Megatest

Diff
Login

Differences From Artifact [4e784cfd15]:

To Artifact [4553eebf08]:


316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
	 (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  (get-cpu-load))
	       (disk-free (get-df (current-directory))))
      (let ((new-cpu-load (let* ((load  (get-cpu-load))
				 (delta (abs (- load cpu-load))))
			    (if (> delta 0.6) ;; 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







|







316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
	 (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  (get-cpu-load))
	       (disk-free (get-df (current-directory))))
      (let ((new-cpu-load (let* ((load  (common:get-normalized-cpu-load #f))
				 (delta (abs (- load cpu-load))))
			    (if (> delta 0.6) ;; 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
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
    (if disks 
	(let ((res (common:get-disk-with-most-free-space disks minspace))) ;; min size of 1000, seems tad dumb
	  (if res
	      (cdr res)
	      (begin
		(if (common:low-noise-print 20 "No valid disks or no disk with enough space")
		    (debug:print-error 0 *default-log-port* "No valid disks found in megatest.config. Please add some to your [disks] section and ensure the directory exists and has enough space!\n    You can change minspace in the [setup] section of megatest.config. Current setting is: " minspace))
		(exit 1)))))))

;; Desired directory structure:
;;
;;  <linkdir> - <target> - <testname> -.
;;                                     |
;;                                     v
;;  <rundir>  -  <target>  -    <testname> -|- <itempath(s)>







|







859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
    (if disks 
	(let ((res (common:get-disk-with-most-free-space disks minspace))) ;; min size of 1000, seems tad dumb
	  (if res
	      (cdr res)
	      (begin
		(if (common:low-noise-print 20 "No valid disks or no disk with enough space")
		    (debug:print-error 0 *default-log-port* "No valid disks found in megatest.config. Please add some to your [disks] section and ensure the directory exists and has enough space!\n    You can change minspace in the [setup] section of megatest.config. Current setting is: " minspace))
		(exit 1))))))) ;; TODO - move the exit to the calling location and return #f

;; Desired directory structure:
;;
;;  <linkdir> - <target> - <testname> -.
;;                                     |
;;                                     v
;;  <rundir>  -  <target>  -    <testname> -|- <itempath(s)>
1051
1052
1053
1054
1055
1056
1057


1058
1059
1060
1061
1062
1063
1064
1065
1066
1067

1068
1069
1070
1071
1072
1073

1074
1075
1076
1077
1078
1079


1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
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
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192

1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205


1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219

1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
;; 2. create run dir on disk, path name is meaningful
;; 3. create link from run dir to megatest runs area 
;; 4. remotely run the test on allocated host
;;    - could be ssh to host from hosts table (update regularly with load)
;;    - could be netbatch
;;      (launch-test db (cadr status) test-conf))
(define (launch-test test-id run-id run-info keyvals runname test-conf test-name test-path itemdat params)


  (let loop ((delta        (- (current-seconds) *last-launch*))
	     (launch-delay (string->number (or (configf:lookup *configdat* "setup" "launch-delay") "5"))))
    (if (> launch-delay delta)
	(begin
	  (debug:print-info 0 *default-log-port* "Delaying launch of " test-name " for " (- launch-delay delta) " seconds")
	  (thread-sleep! (- launch-delay delta))
	  (loop (- (current-seconds) *last-launch*) launch-delay))))
  (set! *last-launch* (current-seconds))
  (change-directory *toppath*)
  (alist->env-vars ;; consolidate this code with the code in megatest.scm for "-execute"

   (list ;; (list "MT_TEST_RUN_DIR" work-area)
    (list "MT_RUN_AREA_HOME" *toppath*)
    (list "MT_TEST_NAME" test-name)
    ;; (list "MT_ITEM_INFO" (conc itemdat)) 
    (list "MT_RUNNAME"   runname)
    ;; (list "MT_TARGET"    mt_target)

    ))
  (let* ((tregistry       (tests:get-all))
	 (item-path       (let ((ip (item-list->path itemdat)))
			    (alist->env-vars (list (list "MT_ITEMPATH" ip)))
			    ip))
	 (tconfig         (or (tests:get-testconfig test-name tregistry #t force-create: #t)


			      test-conf)) ;; force re-read now that all vars are set
	 (useshell        (let ((ush (config-lookup *configdat* "jobtools"     "useshell")))
			    (if ush 
				(if (equal? ush "no") ;; must use "no" to NOT use shell
				    #f
				    ush)
				#t)))     ;; default is yes
	 (runscript       (config-lookup tconfig   "setup"        "runscript"))
	 (ezsteps         (> (length (hash-table-ref/default tconfig "ezsteps" '())) 0)) ;; don't send all the steps, could be big
	 (diskspace       (config-lookup tconfig   "requirements" "diskspace"))
	 (memory          (config-lookup tconfig   "requirements" "memory"))
	 (hosts           (config-lookup *configdat* "jobtools"     "workhosts"))
	 (remote-megatest (config-lookup *configdat* "setup" "executable"))
	 (run-time-limit  (or (configf:lookup  tconfig   "requirements" "runtimelim")
			      (configf:lookup  *configdat* "setup" "runtimelim")))
	 ;; FIXME SOMEDAY: not good how this is so obtuse, this hack is to 
	 ;;                allow running from dashboard. Extract the path
	 ;;                from the called megatest and convert dashboard
	 ;;             	  or dboard to megatest
	 (local-megatest  (let* ((lm  (car (argv)))
				 (dir (pathname-directory lm))
				 (exe (pathname-strip-directory lm)))
			    (conc (if dir (conc dir "/") "")
				  (case (string->symbol exe)
				    ((dboard)    "../megatest")
				    ((mtest)     "../megatest")
				    ((dashboard) "megatest")
				    (else exe)))))
	 (launcher        (common:get-launcher *configdat* test-name item-path)) ;; (config-lookup *configdat* "jobtools"     "launcher"))
	 (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") '()))))

    (setenv "MT_ITEMPATH" item-path)
    (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:roll-up-pass-fail-counts run-id test-name item-path #f "LAUNCHED" #f)

    (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))
	(begin
	  (set! work-area (conc test-path "/tmp_run"))
	  (create-directory work-area #t)
	  (debug:print 0 *default-log-port* "WARNING: No disk work area specified - running in the test directory under tmp_run")))
    (set! cmdparms (base64:base64-encode 
		    (z3:encode-buffer 
		     (with-output-to-string
		       (lambda () ;; (list 'hosts     hosts)
			 (write (list (list 'testpath  test-path)
				      (list 'transport (conc *transport-type*))
				      ;; (list 'serverinf *server-info*)
				      (list 'toppath   *toppath*)
				      (list 'work-area work-area)
				      (list 'test-name test-name) 
				      (list 'runscript runscript) 
				      (list 'run-id    run-id   )
				      (list 'test-id   test-id  )
				      ;; (list 'item-path item-path )
				      (list 'itemdat   itemdat  )
				      (list 'megatest  remote-megatest)
				      (list 'ezsteps   ezsteps) 
				      (list 'target    mt_target)
				      (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))))
     (launcher
      (set! fullcmd (append launcher (list remote-megatest "-m" test-sig "-execute" cmdparms) debug-param)))
     ;; (set! fullcmd (append launcher (list remote-megatest test-sig "-execute" cmdparms))))
     (else
      (if (not useshell)(debug:print 0 *default-log-port* "WARNING: internal launching will not work well without \"useshell yes\" in your [jobtools] section"))
      (set! fullcmd (append (list remote-megatest "-m" test-sig "-execute" cmdparms) debug-param (list (if useshell "&" ""))))))
    ;; (set! fullcmd (list remote-megatest test-sig "-execute" cmdparms (if useshell "&" "")))))
    (if (args:get-arg "-xterm")(set! fullcmd (append fullcmd (list "-xterm"))))
    (debug:print 1 *default-log-port* "Launching " work-area)
    ;; set pre-launch-env-vars before launching, keep the vars in prevvals and put the envionment back when done
    (debug:print 4 *default-log-port* "fullcmd: " fullcmd)

    (let* ((commonprevvals (alist->env-vars
			    (hash-table-ref/default *configdat* "env-override" '())))
	   (testprevvals   (alist->env-vars
			    (hash-table-ref/default tconfig "pre-launch-env-overrides" '())))
	   (miscprevvals   (alist->env-vars ;; consolidate this code with the code in megatest.scm for "-execute"
			    (append (list (list "MT_TEST_RUN_DIR" work-area)
					  (list "MT_TEST_NAME" test-name)
					  (list "MT_ITEM_INFO" (conc itemdat)) 
					  (list "MT_RUNNAME"   runname)
					  (list "MT_TARGET"    mt_target)
					  (list "MT_ITEMPATH"  item-path)
					  )
				    itemdat)))


	   ;; Launchwait defaults to true, must override it to turn off wait
	   (launchwait     (if (equal? (configf:lookup *configdat* "setup" "launchwait") "no") #f #t))
	   (launch-results (apply (if launchwait
				      process:cmd-run-with-stderr->list
				      process-run)
				  (if useshell
				      (let ((cmdstr (string-intersperse fullcmd " ")))
					(if launchwait
					    cmdstr
					    (conc cmdstr " >> mt_launch.log 2>&1")))
				      (car fullcmd))
				  (if useshell
				      '()
				      (cdr fullcmd)))))

      (if (not launchwait) ;; give the OS a little time to allow the process to start
	  (thread-sleep! 0.01))
      (with-output-to-file "mt_launch.log"
	(lambda ()
	  (print "LAUNCHCMD: " (string-intersperse fullcmd " "))
	  (if (list? launch-results)
	      (apply print launch-results)
	      (print "NOTE: launched \"" fullcmd "\"\n  but did not wait for it to proceed. Add the following to megatest.config \n[setup]\nlaunchwait yes\n  if you have problems with this"))
	  #:append))
      (debug:print 2 *default-log-port* "Launching completed, updating db")
      (debug:print 2 *default-log-port* "Launch results: " launch-results)
      (if (not launch-results)
          (begin
            (print "ERROR: Failed to run " (string-intersperse fullcmd " ") ", exiting now")
            ;; (sqlite3:finalize! db)
            ;; good ole "exit" seems not to work
            ;; (_exit 9)
            ;; but this hack will work! Thanks go to Alan Post of the Chicken email list
            ;; NB// Is this still needed? Should be safe to go back to "exit" now?
            (process-signal (current-process-id) signal/kill)
            ))
      (alist->env-vars miscprevvals)
      (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







>
>
|
|
|
|
|
|
|
<
|
|
>
|
|
|
<
|
|
>
|
|
<
<
|
|
>
>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
<
<
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
>
|
|
<
<
|
|
|
|
|
|
|
|
|
>
>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|







1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066

1067
1068
1069
1070
1071
1072

1073
1074
1075
1076
1077


1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121


1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
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
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196


1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
;; 2. create run dir on disk, path name is meaningful
;; 3. create link from run dir to megatest runs area 
;; 4. remotely run the test on allocated host
;;    - could be ssh to host from hosts table (update regularly with load)
;;    - could be netbatch
;;      (launch-test db (cadr status) test-conf))
(define (launch-test test-id run-id run-info keyvals runname test-conf test-name test-path itemdat params)
  (mutex-lock! *launch-setup-mutex*) ;; setting variables and processing the testconfig is NOT thread-safe, reuse the launch-setup mutex
  (let* ((item-path       (item-list->path itemdat)))
    (let loop ((delta        (- (current-seconds) *last-launch*))
	       (launch-delay (string->number (or (configf:lookup *configdat* "setup" "launch-delay") "5"))))
      (if (> launch-delay delta)
	  (begin
	    (debug:print-info 0 *default-log-port* "Delaying launch of " test-name " for " (- launch-delay delta) " seconds")
	    (thread-sleep! (- launch-delay delta))
	    (loop (- (current-seconds) *last-launch*) launch-delay))))

    (change-directory *toppath*)
    (alist->env-vars ;; consolidate this code with the code in megatest.scm for "-execute", *maybe* - the longer they are set the longer each launch takes (must be non-overlapping with the vars)
     (append
      (list
       (list "MT_RUN_AREA_HOME" *toppath*)
       (list "MT_TEST_NAME" test-name)

       (list "MT_RUNNAME"   runname)
       (list "MT_ITEMPATH"  item-path)
       )
      itemdat))
    (let* ((tregistry       (tests:get-all)) ;; third param (below) is system-allowed


           ;; for tconfig, why do we allow fallback to test-conf?
	   (tconfig         (or (tests:get-testconfig test-name tregistry #t force-create: #t)
				(begin
                                  (debug:print 0 *default-log-port* "WARNING: falling back to pre-calculated testconfig. This is likely not desired.")
                                  test-conf))) ;; force re-read now that all vars are set
	   (useshell        (let ((ush (config-lookup *configdat* "jobtools"     "useshell")))
			      (if ush 
				  (if (equal? ush "no") ;; must use "no" to NOT use shell
				      #f
				      ush)
				  #t)))     ;; default is yes
	   (runscript       (config-lookup tconfig   "setup"        "runscript"))
	   (ezsteps         (> (length (hash-table-ref/default tconfig "ezsteps" '())) 0)) ;; don't send all the steps, could be big
	   ;; (diskspace       (config-lookup tconfig   "requirements" "diskspace"))
	   ;; (memory          (config-lookup tconfig   "requirements" "memory"))
	   ;; (hosts           (config-lookup *configdat* "jobtools"     "workhosts")) ;; I'm pretty sure this was never completed
	   (remote-megatest (config-lookup *configdat* "setup" "executable"))
	   (run-time-limit  (or (configf:lookup  tconfig   "requirements" "runtimelim")
				(configf:lookup  *configdat* "setup" "runtimelim")))
	   ;; FIXME SOMEDAY: not good how this is so obtuse, this hack is to 
	   ;;                allow running from dashboard. Extract the path
	   ;;                from the called megatest and convert dashboard
	   ;;             	  or dboard to megatest
	   (local-megatest  (let* ((lm  (car (argv)))
				   (dir (pathname-directory lm))
				   (exe (pathname-strip-directory lm)))
			      (conc (if dir (conc dir "/") "")
				    (case (string->symbol exe)
				      ((dboard)    "../megatest")
				      ((mtest)     "../megatest")
				      ((dashboard) "megatest")
				      (else exe)))))
	   (launcher        (common:get-launcher *configdat* test-name item-path)) ;; (config-lookup *configdat* "jobtools"     "launcher"))
	   (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:roll-up-pass-fail-counts 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))
	  (begin
	    (set! work-area (conc test-path "/tmp_run"))
	    (create-directory work-area #t)
	    (debug:print 0 *default-log-port* "WARNING: No disk work area specified - running in the test directory under tmp_run")))
      (set! cmdparms (base64:base64-encode 
		      (z3:encode-buffer 
		       (with-output-to-string
			 (lambda () ;; (list 'hosts     hosts)
			   (write (list (list 'testpath  test-path)
					(list 'transport (conc *transport-type*))
					;; (list 'serverinf *server-info*)
					(list 'toppath   *toppath*)
					(list 'work-area work-area)
					(list 'test-name test-name) 
					(list 'runscript runscript) 
					(list 'run-id    run-id   )
					(list 'test-id   test-id  )
					;; (list 'item-path item-path )
					(list 'itemdat   itemdat  )
					(list 'megatest  remote-megatest)
					(list 'ezsteps   ezsteps) 
					(list 'target    mt_target)
					(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))))
       (launcher
	(set! fullcmd (append launcher (list remote-megatest "-m" test-sig "-execute" cmdparms) debug-param)))
       ;; (set! fullcmd (append launcher (list remote-megatest test-sig "-execute" cmdparms))))
       (else
	(if (not useshell)(debug:print 0 *default-log-port* "WARNING: internal launching will not work well without \"useshell yes\" in your [jobtools] section"))
	(set! fullcmd (append (list remote-megatest "-m" test-sig "-execute" cmdparms) debug-param (list (if useshell "&" ""))))))
      ;; (set! fullcmd (list remote-megatest test-sig "-execute" cmdparms (if useshell "&" "")))))
      (if (args:get-arg "-xterm")(set! fullcmd (append fullcmd (list "-xterm"))))
      (debug:print 1 *default-log-port* "Launching " work-area)
      ;; set pre-launch-env-vars before launching, keep the vars in prevvals and put the envionment back when done
      (debug:print 4 *default-log-port* "fullcmd: " fullcmd)
      (set! *last-launch* (current-seconds)) ;; all that junk above takes time, set this as late as possible.
      (let* ((commonprevvals (alist->env-vars
			      (hash-table-ref/default *configdat* "env-override" '())))


	     (miscprevvals   (alist->env-vars ;; consolidate this code with the code in megatest.scm for "-execute"
			      (append (list (list "MT_TEST_RUN_DIR" work-area)
					    (list "MT_TEST_NAME" test-name)
					    (list "MT_ITEM_INFO" (conc itemdat)) 
					    (list "MT_RUNNAME"   runname)
					    (list "MT_TARGET"    mt_target)
					    (list "MT_ITEMPATH"  item-path)
					    )
				      itemdat)))
	     (testprevvals   (alist->env-vars
			      (hash-table-ref/default tconfig "pre-launch-env-overrides" '())))
	     ;; Launchwait defaults to true, must override it to turn off wait
	     (launchwait     (if (equal? (configf:lookup *configdat* "setup" "launchwait") "no") #f #t))
	     (launch-results (apply (if launchwait
					process:cmd-run-with-stderr->list
					process-run)
				    (if useshell
					(let ((cmdstr (string-intersperse fullcmd " ")))
					  (if launchwait
					      cmdstr
					      (conc cmdstr " >> mt_launch.log 2>&1")))
					(car fullcmd))
				    (if useshell
					'()
					(cdr fullcmd)))))
        (mutex-unlock! *launch-setup-mutex*) ;; yes, really should mutex all the way to here. Need to put this entire process into a fork.
	(if (not launchwait) ;; give the OS a little time to allow the process to start
	    (thread-sleep! 0.01))
	(with-output-to-file "mt_launch.log"
	  (lambda ()
	    (print "LAUNCHCMD: " (string-intersperse fullcmd " "))
	    (if (list? launch-results)
		(apply print launch-results)
		(print "NOTE: launched \"" fullcmd "\"\n  but did not wait for it to proceed. Add the following to megatest.config \n[setup]\nlaunchwait yes\n  if you have problems with this"))
	    #:append))
	(debug:print 2 *default-log-port* "Launching completed, updating db")
	(debug:print 2 *default-log-port* "Launch results: " launch-results)
	(if (not launch-results)
	    (begin
	      (print "ERROR: Failed to run " (string-intersperse fullcmd " ") ", exiting now")
	      ;; (sqlite3:finalize! db)
	      ;; good ole "exit" seems not to work
	      ;; (_exit 9)
	      ;; but this hack will work! Thanks go to Alan Post of the Chicken email list
	      ;; NB// Is this still needed? Should be safe to go back to "exit" now?
	      (process-signal (current-process-id) signal/kill)
	      ))
	(alist->env-vars miscprevvals)
	(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