:print 4 *default-log-port* "script: " script)
    (rmt:teststep-set-status! area-dat run-id test-id stepname "start" "-" #f #f)
    ;; now launch the actual process
    (call-with-environment-variables 
     (list (cons "PATH" (conc (get-environment-variable "PATH") ":.")))
     (lambda () ;; (process-run "/bin/bash" "-c" "exec ls -l /tmp/foobar > /tmp/delme-more.log 2>&1")
       (let* ((cmd (conc stepcmd " > " stepname ".log 2>&1")) ;; >outfile 2>&1 
	      (pid (process-run "/bin/bash" (list "-c" cmd))))

         (with-output-to-file "Makefile.ezsteps"
           (lambda ()
             (print stepname ".log :")
             (print "\t" cmd)
             (if (file-exists? (conc stepname ".logpro"))
                 (print "\tlogpro " stepname ".logpro " stepname ".html < " stepname ".log"))
             (print)
             (print stepname " : " stepname ".log")
             (print))
           #:append)

	 (rmt:test-set-top-process-pid area-dat run-id test-id pid)
	 (let processloop ((i 0))
	   (let-values (((pid-val exit-status exit-code)(process-wait pid #t)))
		       (mutex-lock! m)
		       (launch:einf-pid-set!         exit-info pid)         ;; (vector-set! exit-info 0 pid)
		       (launch:einf-exit-status-set! exit-info exit-status) ;; (vector-set! exit-info 1 exit-status)
		       (launch:einf-exit-code-set!   exit-info exit-code)   ;; (vector-set! exit-info 2 exit-code)
		       (mutex-unlock! m)
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
	  (logfna (if logpro-used (conc stepname ".html") ""))
	  (comment #f))
      (if logpro-used
	  (let ((datfile (conc stepname ".dat")))
	    ;; load the .dat file into the test_data table if it exists
	    (if (file-exists? datfile)
		(set! comment (launch:load-logpro-dat run-id test-id stepname)))
	    (rmt:test-set-log! run-id test-id (conc stepname ".html"))))
      (rmt:teststep-set-status! run-id test-id stepname "end" exinfo comment logfna))
    ;; set the test final status
    (let* ((process-exit-status (launch:einf-exit-code exit-info)) ;; (vector-ref exit-info 2))
	   (this-step-status (cond
			      ((and (eq? process-exit-status 2) logpro-used) 'warn)   ;; logpro 2 = warnings
			      ((and (eq? process-exit-status 3) logpro-used) 'check)  ;; logpro 3 = check
			      ((and (eq? process-exit-status 4) logpro-used) 'waived) ;; logpro 4 = waived
			      ((and (eq? process-exit-status 5) logpro-used) 'abort)  ;; logpro 5 = abort







|
|







172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
	  (logfna (if logpro-used (conc stepname ".html") ""))
	  (comment #f))
      (if logpro-used
	  (let ((datfile (conc stepname ".dat")))
	    ;; load the .dat file into the test_data table if it exists
	    (if (file-exists? datfile)
		(set! comment (launch:load-logpro-dat run-id test-id stepname)))
	    (rmt:test-set-log! area-dat run-id test-id (conc stepname ".html"))))
      (rmt:teststep-set-status! area-dat run-id test-id stepname "end" exinfo comment logfna))
    ;; set the test final status
    (let* ((process-exit-status (launch:einf-exit-code exit-info)) ;; (vector-ref exit-info 2))
	   (this-step-status (cond
			      ((and (eq? process-exit-status 2) logpro-used) 'warn)   ;; logpro 2 = warnings
			      ((and (eq? process-exit-status 3) logpro-used) 'check)  ;; logpro 3 = check
			      ((and (eq? process-exit-status 4) logpro-used) 'waived) ;; logpro 4 = waived
			      ((and (eq? process-exit-status 5) logpro-used) 'abort)  ;; logpro 5 = abort
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
      (debug:print 4 *default-log-port* "Exit value received: " (launch:einf-exit-code exit-info) " logpro-used: " logpro-used 
		   " this-step-status: " this-step-status " overall-status: " overall-status 
		   " next-status: " next-status " rollup-status: "  (launch:einf-rollup-status exit-info)) ;; (vector-ref exit-info 3))
      (case next-status
	((warn)
	 (launch:einf-rollup-status-set! exit-info 2) ;; (vector-set! exit-info 3 2) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! run-id test-id next-state "WARN" 
				 (if (eq? this-step-status 'warn) "Logpro warning found" #f)
				 #f))
	((check)
	 (launch:einf-rollup-status-set! exit-info 3) ;; (vector-set! exit-info 3 3) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! run-id test-id next-state "CHECK" 
				 (if (eq? this-step-status 'check) "Logpro check found" #f)
				 #f))
	((waived)
	 (launch:einf-rollup-status-set! exit-info 4) ;; (vector-set! exit-info 3 3) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! run-id test-id next-state "WAIVED" 
				 (if (eq? this-step-status 'check) "Logpro waived found" #f)
				 #f))
	((abort)
	 (launch:einf-rollup-status-set! exit-info 5) ;; (vector-set! exit-info 3 4) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! run-id test-id next-state "ABORT" 
				 (if (eq? this-step-status 'abort) "Logpro abort found" #f)
				 #f))
	((skip)
	 (launch:einf-rollup-status-set! exit-info 6) ;; (vector-set! exit-info 3 4) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! run-id test-id next-state "SKIP" 
				 (if (eq? this-step-status 'skip) "Logpro skip found" #f)
				 #f))
	((pass)
	 (tests:test-set-status! run-id test-id next-state "PASS" #f #f))
	(else ;; 'fail
	 (launch:einf-rollup-status-set! exit-info 1) ;; (vector-set! exit-info 3 1) ;; force fail, this used to be next-state but that doesn't make sense. should always be "COMPLETED" 
	 (tests:test-set-status! run-id test-id "COMPLETED" "FAIL" (conc "Failed at step " stepname) #f)
	 )))
    logpro-used))

(define (launch:manage-steps run-id test-id item-path fullrunscript ezsteps test-name tconfigreg exit-info m)
  ;; (let-values
  ;;  (((pid exit-status exit-code)
  ;;    (run-n-wait fullrunscript)))
  ;; (tests:test-set-status! test-id "RUNNING" "n/a" #f #f)
  ;; Since we should have a clean slate at this time there is no need to do 
  ;; any of the other stuff that tests:test-set-status! does. Let's just 
  ;; force RUNNING/n/a

  ;; (thread-sleep! 0.3)
  ;; (tests:test-force-state-status! run-id test-id "RUNNING" "n/a")
  (rmt:set-state-status-and-roll-up-items run-id test-name item-path "RUNNING" #f #f) 
  ;; (thread-sleep! 0.3) ;; NFS slowness has caused grief here

  ;; if there is a runscript do it first
  (if fullrunscript
      (let ((pid (process-run fullrunscript)))
	(rmt:test-set-top-process-pid run-id test-id pid)
	(let loop ((i 0))
	  (let-values
	   (((pid-val exit-status exit-code) (process-wait pid #t)))
	   (mutex-lock! m)
	   (launch:einf-pid-set!           exit-info  pid)         ;; (vector-set! exit-info 0 pid)
	   (launch:einf-exit-status-set!   exit-info  exit-status) ;; (vector-set! exit-info 1 exit-status)
	   (launch:einf-exit-code-set!     exit-info  exit-code)   ;; (vector-set! exit-info 2 exit-code)







|





|





|





|





|



|


|



|










|





|







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
      (debug:print 4 *default-log-port* "Exit value received: " (launch:einf-exit-code exit-info) " logpro-used: " logpro-used 
		   " this-step-status: " this-step-status " overall-status: " overall-status 
		   " next-status: " next-status " rollup-status: "  (launch:einf-rollup-status exit-info)) ;; (vector-ref exit-info 3))
      (case next-status
	((warn)
	 (launch:einf-rollup-status-set! exit-info 2) ;; (vector-set! exit-info 3 2) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! area-dat run-id test-id next-state "WARN" 
				 (if (eq? this-step-status 'warn) "Logpro warning found" #f)
				 #f))
	((check)
	 (launch:einf-rollup-status-set! exit-info 3) ;; (vector-set! exit-info 3 3) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! area-dat run-id test-id next-state "CHECK" 
				 (if (eq? this-step-status 'check) "Logpro check found" #f)
				 #f))
	((waived)
	 (launch:einf-rollup-status-set! exit-info 4) ;; (vector-set! exit-info 3 3) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! area-dat run-id test-id next-state "WAIVED" 
				 (if (eq? this-step-status 'check) "Logpro waived found" #f)
				 #f))
	((abort)
	 (launch:einf-rollup-status-set! exit-info 5) ;; (vector-set! exit-info 3 4) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! area-dat run-id test-id next-state "ABORT" 
				 (if (eq? this-step-status 'abort) "Logpro abort found" #f)
				 #f))
	((skip)
	 (launch:einf-rollup-status-set! exit-info 6) ;; (vector-set! exit-info 3 4) ;; rollup-status
	 ;; NB// test-set-status! does rdb calls under the hood
	 (tests:test-set-status! area-dat run-id test-id next-state "SKIP" 
				 (if (eq? this-step-status 'skip) "Logpro skip found" #f)
				 #f))
	((pass)
	 (tests:test-set-status! area-dat run-id test-id next-state "PASS" #f #f))
	(else ;; 'fail
	 (launch:einf-rollup-status-set! exit-info 1) ;; (vector-set! exit-info 3 1) ;; force fail, this used to be next-state but that doesn't make sense. should always be "COMPLETED" 
	 (tests:test-set-status! area-dat run-id test-id "COMPLETED" "FAIL" (conc "Failed at step " stepname) #f)
	 )))
    logpro-used))

(define (launch:manage-steps area-dat run-id test-id item-path fullrunscript ezsteps test-name tconfigreg exit-info m)
  ;; (let-values
  ;;  (((pid exit-status exit-code)
  ;;    (run-n-wait fullrunscript)))
  ;; (tests:test-set-status! test-id "RUNNING" "n/a" #f #f)
  ;; Since we should have a clean slate at this time there is no need to do 
  ;; any of the other stuff that tests:test-set-status! does. Let's just 
  ;; force RUNNING/n/a

  ;; (thread-sleep! 0.3)
  ;; (tests:test-force-state-status! run-id test-id "RUNNING" "n/a")
  (rmt:set-state-status-and-roll-up-items area-dat run-id test-name item-path "RUNNING" #f #f) 
  ;; (thread-sleep! 0.3) ;; NFS slowness has caused grief here

  ;; if there is a runscript do it first
  (if fullrunscript
      (let ((pid (process-run fullrunscript)))
	(rmt:test-set-top-process-pid area-dat run-id test-id pid)
	(let loop ((i 0))
	  (let-values
	   (((pid-val exit-status exit-code) (process-wait pid #t)))
	   (mutex-lock! m)
	   (launch:einf-pid-set!           exit-info  pid)         ;; (vector-set! exit-info 0 pid)
	   (launch:einf-exit-status-set!   exit-info  exit-status) ;; (vector-set! exit-info 1 exit-status)
	   (launch:einf-exit-code-set!     exit-info  exit-code)   ;; (vector-set! exit-info 2 exit-code)
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
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
			(launch:load-logpro-dat run-id test-id stepname))
		    (if (steprun-good? logpro-used (launch:einf-exit-code exit-info))
			(if (not (null? tal))
			    (loop (car tal) (cdr tal) stepname))
			(debug:print 4 *default-log-port* "WARNING: step " (car ezstep) " failed. Stopping")))
		  (debug:print 4 *default-log-port* "WARNING: a prior step failed, stopping at " ezstep)))))))

(define (launch:monitor-job run-id test-id item-path fullrunscript ezsteps test-name tconfigreg exit-info m work-area runtlim misc-flags)
  (let* ((start-seconds (current-seconds))
	 (calc-minutes  (lambda ()
			  (inexact->exact 
			   (round 
			    (- 
			     (current-seconds) 
			     start-seconds)))))
	 (kill-tries 0))
    ;; (tests:set-full-meta-info #f test-id run-id (calc-minutes) work-area)
    ;; (tests:set-full-meta-info test-id run-id (calc-minutes) work-area)
    (tests:set-full-meta-info #f test-id run-id (calc-minutes) work-area 10)
    (let loop ((minutes   (calc-minutes))
	       (cpu-load  (alist-ref 'adj-core-load (common:get-normalized-cpu-load #f)))
	       (disk-free (get-df (current-directory))))
      (let ((new-cpu-load (let* ((load  (alist-ref 'adj-core-load (common:get-normalized-cpu-load #f)))
				 (delta (abs (- load cpu-load))))
			    (if (> delta 0.1) ;; don't bother updating with small changes
				load
				#f)))
	    (new-disk-free (let* ((df    (get-df (current-directory)))
				  (delta (abs (- df disk-free))))
			     (if (> delta 200) ;; ignore changes under 200 Meg
				 df
				 #f))))
	(set! kill-job? (or (test-get-kill-request run-id test-id) ;; run-id test-name itemdat))
			    (and runtlim (let* ((run-seconds   (- (current-seconds) start-seconds))
						(time-exceeded (> run-seconds runtlim)))
					   (if time-exceeded
					       (begin
						 (debug:print-info 0 *default-log-port* "KILLING TEST DUE TO TIME LIMIT EXCEEDED! Runtime=" run-seconds " seconds, limit=" runtlim)
						 #t)
					       #f)))))
	(tests:update-central-meta-info run-id test-id new-cpu-load new-disk-free (calc-minutes) #f #f)
	(if kill-job? 
	    (begin
	      (mutex-lock! m)
	      ;; NOTE: The pid can change as different steps are run. Do we need handshaking between this
	      ;;       section and the runit section? Or add a loop that tries three times with a 1/4 second
	      ;;       between tries?
	      (let* ((pid1 (launch:einf-pid exit-info)) ;; (vector-ref exit-info 0))
		     (pid2 (rmt:test-get-top-process-pid run-id test-id))
		     (pids (delete-duplicates (filter number? (list pid1 pid2)))))
		(if (not (null? pids))
		    (begin
		      (for-each
		       (lambda (pid)
			 (handle-exceptions
			  exn







|










|













|







|







|







314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
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
			(launch:load-logpro-dat run-id test-id stepname))
		    (if (steprun-good? logpro-used (launch:einf-exit-code exit-info))
			(if (not (null? tal))
			    (loop (car tal) (cdr tal) stepname))
			(debug:print 4 *default-log-port* "WARNING: step " (car ezstep) " failed. Stopping")))
		  (debug:print 4 *default-log-port* "WARNING: a prior step failed, stopping at " ezstep)))))))

(define (launch:monitor-job area-dat run-id test-id item-path fullrunscript ezsteps test-name tconfigreg exit-info m work-area runtlim misc-flags)
  (let* ((start-seconds (current-seconds))
	 (calc-minutes  (lambda ()
			  (inexact->exact 
			   (round 
			    (- 
			     (current-seconds) 
			     start-seconds)))))
	 (kill-tries 0))
    ;; (tests:set-full-meta-info #f test-id run-id (calc-minutes) work-area)
    ;; (tests:set-full-meta-info test-id run-id (calc-minutes) work-area)
    (tests:set-full-meta-info #f area-dat test-id run-id (calc-minutes) work-area 10)
    (let loop ((minutes   (calc-minutes))
	       (cpu-load  (alist-ref 'adj-core-load (common:get-normalized-cpu-load #f)))
	       (disk-free (get-df (current-directory))))
      (let ((new-cpu-load (let* ((load  (alist-ref 'adj-core-load (common:get-normalized-cpu-load #f)))
				 (delta (abs (- load cpu-load))))
			    (if (> delta 0.1) ;; don't bother updating with small changes
				load
				#f)))
	    (new-disk-free (let* ((df    (get-df (current-directory)))
				  (delta (abs (- df disk-free))))
			     (if (> delta 200) ;; ignore changes under 200 Meg
				 df
				 #f))))
	(set! kill-job? (or (test-get-kill-request area-dat run-id test-id) ;; run-id test-name itemdat))
			    (and runtlim (let* ((run-seconds   (- (current-seconds) start-seconds))
						(time-exceeded (> run-seconds runtlim)))
					   (if time-exceeded
					       (begin
						 (debug:print-info 0 *default-log-port* "KILLING TEST DUE TO TIME LIMIT EXCEEDED! Runtime=" run-seconds " seconds, limit=" runtlim)
						 #t)
					       #f)))))
	(tests:update-central-meta-info area-dat run-id test-id new-cpu-load new-disk-free (calc-minutes) #f #f)
	(if kill-job? 
	    (begin
	      (mutex-lock! m)
	      ;; NOTE: The pid can change as different steps are run. Do we need handshaking between this
	      ;;       section and the runit section? Or add a loop that tries three times with a 1/4 second
	      ;;       between tries?
	      (let* ((pid1 (launch:einf-pid exit-info)) ;; (vector-ref exit-info 0))
		     (pid2 (rmt:test-get-top-process-pid area-dat run-id test-id))
		     (pids (delete-duplicates (filter number? (list pid1 pid2)))))
		(if (not (null? pids))
		    (begin
		      (for-each
		       (lambda (pid)
			 (handle-exceptions
			  exn
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
				 (handle-exceptions
				  exn
				  #f
				  (process-signal pid-num signal/kill)))
			       (process:get-sub-pids pid))))
		       ;;    (debug:print-info 0 *default-log-port* "not killing process " pid " as it is not alive"))))
		       pids)
		      (tests:test-set-status! run-id test-id "KILLED"  "KILLED" (args:get-arg "-m") #f))
		    (begin
		      (debug:print-error 0 *default-log-port* "Nothing to kill, pid1=" pid1 ", pid2=" pid2)
		      (tests:test-set-status! run-id test-id "KILLED"  "FAILED TO KILL" (args:get-arg "-m") #f)
		      )))
	      (mutex-unlock! m)
	      ;; no point in sticking around. Exit now.
	      (exit)))
	(if (hash-table-ref/default misc-flags 'keep-going #f)
	    (begin
	      (thread-sleep! 3) ;; (+ 3 (random 6))) ;; add some jitter to the call home time to spread out the db accesses
	      (if (hash-table-ref/default misc-flags 'keep-going #f)  ;; keep originals for cpu-load and disk-free unless they change more than the allowed delta
		  (loop (calc-minutes) (or new-cpu-load cpu-load) (or new-disk-free disk-free)))))))
    (tests:update-central-meta-info run-id test-id (get-cpu-load) (get-df (current-directory))(calc-minutes) #f #f))) ;; NOTE: Checking twice for keep-going is intentional

(define (launch:execute encoded-cmd)
  (let* ((cmdinfo    (common:read-encoded-string encoded-cmd))
	 (tconfigreg #f))
    (setenv "MT_CMDINFO" encoded-cmd)
    (if (list? cmdinfo) ;; ((testpath /tmp/mrwellan/jazzmind/src/example_run/tests/sqlitespeed)
	;; (test-name sqlitespeed) (runscript runscript.rb) (db-host localhost) (run-id 1))







|


|









|







383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
				 (handle-exceptions
				  exn
				  #f
				  (process-signal pid-num signal/kill)))
			       (process:get-sub-pids pid))))
		       ;;    (debug:print-info 0 *default-log-port* "not killing process " pid " as it is not alive"))))
		       pids)
		      (tests:test-set-status! area-dat run-id test-id "KILLED"  "KILLED" (args:get-arg "-m") #f))
		    (begin
		      (debug:print-error 0 *default-log-port* "Nothing to kill, pid1=" pid1 ", pid2=" pid2)
		      (tests:test-set-status! area-dat run-id test-id "KILLED"  "FAILED TO KILL" (args:get-arg "-m") #f)
		      )))
	      (mutex-unlock! m)
	      ;; no point in sticking around. Exit now.
	      (exit)))
	(if (hash-table-ref/default misc-flags 'keep-going #f)
	    (begin
	      (thread-sleep! 3) ;; (+ 3 (random 6))) ;; add some jitter to the call home time to spread out the db accesses
	      (if (hash-table-ref/default misc-flags 'keep-going #f)  ;; keep originals for cpu-load and disk-free unless they change more than the allowed delta
		  (loop (calc-minutes) (or new-cpu-load cpu-load) (or new-disk-free disk-free)))))))
    (tests:update-central-meta-info area-dat run-id test-id (get-cpu-load) (get-df (current-directory))(calc-minutes) #f #f))) ;; NOTE: Checking twice for keep-going is intentional

(define (launch:execute encoded-cmd)
  (let* ((cmdinfo    (common:read-encoded-string encoded-cmd))
	 (tconfigreg #f))
    (setenv "MT_CMDINFO" encoded-cmd)
    (if (list? cmdinfo) ;; ((testpath /tmp/mrwellan/jazzmind/src/example_run/tests/sqlitespeed)
	;; (test-name sqlitespeed) (runscript runscript.rb) (db-host localhost) (run-id 1))
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
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
                                  (if (substring-index "/" runscript)
                                      runscript ;; use unadultered if contains slashes
                                      (let ((fulln (conc testpath "/" runscript)))
	                                  (if (and (file-exists? fulln)
                                                   (file-execute-access? fulln))
                                              fulln
                                              runscript))))) ;; assume it is on the path
	       ) ;; (rollup-status 0)

	  ;; NFS might not have propagated the directory meta data to the run host - give it time if needed
	  (let loop ((count 0))
	    (if (or (file-exists? top-path)
		    (> count 10))
		(change-directory top-path)
		(begin
		  (debug:print 0 *default-log-port* "INFO: Not starting job yet - directory " top-path " not found")
		  (thread-sleep! 10)
		  (loop (+ count 1)))))
	  (launch:setup) ;; should be properly in the top-path now
	  (set! tconfigreg (tests:get-all))
	  (let ((sighand (lambda (signum)
			   ;; (signal-mask! signum) ;; to mask or not? seems to cause issues in exiting
			   (if (eq? signum signal/stop)
			       (debug:print-error 0 *default-log-port* "attempt to STOP process. Exiting."))
			   (set! *time-to-exit* #t)
			   (print "Received signal " signum ", cleaning up before exit. Please wait...")
			   (let ((th1 (make-thread (lambda ()
						     (rmt:test-set-state-status run-id test-id "INCOMPLETE" "KILLED" #f)
						     (print "Killed by signal " signum ". Exiting")
						     (thread-sleep! 1)
						     (exit 1))))
				 (th2 (make-thread (lambda ()
						     (thread-sleep! 2)
						     (debug:print 0 *default-log-port* "Done")
						     (exit 4)))))
			     (thread-start! th2)
			     (thread-start! th1)
			     (thread-join! th2)))))
	    (set-signal-handler! signal/int sighand)
	    (set-signal-handler! signal/term sighand)
	    ) ;; (set-signal-handler! signal/stop sighand)
	  
	  ;; Do not run the test if it is REMOVING, RUNNING, KILLREQ or REMOTEHOSTSTART,
	  ;; Mark the test as REMOTEHOSTSTART *IMMEDIATELY*
	  ;;
	  (let* ((test-info (rmt:get-test-info-by-id run-id test-id))
		 (test-host (db:test-get-host        test-info))
		 (test-pid  (db:test-get-process_id  test-info)))
	    (cond
	     ((member (db:test-get-state test-info) '("INCOMPLETE" "KILLED" "UNKNOWN" "KILLREQ" "STUCK")) ;; prior run of this test didn't complete, go ahead and try to rerun
	      (debug:print 0 *default-log-port* "INFO: test is INCOMPLETE or KILLED, treat this execute call as a rerun request")
	      ;; (tests:test-force-state-status! run-id test-id "REMOTEHOSTSTART" "n/a")
	      (rmt:test-set-state-status run-id test-id "REMOTEHOSTSTART" "n/a" #f)
	      ) ;; prime it for running
	     ((member (db:test-get-state test-info) '("RUNNING" "REMOTEHOSTSTART"))
	      (if (process:alive-on-host? test-host test-pid)
		  (debug:print-error 0 *default-log-port* "test state is "  (db:test-get-state test-info) " and process " test-pid " is still running on host " test-host ", cannot proceed")
		  ;; (tests:test-force-state-status! run-id test-id "REMOTEHOSTSTART" "n/a")
		  (rmt:test-set-state-status run-id test-id "REMOTEHOSTSTART" "n/a" #f)
		  ))
	     ((not (member (db:test-get-state test-info) '("REMOVING" "REMOTEHOSTSTART" "RUNNING" "KILLREQ")))
	      ;; (tests:test-force-state-status! run-id test-id "REMOTEHOSTSTART" "n/a")
	      (rmt:test-set-state-status run-id test-id "REMOTEHOSTSTART" "n/a" #f)
	      )
	     (else ;; (member (db:test-get-state test-info) '("REMOVING" "REMOTEHOSTSTART" "RUNNING" "KILLREQ"))
	      (debug:print-error 0 *default-log-port* "test state is " (db:test-get-state test-info) ", cannot proceed")
	      (exit))))
	  
	  (debug:print 2 *default-log-port* "Exectuing " test-name " (id: " test-id ") on " (get-host-name))
	  (set! keys       (rmt:get-keys))
	  ;; (runs:set-megatest-env-vars run-id inkeys: keys inkeyvals: keyvals) ;; these may be needed by the launching process
	  ;; one of these is defunct/redundant ...
	  (if (not (launch:setup force: #t))
	      (begin
		(debug:print 0 *default-log-port* "Failed to setup, exiting") 
		;; (sqlite3:finalize! db)
		;; (sqlite3:finalize! tdb)
		(exit 1)))







|



















|

















|






|





|



|






|
<







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
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505

506
507
508
509
510
511
512
                                  (if (substring-index "/" runscript)
                                      runscript ;; use unadultered if contains slashes
                                      (let ((fulln (conc testpath "/" runscript)))
	                                  (if (and (file-exists? fulln)
                                                   (file-execute-access? fulln))
                                              fulln
                                              runscript))))) ;; assume it is on the path
	       (area-dat (make-remote))) ;; (rollup-status 0)

	  ;; NFS might not have propagated the directory meta data to the run host - give it time if needed
	  (let loop ((count 0))
	    (if (or (file-exists? top-path)
		    (> count 10))
		(change-directory top-path)
		(begin
		  (debug:print 0 *default-log-port* "INFO: Not starting job yet - directory " top-path " not found")
		  (thread-sleep! 10)
		  (loop (+ count 1)))))
	  (launch:setup) ;; should be properly in the top-path now
	  (set! tconfigreg (tests:get-all))
	  (let ((sighand (lambda (signum)
			   ;; (signal-mask! signum) ;; to mask or not? seems to cause issues in exiting
			   (if (eq? signum signal/stop)
			       (debug:print-error 0 *default-log-port* "attempt to STOP process. Exiting."))
			   (set! *time-to-exit* #t)
			   (print "Received signal " signum ", cleaning up before exit. Please wait...")
			   (let ((th1 (make-thread (lambda ()
						     (rmt:test-set-state-status area-dat run-id test-id "INCOMPLETE" "KILLED" #f)
						     (print "Killed by signal " signum ". Exiting")
						     (thread-sleep! 1)
						     (exit 1))))
				 (th2 (make-thread (lambda ()
						     (thread-sleep! 2)
						     (debug:print 0 *default-log-port* "Done")
						     (exit 4)))))
			     (thread-start! th2)
			     (thread-start! th1)
			     (thread-join! th2)))))
	    (set-signal-handler! signal/int sighand)
	    (set-signal-handler! signal/term sighand)
	    ) ;; (set-signal-handler! signal/stop sighand)
	  
	  ;; Do not run the test if it is REMOVING, RUNNING, KILLREQ or REMOTEHOSTSTART,
	  ;; Mark the test as REMOTEHOSTSTART *IMMEDIATELY*
	  ;;
	  (let* ((test-info (rmt:get-test-info-by-id area-dat run-id test-id))
		 (test-host (db:test-get-host        test-info))
		 (test-pid  (db:test-get-process_id  test-info)))
	    (cond
	     ((member (db:test-get-state test-info) '("INCOMPLETE" "KILLED" "UNKNOWN" "KILLREQ" "STUCK")) ;; prior run of this test didn't complete, go ahead and try to rerun
	      (debug:print 0 *default-log-port* "INFO: test is INCOMPLETE or KILLED, treat this execute call as a rerun request")
	      ;; (tests:test-force-state-status! run-id test-id "REMOTEHOSTSTART" "n/a")
	      (rmt:test-set-state-status area-dat run-id test-id "REMOTEHOSTSTART" "n/a" #f)
	      ) ;; prime it for running
	     ((member (db:test-get-state test-info) '("RUNNING" "REMOTEHOSTSTART"))
	      (if (process:alive-on-host? test-host test-pid)
		  (debug:print-error 0 *default-log-port* "test state is "  (db:test-get-state test-info) " and process " test-pid " is still running on host " test-host ", cannot proceed")
		  ;; (tests:test-force-state-status! run-id test-id "REMOTEHOSTSTART" "n/a")
		  (rmt:test-set-state-status area-dat run-id test-id "REMOTEHOSTSTART" "n/a" #f)
		  ))
	     ((not (member (db:test-get-state test-info) '("REMOVING" "REMOTEHOSTSTART" "RUNNING" "KILLREQ")))
	      ;; (tests:test-force-state-status! run-id test-id "REMOTEHOSTSTART" "n/a")
	      (rmt:test-set-state-status area-dat run-id test-id "REMOTEHOSTSTART" "n/a" #f)
	      )
	     (else ;; (member (db:test-get-state test-info) '("REMOVING" "REMOTEHOSTSTART" "RUNNING" "KILLREQ"))
	      (debug:print-error 0 *default-log-port* "test state is " (db:test-get-state test-info) ", cannot proceed")
	      (exit))))
	  
	  (debug:print 2 *default-log-port* "Exectuing " test-name " (id: " test-id ") on " (get-host-name))
	  (set! keys       (rmt:get-keys area-dat))

	  ;; one of these is defunct/redundant ...
	  (if (not (launch:setup force: #t))
	      (begin
		(debug:print 0 *default-log-port* "Failed to setup, exiting") 
		;; (sqlite3:finalize! db)
		;; (sqlite3:finalize! tdb)
		(exit 1)))
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
	  ;; (change-directory top-path)
	  ;; Can setup as client for server mode now
	  ;; (client:setup)

	  
	  ;; environment overrides are done *before* the remaining critical envars.
	  (alist->env-vars env-ovrd)
	  (runs:set-megatest-env-vars run-id inkeys: keys inkeyvals: keyvals)
	  (set-item-env-vars itemdat)
	  (save-environment-as-files "megatest")
	  ;; open-run-close not needed for test-set-meta-info
	  ;; (tests:set-full-meta-info #f test-id run-id 0 work-area)
	  ;; (tests:set-full-meta-info test-id run-id 0 work-area)
	  (tests:set-full-meta-info #f test-id run-id 0 work-area 10)

	  ;; (thread-sleep! 0.3) ;; NFS slowness has caused grief here

	  (if (args:get-arg "-xterm")
	      (set! fullrunscript "xterm")
	      (if (and fullrunscript 
		       (file-exists? fullrunscript)







|





|







580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
	  ;; (change-directory top-path)
	  ;; Can setup as client for server mode now
	  ;; (client:setup)

	  
	  ;; environment overrides are done *before* the remaining critical envars.
	  (alist->env-vars env-ovrd)
	  (runs:set-megatest-env-vars area-dat run-id inkeys: keys inkeyvals: keyvals)
	  (set-item-env-vars itemdat)
	  (save-environment-as-files "megatest")
	  ;; open-run-close not needed for test-set-meta-info
	  ;; (tests:set-full-meta-info #f test-id run-id 0 work-area)
	  ;; (tests:set-full-meta-info test-id run-id 0 work-area)
	  (tests:set-full-meta-info #f area-dat test-id run-id 0 work-area 10)

	  ;; (thread-sleep! 0.3) ;; NFS slowness has caused grief here

	  (if (args:get-arg "-xterm")
	      (set! fullrunscript "xterm")
	      (if (and fullrunscript 
		       (file-exists? fullrunscript)
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
	    (debug:print-info 0 *default-log-port* "Megatest exectute of test " test-name ", item path " item-path " complete. Notifying the db ...")
	    (hash-table-set! misc-flags 'keep-going #f)
	    (thread-join! th1)
	    (thread-sleep! 1)       ;; givbe thread th1 a chance to be done TODO: Verify this is needed. At 0.1 I was getting fail to stop, increased to total of 1.1 sec.
	    (mutex-lock! m)
	    (let* ((item-path (item-list->path itemdat))
		   ;; only state and status needed - use lazy routine
		   (testinfo  (rmt:get-testinfo-state-status run-id test-id)))
	      ;; Am I completed?
	      (if (member (db:test-get-state testinfo) '("REMOTEHOSTSTART" "RUNNING")) ;; NOTE: It should *not* be REMOTEHOSTSTART but for reasons I don't yet understand it sometimes gets stuck in that state ;; (not (equal? (db:test-get-state testinfo) "COMPLETED"))
		  (let ((new-state  (if kill-job? "KILLED" "COMPLETED") ;; (if (eq? (vector-ref exit-info 2) 0) ;; exited with "good" status
				                                        ;; "COMPLETED"
							                ;; (db:test-get-state testinfo)))   ;; else preseve the state as set within the test
				    )
			(new-status (cond







|







627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
	    (debug:print-info 0 *default-log-port* "Megatest exectute of test " test-name ", item path " item-path " complete. Notifying the db ...")
	    (hash-table-set! misc-flags 'keep-going #f)
	    (thread-join! th1)
	    (thread-sleep! 1)       ;; givbe thread th1 a chance to be done TODO: Verify this is needed. At 0.1 I was getting fail to stop, increased to total of 1.1 sec.
	    (mutex-lock! m)
	    (let* ((item-path (item-list->path itemdat))
		   ;; only state and status needed - use lazy routine
		   (testinfo  (rmt:get-testinfo-state-status area-dat run-id test-id)))
	      ;; Am I completed?
	      (if (member (db:test-get-state testinfo) '("REMOTEHOSTSTART" "RUNNING")) ;; NOTE: It should *not* be REMOTEHOSTSTART but for reasons I don't yet understand it sometimes gets stuck in that state ;; (not (equal? (db:test-get-state testinfo) "COMPLETED"))
		  (let ((new-state  (if kill-job? "KILLED" "COMPLETED") ;; (if (eq? (vector-ref exit-info 2) 0) ;; exited with "good" status
				                                        ;; "COMPLETED"
							                ;; (db:test-get-state testinfo)))   ;; else preseve the state as set within the test
				    )
			(new-status (cond
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
				      (if (equal? (db:test-get-status testinfo) "AUTO") "AUTO-WARN" "WARN"))
				     ((eq? (launch:einf-rollup-status exit-info) 3) "CHECK")
				     ((eq? (launch:einf-rollup-status exit-info) 4) "WAIVED")
				     ((eq? (launch:einf-rollup-status exit-info) 5) "ABORT")
				     ((eq? (launch:einf-rollup-status exit-info) 6) "SKIP")
				     (else "FAIL")))) ;; (db:test-get-status testinfo)))
		    (debug:print-info 1 *default-log-port* "Test exited in state=" (db:test-get-state testinfo) ", setting state/status based on exit code of " (launch:einf-exit-status exit-info) " and rollup-status of " (launch:einf-rollup-status exit-info))
		    (tests:test-set-status! run-id 
					    test-id 
					    new-state
					    new-status
					    (args:get-arg "-m") #f)
		    ;; need to update the top test record if PASS or FAIL and this is a subtest
		    ;; NO NEED TO CALL set-state-status-and-roll-up-items HERE, THIS IS DONE IN set-state-status-and-roll-up-items called by tests:test-set-status!
		    ))
	      ;; for automated creation of the rollup html file this is a good place...
	      (if (not (equal? item-path ""))
		  (tests:summarize-items run-id test-id test-name #f))
	      (tests:summarize-test run-id test-id)  ;; don't force - just update if no
	      (rmt:update-run-stats run-id (rmt:get-raw-run-stats run-id)))
	    (mutex-unlock! m)
	    (debug:print 2 *default-log-port* "Output from running " fullrunscript ", pid " (launch:einf-pid exit-info) " in work area " 
			 work-area ":\n====\n exit code " (launch:einf-exit-code exit-info) "\n" "====\n")
	    (if (not (launch:einf-exit-status exit-info))
		(exit 4)))))))

(define (launch:cache-config)







|









|
|
|







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
				      (if (equal? (db:test-get-status testinfo) "AUTO") "AUTO-WARN" "WARN"))
				     ((eq? (launch:einf-rollup-status exit-info) 3) "CHECK")
				     ((eq? (launch:einf-rollup-status exit-info) 4) "WAIVED")
				     ((eq? (launch:einf-rollup-status exit-info) 5) "ABORT")
				     ((eq? (launch:einf-rollup-status exit-info) 6) "SKIP")
				     (else "FAIL")))) ;; (db:test-get-status testinfo)))
		    (debug:print-info 1 *default-log-port* "Test exited in state=" (db:test-get-state testinfo) ", setting state/status based on exit code of " (launch:einf-exit-status exit-info) " and rollup-status of " (launch:einf-rollup-status exit-info))
		    (tests:test-set-status! area-dat run-id 
					    test-id 
					    new-state
					    new-status
					    (args:get-arg "-m") #f)
		    ;; need to update the top test record if PASS or FAIL and this is a subtest
		    ;; NO NEED TO CALL set-state-status-and-roll-up-items HERE, THIS IS DONE IN set-state-status-and-roll-up-items called by tests:test-set-status!
		    ))
	      ;; for automated creation of the rollup html file this is a good place...
	      (if (not (equal? item-path ""))
		  (tests:summarize-items area-dat run-id test-id test-name #f))
	      (tests:summarize-test area-dat run-id test-id)  ;; don't force - just update if no
	      (rmt:update-run-stats area-dat run-id (rmt:get-raw-run-stats area-dat run-id)))
	    (mutex-unlock! m)
	    (debug:print 2 *default-log-port* "Output from running " fullrunscript ", pid " (launch:einf-pid exit-info) " in work area " 
			 work-area ":\n====\n exit code " (launch:einf-exit-code exit-info) "\n" "====\n")
	    (if (not (launch:einf-exit-status exit-info))
		(exit 4)))))))

(define (launch:cache-config)
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
	(debug:print 0 *default-log-port* "NOTE: skipping launch:setup-body call since we have fulldata")
	(mutex-unlock! *launch-setup-mutex*)
	*toppath*)
      (let ((res (launch:setup-body force: force)))
	(mutex-unlock! *launch-setup-mutex*)
	res)))

(define (launch:setup-body #!key (force #f))
  (let* ((toppath  (or *toppath* (getenv "MT_RUN_AREA_HOME"))) ;; preserve toppath
	 (runname  (common:args-get-runname))
	 (target   (common:args-get-target))
	 (linktree (common:get-linktree))
	 (sections (if target (list "default" target) #f)) ;; for runconfigs
	 (mtconfig (or (args:get-arg "-config") "megatest.config")) ;; allow overriding megatest.config 
	 (rundir   (if (and runname target linktree)(conc linktree "/" target "/" runname) #f))







|







730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
	(debug:print 0 *default-log-port* "NOTE: skipping launch:setup-body call since we have fulldata")
	(mutex-unlock! *launch-setup-mutex*)
	*toppath*)
      (let ((res (launch:setup-body force: force)))
	(mutex-unlock! *launch-setup-mutex*)
	res)))

(define (launch:setup-body area-dat #!key (force #f))
  (let* ((toppath  (or *toppath* (getenv "MT_RUN_AREA_HOME"))) ;; preserve toppath
	 (runname  (common:args-get-runname))
	 (target   (common:args-get-target))
	 (linktree (common:get-linktree))
	 (sections (if target (list "default" target) #f)) ;; for runconfigs
	 (mtconfig (or (args:get-arg "-config") "megatest.config")) ;; allow overriding megatest.config 
	 (rundir   (if (and runname target linktree)(conc linktree "/" target "/" runname) #f))
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
	      (set! toppath      *toppath*)
	      (if (not *toppath*)
		  (begin
		    (debug:print-error 0 *default-log-port* "you are not in a megatest area!")
		    (exit 1)))
	      (setenv "MT_RUN_AREA_HOME" *toppath*)
	      ;; the seed read is done, now read runconfigs, cache it then read megatest.config one more time and cache it
	      (let* ((keys         (rmt:get-keys))
		     (key-vals     (keys:target->keyval keys target))
		     (linktree     (or (getenv "MT_LINKTREE")
				       (if *configdat* (configf:lookup *configdat* "setup" "linktree") #f)))
		     (second-pass  (find-and-read-config
				    mtconfig
				    environ-patt: "env-override"
				    given-toppath: toppath







|







790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
	      (set! toppath      *toppath*)
	      (if (not *toppath*)
		  (begin
		    (debug:print-error 0 *default-log-port* "you are not in a megatest area!")
		    (exit 1)))
	      (setenv "MT_RUN_AREA_HOME" *toppath*)
	      ;; the seed read is done, now read runconfigs, cache it then read megatest.config one more time and cache it
	      (let* ((keys         (rmt:get-keys area-dat))
		     (key-vals     (keys:target->keyval keys target))
		     (linktree     (or (getenv "MT_LINKTREE")
				       (if *configdat* (configf:lookup *configdat* "setup" "linktree") #f)))
		     (second-pass  (find-and-read-config
				    mtconfig
				    environ-patt: "env-override"
				    given-toppath: toppath
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
	 (lnkbase   (conc linktree "/" target "/" runname))
	 (lnkpath   (conc lnkbase "/" testname))
	 (lnkpathf  (conc lnkpath (if not-iterated "" "/") item-path))
	 (lnktarget (conc lnkpath "/" item-path)))

    ;; Update the rundir path in the test record for all, rundir=physical, shortdir=logical
    ;;                                                 rundir   shortdir
    (rmt:general-call 'test-set-rundir-shortdir run-id lnkpathf test-path testname item-path run-id)

    (debug:print 2 *default-log-port* "INFO:\n       lnkbase=" lnkbase "\n       lnkpath=" lnkpath "\n  toptest-path=" toptest-path "\n     test-path=" test-path)
    (if (not (file-exists? linktree))
	(begin
	  (debug:print 0 *default-log-port* "WARNING: linktree did not exist! Creating it now at " linktree)
	  (create-directory linktree #t))) ;; (system (conc "mkdir -p " linktree))))
    ;; create the directory for the tests dir links, this is needed no matter what...







|







928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
	 (lnkbase   (conc linktree "/" target "/" runname))
	 (lnkpath   (conc lnkbase "/" testname))
	 (lnkpathf  (conc lnkpath (if not-iterated "" "/") item-path))
	 (lnktarget (conc lnkpath "/" item-path)))

    ;; Update the rundir path in the test record for all, rundir=physical, shortdir=logical
    ;;                                                 rundir   shortdir
    (rmt:general-call area-dat 'test-set-rundir-shortdir run-id lnkpathf test-path testname item-path run-id)

    (debug:print 2 *default-log-port* "INFO:\n       lnkbase=" lnkbase "\n       lnkpath=" lnkpath "\n  toptest-path=" toptest-path "\n     test-path=" test-path)
    (if (not (file-exists? linktree))
	(begin
	  (debug:print 0 *default-log-port* "WARNING: linktree did not exist! Creating it now at " linktree)
	  (create-directory linktree #t))) ;; (system (conc "mkdir -p " linktree))))
    ;; create the directory for the tests dir links, this is needed no matter what...
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
    ;; NB - This was not working right - some top tests are not getting the path set!!!
    ;;
    ;; Do the setting of this record after the paths are created so that the shortdir can 
    ;; be set to the real directory location. This is safer for future clean up if the link
    ;; tree is damaged or lost.
    ;; 
    (if (not (hash-table-ref/default *toptest-paths* testname #f))
	(let* ((testinfo       (rmt:get-test-info-by-id run-id test-id)) ;;  run-id testname item-path))
	       (curr-test-path (if testinfo ;; (filedb:get-path *fdb*
							     ;; (db:get-path dbstruct
				   ;; (rmt:sdb-qry 'getstr 
				   (db:test-get-rundir testinfo) ;; ) ;; )
				   #f)))
	  (hash-table-set! *toptest-paths* testname curr-test-path)
	  ;; NB// Was this for the test or for the parent in an iterated test?
	  (rmt:general-call 'test-set-rundir-shortdir run-id lnkpath 
			    (if (file-exists? lnkpath)
				;; (resolve-pathname lnkpath)
				(common:nice-path lnkpath)
				lnkpath)
			    testname "" run-id)
	  ;; (rmt:general-call 'test-set-rundir run-id lnkpath testname "") ;; toptest-path)
	  (if (or (not curr-test-path)
		  (not (directory-exists? toptest-path)))
	      (begin
		(debug:print-info 2 *default-log-port* "Creating " toptest-path " and link " lnkpath)
		(handle-exceptions
		 exn
		 #f ;; don't care to catch and deal with errors here for now.







|


|




|





|







988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
    ;; NB - This was not working right - some top tests are not getting the path set!!!
    ;;
    ;; Do the setting of this record after the paths are created so that the shortdir can 
    ;; be set to the real directory location. This is safer for future clean up if the link
    ;; tree is damaged or lost.
    ;; 
    (if (not (hash-table-ref/default *toptest-paths* testname #f))
	(let* ((testinfo       (rmt:get-test-info-by-id area-dat run-id test-id)) ;;  run-id testname item-path))
	       (curr-test-path (if testinfo ;; (filedb:get-path *fdb*
							     ;; (db:get-path dbstruct
				   ;; (rmt:sdb-qry area-dat 'getstr 
				   (db:test-get-rundir testinfo) ;; ) ;; )
				   #f)))
	  (hash-table-set! *toptest-paths* testname curr-test-path)
	  ;; NB// Was this for the test or for the parent in an iterated test?
	  (rmt:general-call area-dat 'test-set-rundir-shortdir run-id lnkpath 
			    (if (file-exists? lnkpath)
				;; (resolve-pathname lnkpath)
				(common:nice-path lnkpath)
				lnkpath)
			    testname "" run-id)
	  ;; (rmt:general-call area-dat 'test-set-rundir run-id lnkpath testname "") ;; toptest-path)
	  (if (or (not curr-test-path)
		  (not (directory-exists? toptest-path)))
	      (begin
		(debug:print-info 2 *default-log-port* "Creating " toptest-path " and link " lnkpath)
		(handle-exceptions
		 exn
		 #f ;; don't care to catch and deal with errors here for now.
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
	   (test-sig        (conc (common:get-testsuite-name) ":" test-name ":" item-path)) ;; (item-list->path itemdat))) ;; test-path is the full path including the item-path
	   (work-area       #f)
	   (toptest-work-area #f) ;; for iterated tests the top test contains data relevant for all
	   (diskpath   #f)
	   (cmdparms   #f)
	   (fullcmd    #f) ;; (define a (with-output-to-string (lambda ()(write x))))
	   (mt-bindir-path #f)
	   (testinfo   (rmt:get-test-info-by-id run-id test-id))
	   (mt_target  (string-intersperse (map cadr keyvals) "/"))
	   (debug-param (append (if (args:get-arg "-debug")  (list "-debug" (args:get-arg "-debug")) '())
				(if (args:get-arg "-logging")(list "-logging") '()))))
      ;; (if hosts (set! hosts (string-split hosts)))
      ;; set the megatest to be called on the remote host
      (if (not remote-megatest)(set! remote-megatest local-megatest)) ;; "megatest"))
      (set! mt-bindir-path (pathname-directory remote-megatest))
      (if launcher (set! launcher (string-split launcher)))
      ;; set up the run work area for this test
      (if (and (args:get-arg "-preclean") ;; user has requested to preclean for this run
	       (not (member (db:test-get-rundir testinfo)(list "n/a" "/tmp/badname")))) ;; n/a is a placeholder and thus not a read dir
	  (begin
	    (debug:print-info 0 *default-log-port* "attempting to preclean directory " (db:test-get-rundir testinfo) " for test " test-name "/" item-path)
	    (runs:remove-test-directory testinfo 'remove-data-only))) ;; remove data only, do not perturb the record
      
      ;; prevent overlapping actions - set to LAUNCHED as early as possible
      ;;
      ;; the following call handles waiver propogation. cannot yet condense into roll-up-pass-fail
      (tests:test-set-status! run-id test-id "LAUNCHED" "n/a" #f #f) ;; (if launch-results launch-results "FAILED"))
      (rmt:set-state-status-and-roll-up-items run-id test-name item-path #f "LAUNCHED" #f)
      ;; (pp (hash-table->alist tconfig))
      (set! diskpath (get-best-disk *configdat* tconfig))
      (if diskpath
	  (let ((dat  (create-work-area run-id run-info keyvals test-id test-path diskpath test-name itemdat)))
	    (set! work-area (car dat))
	    (set! toptest-work-area (cadr dat))
	    (debug:print-info 2 *default-log-port* "Using work area " work-area))







|













|




|
|







1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
	   (test-sig        (conc (common:get-testsuite-name) ":" test-name ":" item-path)) ;; (item-list->path itemdat))) ;; test-path is the full path including the item-path
	   (work-area       #f)
	   (toptest-work-area #f) ;; for iterated tests the top test contains data relevant for all
	   (diskpath   #f)
	   (cmdparms   #f)
	   (fullcmd    #f) ;; (define a (with-output-to-string (lambda ()(write x))))
	   (mt-bindir-path #f)
	   (testinfo   (rmt:get-test-info-by-id area-dat run-id test-id))
	   (mt_target  (string-intersperse (map cadr keyvals) "/"))
	   (debug-param (append (if (args:get-arg "-debug")  (list "-debug" (args:get-arg "-debug")) '())
				(if (args:get-arg "-logging")(list "-logging") '()))))
      ;; (if hosts (set! hosts (string-split hosts)))
      ;; set the megatest to be called on the remote host
      (if (not remote-megatest)(set! remote-megatest local-megatest)) ;; "megatest"))
      (set! mt-bindir-path (pathname-directory remote-megatest))
      (if launcher (set! launcher (string-split launcher)))
      ;; set up the run work area for this test
      (if (and (args:get-arg "-preclean") ;; user has requested to preclean for this run
	       (not (member (db:test-get-rundir testinfo)(list "n/a" "/tmp/badname")))) ;; n/a is a placeholder and thus not a read dir
	  (begin
	    (debug:print-info 0 *default-log-port* "attempting to preclean directory " (db:test-get-rundir testinfo) " for test " test-name "/" item-path)
	    (runs:remove-test-directory area-dat testinfo 'remove-data-only))) ;; remove data only, do not perturb the record
      
      ;; prevent overlapping actions - set to LAUNCHED as early as possible
      ;;
      ;; the following call handles waiver propogation. cannot yet condense into roll-up-pass-fail
      (tests:test-set-status! area-dat run-id test-id "LAUNCHED" "n/a" #f #f) ;; (if launch-results launch-results "FAILED"))
      (rmt:set-state-status-and-roll-up-items area-dat run-id test-name item-path #f "LAUNCHED" #f)
      ;; (pp (hash-table->alist tconfig))
      (set! diskpath (get-best-disk *configdat* tconfig))
      (if diskpath
	  (let ((dat  (create-work-area run-id run-info keyvals test-id test-path diskpath test-name itemdat)))
	    (set! work-area (car dat))
	    (set! toptest-work-area (cadr dat))
	    (debug:print-info 2 *default-log-port* "Using work area " work-area))
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
					(list 'runtlim   (if run-time-limit (common:hms-string->seconds run-time-limit) #f))
					(list 'env-ovrd  (hash-table-ref/default *configdat* "env-override" '())) 
					(list 'set-vars  (if params (hash-table-ref/default params "-setvars" #f)))
					(list 'runname   runname)
					(list 'mt-bindir-path mt-bindir-path))))))))
      
      ;; clean out step records from previous run if they exist
      ;; (rmt:delete-test-step-records run-id test-id)
      ;; if the dir does not exist we may have a itempath where individual variables are a path, launch anyway
      (if (file-exists? work-area)
	  (change-directory work-area)) ;; so that log files from the launch process don't clutter the test dir
      (cond
       ;; ((and launcher hosts) ;; must be using ssh hostname
       ;;    (set! fullcmd (append launcher (car hosts)(list remote-megatest "-m" test-sig "-execute" cmdparms) debug-param)))
       ;; (set! fullcmd (append launcher (car hosts)(list remote-megatest test-sig "-execute" cmdparms))))







|







1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
					(list 'runtlim   (if run-time-limit (common:hms-string->seconds run-time-limit) #f))
					(list 'env-ovrd  (hash-table-ref/default *configdat* "env-override" '())) 
					(list 'set-vars  (if params (hash-table-ref/default params "-setvars" #f)))
					(list 'runname   runname)
					(list 'mt-bindir-path mt-bindir-path))))))))
      
      ;; clean out step records from previous run if they exist
      ;; (rmt:delete-test-step-records area-dat run-id test-id)
      ;; if the dir does not exist we may have a itempath where individual variables are a path, launch anyway
      (if (file-exists? work-area)
	  (change-directory work-area)) ;; so that log files from the launch process don't clutter the test dir
      (cond
       ;; ((and launcher hosts) ;; must be using ssh hostname
       ;;    (set! fullcmd (append launcher (car hosts)(list remote-megatest "-m" test-sig "-execute" cmdparms) debug-param)))
       ;; (set! fullcmd (append launcher (car hosts)(list remote-megatest test-sig "-execute" cmdparms))))
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
	(alist->env-vars testprevvals)
	(alist->env-vars commonprevvals)
	launch-results))
    (change-directory *toppath*)))

;; recover a test where the top controlling mtest may have died
;;
(define (launch:recover-test run-id test-id)
  ;; this function is called on the test run host via ssh
  ;;
  ;; 1. look at the process from pid
  ;;    - is it owned by calling user
  ;;    - it it's run directory correct for the test
  ;;    - is there a controlling mtest (maybe stuck)
  ;; 2. if recovery is needed watch pid
  ;;    - when it exits take the exit code and do the needful
  ;;
  (let* ((pid (rmt:test-get-top-process-id run-id test-id))
	 (psres (with-input-from-pipe
		 (conc "ps -F -u " (current-user-name) " | grep -E '" pid " ' | grep -v 'grep -E " pid "'")
		 (lambda ()
		   (read-line))))
	 (rundir (if (string? psres) ;; real process owned by user
		     (read-symbolic-link (conc "/proc/" pid "/cwd"))
		     #f)))
    ;; now wait on that process if all is correct
    ;; periodically update the db with runtime
    ;; when the process exits look at the db, if still RUNNING after 10 seconds set
    ;; state/status appropriately
    (process-wait pid)))







|









|












1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
	(alist->env-vars testprevvals)
	(alist->env-vars commonprevvals)
	launch-results))
    (change-directory *toppath*)))

;; recover a test where the top controlling mtest may have died
;;
(define (launch:recover-test area-dat run-id test-id)
  ;; this function is called on the test run host via ssh
  ;;
  ;; 1. look at the process from pid
  ;;    - is it owned by calling user
  ;;    - it it's run directory correct for the test
  ;;    - is there a controlling mtest (maybe stuck)
  ;; 2. if recovery is needed watch pid
  ;;    - when it exits take the exit code and do the needful
  ;;
  (let* ((pid (rmt:test-get-top-process-id area-dat run-id test-id))
	 (psres (with-input-from-pipe
		 (conc "ps -F -u " (current-user-name) " | grep -E '" pid " ' | grep -v 'grep -E " pid "'")
		 (lambda ()
		   (read-line))))
	 (rundir (if (string? psres) ;; real process owned by user
		     (read-symbolic-link (conc "/proc/" pid "/cwd"))
		     #f)))
    ;; now wait on that process if all is correct
    ;; periodically update the db with runtime
    ;; when the process exits look at the db, if still RUNNING after 10 seconds set
    ;; state/status appropriately
    (process-wait pid)))