︙ | | | ︙ | |
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
|
(exit 4)))))
(thread-start! th2)
(thread-start! th1)
(thread-join! th2)))))
(set-signal-handler! signal/int sighand)
(set-signal-handler! signal/term sighand))
;; force the starting of a server
(debug:print 0 *default-log-port* "waiting on server...")
(server:start-and-wait *toppath*)
(runs:set-megatest-env-vars run-id inkeys: keys inrunname: runname) ;; these may be needed by the launching process
(set! runconf (if (common:file-exists? runconfigf)
(setup-env-defaults runconfigf run-id *already-seen-runconfig-info* keyvals target)
(begin
(debug:print 0 *default-log-port* "WARNING: You do not have a run config file: " runconfigf)
#f)))
|
|
|
|
|
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
|
(exit 4)))))
(thread-start! th2)
(thread-start! th1)
(thread-join! th2)))))
(set-signal-handler! signal/int sighand)
(set-signal-handler! signal/term sighand))
;; force the starting of a server -- removed BB 17ww28 - no longer needed.
;;(debug:print 0 *default-log-port* "waiting on server...")
;;(server:start-and-wait *toppath*)
(runs:set-megatest-env-vars run-id inkeys: keys inrunname: runname) ;; these may be needed by the launching process
(set! runconf (if (common:file-exists? runconfigf)
(setup-env-defaults runconfigf run-id *already-seen-runconfig-info* keyvals target)
(begin
(debug:print 0 *default-log-port* "WARNING: You do not have a run config file: " runconfigf)
#f)))
|
︙ | | | ︙ | |
593
594
595
596
597
598
599
600
601
602
603
604
605
606
|
(cdr reg)
(if (null? tal) ;; if tal is null and reg not full then '() as reg contents moved to tal
'()
reg)))
(define runs:nothing-left-in-queue-count 0)
(define (runs:expand-items hed tal reg reruns regfull newtal jobgroup max-concurrent-jobs run-id waitons item-path testmode test-record can-run-more items runname tconfig reglen test-registry test-records itemmaps)
(let* ((loop-list (list hed tal reg reruns))
(prereqs-not-met (let ((res (rmt:get-prereqs-not-met run-id waitons hed item-path mode: testmode itemmaps: itemmaps)))
(if (list? res)
res
(begin
(debug:print 0 *default-log-port*
|
>
|
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
|
(cdr reg)
(if (null? tal) ;; if tal is null and reg not full then '() as reg contents moved to tal
'()
reg)))
(define runs:nothing-left-in-queue-count 0)
;; BB: for future reference - suspect target vars are not expanded to env vars at this point (item expansion using [items]\nwhatever [system echo $TARGETVAR] doesnt work right whereas [system echo #{targetvar}] does.. Tal and Randy have tix on this. on first pass, var not set, on second pass, ok.
(define (runs:expand-items hed tal reg reruns regfull newtal jobgroup max-concurrent-jobs run-id waitons item-path testmode test-record can-run-more items runname tconfig reglen test-registry test-records itemmaps)
(let* ((loop-list (list hed tal reg reruns))
(prereqs-not-met (let ((res (rmt:get-prereqs-not-met run-id waitons hed item-path mode: testmode itemmaps: itemmaps)))
(if (list? res)
res
(begin
(debug:print 0 *default-log-port*
|
︙ | | | ︙ | |
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
|
testmode: testmode
newtal: newtal
itemmaps: itemmaps
;; prereqs-not-met: prereqs-not-met
)))
(runs:dat-regfull-set! runsdat regfull)
;; every 15 minutes verify the server is there for this run
(if (and (common:low-noise-print 240 "try start server" run-id)
(not (or (and *runremote*
(remote-server-url *runremote*)
(server:ping (remote-server-url *runremote*)))
(server:check-if-running *toppath*))))
(server:kind-run *toppath*))
(if (> num-running 0)
(set! last-time-some-running (current-seconds)))
(if (> (current-seconds)(+ last-time-some-running (or (configf:lookup *configdat* "setup" "give-up-waiting") 36000)))
(hash-table-set! *max-tries-hash* tfullname (+ (hash-table-ref/default *max-tries-hash* tfullname 0) 1)))
;; (debug:print 0 *default-log-port* "max-tries-hash: " (hash-table->alist *max-tries-hash*))
|
>
|
|
|
|
|
|
|
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
|
testmode: testmode
newtal: newtal
itemmaps: itemmaps
;; prereqs-not-met: prereqs-not-met
)))
(runs:dat-regfull-set! runsdat regfull)
;; -- removed BB 17ww28 - no longer needed.
;; every 15 minutes verify the server is there for this run
;; (if (and (common:low-noise-print 240 "try start server" run-id)
;; (not (or (and *runremote*
;; (remote-server-url *runremote*)
;; (server:ping (remote-server-url *runremote*)))
;; (server:check-if-running *toppath*))))
;; (server:kind-run *toppath*))
(if (> num-running 0)
(set! last-time-some-running (current-seconds)))
(if (> (current-seconds)(+ last-time-some-running (or (configf:lookup *configdat* "setup" "give-up-waiting") 36000)))
(hash-table-set! *max-tries-hash* tfullname (+ (hash-table-ref/default *max-tries-hash* tfullname 0) 1)))
;; (debug:print 0 *default-log-port* "max-tries-hash: " (hash-table->alist *max-tries-hash*))
|
︙ | | | ︙ | |
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
|
(if (null? tal)
#f
(loop (car tal)(cdr tal) reg reruns)))
;; if items is a proc then need to run items:get-items-from-config, get the list and loop
;; - but only do that if resources exist to kick off the job
;; EXPAND ITEMS
((or (procedure? items)(eq? items 'have-procedure))
(let ((can-run-more (runs:can-run-more-tests runsdat run-id jobgroup max-concurrent-jobs)))
(if (and (list? can-run-more)
(car can-run-more))
(let ((loop-list (runs:expand-items hed tal reg reruns regfull newtal jobgroup max-concurrent-jobs run-id waitons item-path testmode test-record can-run-more items runname tconfig reglen test-registry test-records itemmaps)))
(if loop-list
(apply loop loop-list)))
;; if can't run more just loop with next possible test
|
|
|
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
|
(if (null? tal)
#f
(loop (car tal)(cdr tal) reg reruns)))
;; if items is a proc then need to run items:get-items-from-config, get the list and loop
;; - but only do that if resources exist to kick off the job
;; EXPAND ITEMS
((or (procedure? items)(eq? items 'have-procedure)) ;; BB - target vars are env vars here? to allow expansion of [items]\nsomething [system echo $SOMETARGVAR], which is wonky
(let ((can-run-more (runs:can-run-more-tests runsdat run-id jobgroup max-concurrent-jobs)))
(if (and (list? can-run-more)
(car can-run-more))
(let ((loop-list (runs:expand-items hed tal reg reruns regfull newtal jobgroup max-concurrent-jobs run-id waitons item-path testmode test-record can-run-more items runname tconfig reglen test-registry test-records itemmaps)))
(if loop-list
(apply loop loop-list)))
;; if can't run more just loop with next possible test
|
︙ | | | ︙ | |
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
|
(define (runs:remove-test-directory test mode) ;; remove-data-only)
(let* ((run-dir (db:test-get-rundir test)) ;; run dir is from the link tree
(real-dir (if (common:file-exists? run-dir)
;; (resolve-pathname run-dir)
(common:nice-path run-dir)
#f))
(clean-mode (or mode 'remove-all)))
(case clean-mode
((remove-data-only)(mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "CLEANING" "LOCKED" #f))
((remove-all) (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "REMOVING" "LOCKED" #f))
((archive-remove) (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "ARCHIVE_REMOVING" #f #f)))
(debug:print-info 1 *default-log-port* "Attempting to remove " (if real-dir (conc " dir " real-dir " and ") "") " link " run-dir)
(if (and real-dir
(> (string-length real-dir) 5)
|
|
>
>
>
>
>
>
>
>
>
>
>
>
>
>
>
|
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
|
(define (runs:remove-test-directory test mode) ;; remove-data-only)
(let* ((run-dir (db:test-get-rundir test)) ;; run dir is from the link tree
(real-dir (if (common:file-exists? run-dir)
;; (resolve-pathname run-dir)
(common:nice-path run-dir)
#f))
(clean-mode (or mode 'remove-all))
(test-id (db:test-get-id test))
;; (lock-key (conc "test-" test-id))
;; (got-lock (let loop ((lock (rmt:no-sync-get-lock lock-key))
;; (expire-time (+ (current-seconds) 30))) ;; give up on getting the lock and steal it after 15 seconds
;; (if (car lock)
;; #t
;; (if (> (current-seconds) expire-time)
;; (begin
;; (debug:print-info 0 *default-log-port* "Timed out waiting for a lock to clean test with id " test-id)
;; (rmt:no-sync-del! lock-key) ;; destroy the lock
;; (loop (rmt:no-sync-get-lock lock-key) expire-time)) ;;
;; (begin
;; (thread-sleep! 1)
;; (loop (rmt:no-sync-get-lock lock-key) expire-time)))))))
)
(case clean-mode
((remove-data-only)(mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "CLEANING" "LOCKED" #f))
((remove-all) (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "REMOVING" "LOCKED" #f))
((archive-remove) (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "ARCHIVE_REMOVING" #f #f)))
(debug:print-info 1 *default-log-port* "Attempting to remove " (if real-dir (conc " dir " real-dir " and ") "") " link " run-dir)
(if (and real-dir
(> (string-length real-dir) 5)
|
︙ | | | ︙ | |
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
|
(debug:print 0 *default-log-port* "WARNING: not removing " run-dir " as it either doesn't exist or is not a symlink")
(debug:print 0 *default-log-port* "NOTE: the run dir for this test is undefined. Test may have already been deleted."))
))
;; Only delete the records *after* removing the directory. If things fail we have a record
(case clean-mode
((remove-data-only)(mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) (db:test-get-state test)(db:test-get-status test) #f))
((archive-remove) (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "ARCHIVED" #f #f))
(else (rmt:delete-test-records (db:test-get-run_id test) (db:test-get-id test))))))
;;======================================================================
;; Routines for manipulating runs
;;======================================================================
;; Since many calls to a run require pretty much the same setup
;; this wrapper is used to reduce the replication of code
|
|
>
>
|
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
|
(debug:print 0 *default-log-port* "WARNING: not removing " run-dir " as it either doesn't exist or is not a symlink")
(debug:print 0 *default-log-port* "NOTE: the run dir for this test is undefined. Test may have already been deleted."))
))
;; Only delete the records *after* removing the directory. If things fail we have a record
(case clean-mode
((remove-data-only)(mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) (db:test-get-state test)(db:test-get-status test) #f))
((archive-remove) (mt:test-set-state-status-by-id (db:test-get-run_id test)(db:test-get-id test) "ARCHIVED" #f #f))
(else (rmt:delete-test-records (db:test-get-run_id test) (db:test-get-id test))))
;; (rmt:no-sync-del! lock-key)
))
;;======================================================================
;; Routines for manipulating runs
;;======================================================================
;; Since many calls to a run require pretty much the same setup
;; this wrapper is used to reduce the replication of code
|
︙ | | | ︙ | |