;; Copyright 2006-2013, Matthew Welland.
;;
;; This program is made available under the GNU GPL version 2.0 or
;; greater. See the accompanying file COPYING for details.
;;
;; This program is distributed WITHOUT ANY WARRANTY; without even the
;; implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR
;; PURPOSE.
;;======================================================================
;; launch a task - this runs on the originating host, tests themselves
;;
;;======================================================================
(use regex regex-case base64 sqlite3 srfi-18 directory-utils posix-extras z3 call-with-environment-variables)
(import (prefix base64 base64:))
(import (prefix sqlite3 sqlite3:))
(declare (unit launch))
(declare (uses common))
(declare (uses configf))
(declare (uses db))
;; (declare (uses sdb))
(declare (uses tdb))
;; (declare (uses filedb))
(include "common_records.scm")
(include "key_records.scm")
(include "db_records.scm")
;;======================================================================
;; ezsteps
;;======================================================================
;; ezsteps were going to be coded as
;; stepname[,predstep1,predstep2 ...] [{VAR1=first,second,third}] command to execute
;; BUT
;; now are
;; stepname {VAR=first,second,third ...} command ...
;; where the {VAR=first,second,third ...} is optional.
;; given an exit code and whether or not logpro was used calculate OK/BAD
;; return #t if we are ok, #f otherwise
(define (steprun-good? logpro exitcode)
(or (eq? exitcode 0)
(and logpro (eq? exitcode 2))))
;; if handed a string, process it, else look for MT_CMDINFO
(define (launch:get-cmdinfo-assoc-list #!key (encoded-cmd #f))
(let ((enccmd (if encoded-cmd encoded-cmd (getenv "MT_CMDINFO"))))
(if enccmd
(common:read-encoded-string enccmd)
'())))
(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"))
(html-file (conc stepname ".html"))
(tconfig-logpro (configf:lookup testconfig "logpro" stepname))
(logpro-used (file-exists? logpro-file)))
(if (and tconfig-logpro
(not logpro-used)) ;; no logpro file found but have a defn in the testconfig
(begin
(with-output-to-file logpro-file
(lambda ()
(print ";; logpro file extracted from testconfig\n"
";;")
(print tconfig-logpro)))
(set! logpro-used #t)))
;; NB// can safely assume we are in test-area directory
(debug:print 4 "ezsteps:\n stepname: " stepname " stepinfo: " stepinfo " stepparts: " stepparts
" stepparms: " stepparms " stepcmd: " stepcmd)
;; ;; first source the previous environment
;; (let ((prev-env (conc ".ezsteps/" prevstep (if (string-search (regexp "csh")
;; (get-environment-variable "SHELL")) ".csh" ".sh"))))
;; (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 "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))))
(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)
(vector-set! exit-info 0 pid)
(vector-set! exit-info 1 exit-status)
(vector-set! exit-info 2 exit-code)
(mutex-unlock! m)
(if (eq? pid-val 0)
(begin
(thread-sleep! 2)
(processloop (+ i 1))))
)))))
(debug:print-info 0 "step " stepname " completed with exit code " (vector-ref exit-info 2))
;; now run logpro if needed
(if logpro-used
(let ((pid (process-run (conc "logpro " logpro-file " " (conc stepname ".html") " < " stepname ".log"))))
(let processloop ((i 0))
(let-values (((pid-val exit-status exit-code)(process-wait pid #t)))
(mutex-lock! m)
(vector-set! exit-info 0 pid)
(vector-set! exit-info 1 exit-status)
(vector-set! exit-info 2 exit-code)
(mutex-unlock! m)
(if (eq? pid-val 0)
(begin
(thread-sleep! 2)
(processloop (+ i 1)))))
(debug:print-info 0 "logpro for step " stepname " exited with code " (vector-ref exit-info 2)))))
(let ((exinfo (vector-ref exit-info 2))
(logfna (if logpro-used (conc stepname ".html") "")))
(rmt:teststep-set-status! run-id test-id stepname "end" exinfo #f logfna))
(if logpro-used
(rmt:test-set-log! run-id test-id (conc stepname ".html")))
;; set the test final status
(let* ((this-step-status (cond
((and (eq? (vector-ref exit-info 2) 2) logpro-used) 'warn)
((eq? (vector-ref exit-info 2) 0) 'pass)
(else 'fail)))
(overall-status (cond
((eq? (vector-ref exit-info 3) 2) 'warn) ;; rollup-status
((eq? (vector-ref exit-info 3) 0) 'pass)
(else 'fail)))
(next-status (cond
((eq? overall-status 'pass) this-step-status)
((eq? overall-status 'warn)
(if (eq? this-step-status 'fail) 'fail 'warn))
(else 'fail)))
(next-state ;; "RUNNING") ;; WHY WAS THIS CHANGED TO NOT USE (null? tal) ??
(cond
((null? tal) ;; more to run?
"COMPLETED")
(else "RUNNING")))
)
(debug:print 4 "Exit value received: " (vector-ref exit-info 2) " logpro-used: " logpro-used
" this-step-status: " this-step-status " overall-status: " overall-status
" next-status: " next-status " rollup-status: " (vector-ref exit-info 3))
(case next-status
((warn)
(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))
((pass)
(tests:test-set-status! run-id test-id next-state "PASS" #f #f))
(else ;; 'fail
(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:execute encoded-cmd)
(let* ((cmdinfo (common:read-encoded-string encoded-cmd))
(tconfigreg (make-hash-table)))
(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))
(let* ((testpath (assoc/default 'testpath cmdinfo)) ;; testpath is the test spec area
(top-path (assoc/default 'toppath cmdinfo))
(work-area (assoc/default 'work-area cmdinfo)) ;; work-area is the test run area
(test-name (assoc/default 'test-name cmdinfo))
(runscript (assoc/default 'runscript cmdinfo))
(ezsteps (assoc/default 'ezsteps cmdinfo))
;; (runremote (assoc/default 'runremote cmdinfo))
(transport (assoc/default 'transport cmdinfo))
;; (serverinf (assoc/default 'serverinf cmdinfo))
(port (assoc/default 'port cmdinfo))
(run-id (assoc/default 'run-id cmdinfo))
(test-id (assoc/default 'test-id cmdinfo))
(target (assoc/default 'target cmdinfo))
(itemdat (assoc/default 'itemdat cmdinfo))
(env-ovrd (assoc/default 'env-ovrd cmdinfo))
(set-vars (assoc/default 'set-vars cmdinfo)) ;; pre-overrides from -setvar
(runname (assoc/default 'runname cmdinfo))
(megatest (assoc/default 'megatest cmdinfo))
(runtlim (assoc/default 'runtlim cmdinfo))
(item-path (item-list->path itemdat))
(mt-bindir-path (assoc/default 'mt-bindir-path cmdinfo))
(keys #f)
(keyvals #f)
(fullrunscript (if (not runscript)
#f
(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)
)
(change-directory top-path)
(let ((sighand (lambda (signum)
;; (signal-mask! signum) ;; to mask or not? seems to cause issues in exiting
(if (eq? signum signal/stop)
(debug:print 0 "ERROR: 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 ()
(tests:test-force-state-status! run-id test-id "INCOMPLETE" "KILLED")
(print "Killed by signal " signum ". Exiting")
(thread-sleep! 1)
(exit 1))))
(th2 (make-thread (lambda ()
(thread-sleep! 2)
(debug:print 0 "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))
;; (set-signal-handler! signal/int (lambda ()
;; Do not run the test if it is REMOVING, RUNNING, KILLREQ or REMOTEHOSTSTART,
;; Mark the test as REMOTEHOSTSTART *IMMEDIATELY*
;;
(let ((test-info (rmt:get-testinfo-state-status run-id test-id)))
(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 "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")) ;; prime it for running
((not (member (db:test-get-state test-info) '("REMOVING" "REMOTEHOSTSTART" "RUNNING" "KILLREQ")))
(tests:test-force-state-status! run-id test-id "REMOTEHOSTSTART" "n/a"))
(else ;; (member (db:test-get-state test-info) '("REMOVING" "REMOTEHOSTSTART" "RUNNING" "KILLREQ"))
(debug:print 0 "ERROR: test state is " (db:test-get-state test-info) ", cannot proceed")
(exit))))
(debug:print 2 "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-for-run force: #t))
(begin
(debug:print 0 "Failed to setup, exiting")
;; (sqlite3:finalize! db)
;; (sqlite3:finalize! tdb)
(exit 1)))
(change-directory *toppath*)
;; NOTE: Current order is to process runconfigs *before* setting the MT_ vars. This
;; seems non-ideal but could well break stuff
;; BUG? BUG? BUG?
(let ((rconfig (full-runconfigs-read))) ;; (read-config (conc *toppath* "/runconfigs.config") #f #t sections: (list "default" target))))
;; (setup-env-defaults (conc *toppath* "/runconfigs.config") run-id (make-hash-table) keyvals target)
;; (set-run-config-vars run-id keyvals target) ;; (db:get-target db run-id))
;; Now have runconfigs data loaded, set environment vars
(for-each (lambda (section)
(for-each (lambda (varval)
(let ((var (car varval))
(val (cadr varval)))
(if (and (string? var)(string? val))
(begin
(setenv var (config:eval-string-in-environment val))) ;; val)
(debug:print 0 "ERROR: bad variable spec, " var "=" val))))
(configf:get-section rconfig section)))
(list "default" target)))
(change-directory work-area)
(set! keyvals (keys:target->keyval keys target))
;; apply pre-overrides before other variables. The pre-override vars must not
;; clobbers things from the official sources such as megatest.config and runconfigs.config
(if (string? set-vars)
(let ((varpairs (string-split set-vars ",")))
(debug:print 4 "varpairs: " varpairs)
(map (lambda (varpair)
(let ((varval (string-split varpair "=")))
(if (eq? (length varval) 2)
(let ((var (car varval))
(val (cadr varval)))
(debug:print 1 "Adding pre-var/val " var " = " val " to the environment")
(setenv var val)))))
varpairs)))
(for-each
(lambda (varval)
(let ((var (car varval))
(val (cadr varval)))
(if val
(setenv var val)
(begin
(debug:print 0 "ERROR: required variable " var " does not have a valid value. Exiting")
(exit)))))
(list
(list "MT_TEST_RUN_DIR" work-area)
(list "MT_TEST_NAME" test-name)
(list "MT_ITEM_INFO" (conc itemdat))
(list "MT_ITEMPATH" item-path)
(list "MT_RUNNAME" runname)
(list "MT_MEGATEST" megatest)
(list "MT_TARGET" target)
(list "MT_LINKTREE" (configf:lookup *configdat* "setup" "linktree"))
(list "MT_TESTSUITENAME" (common:get-testsuite-name))))
(if mt-bindir-path (setenv "PATH" (conc (getenv "PATH") ":" mt-bindir-path)))
;; (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 (not (file-execute-access? fullrunscript)))
(system (conc "chmod ug+x " fullrunscript))))
;; We are about to actually kick off the test
;; so this is a good place to remove the records for
;; any previous runs
;; (db:test-remove-steps db run-id testname itemdat)
(let* ((m (make-mutex))
(kill-job? #f)
(exit-info (vector #t #t #t 0))
(job-thread #f)
(keep-going #t)
(runit (lambda ()
;; (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:roll-up-pass-fail-counts run-id test-name item-path #f "RUNNING")
;; (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)
(vector-set! exit-info 0 pid)
(vector-set! exit-info 1 exit-status)
(vector-set! exit-info 2 exit-code)
(vector-set! exit-info 3 exit-code) ;; rollup status
(mutex-unlock! m)
(if (eq? pid-val 0)
(begin
(thread-sleep! 2)
(loop (+ i 1)))
)))))
;; then, if runscript ran ok (or did not get called)
;; do all the ezsteps (if any)
(if ezsteps
(let* ((testconfig ;; (read-config (conc work-area "/testconfig") #f #t environ-patt: "pre-launch-env-vars")) ;; FIXME??? is allow-system ok here?
(tests:get-testconfig test-name tconfigreg #t)) ;; 'return-procs)))
(ezstepslst (hash-table-ref/default testconfig "ezsteps" '())))
(hash-table-set! *testconfigs* test-name testconfig) ;; cached for lazy reads later ...
(if (not (file-exists? ".ezsteps"))(create-directory ".ezsteps"))
;; if ezsteps was defined then we are sure to have at least one step but check anyway
(if (not (> (length ezstepslst) 0))
(debug:print 0 "ERROR: ezsteps defined but ezstepslst is zero length")
(let loop ((ezstep (car ezstepslst))
(tal (cdr ezstepslst))
(prevstep #f))
;; check exit-info (vector-ref exit-info 1)
(if (vector-ref exit-info 1)
(let ((logpro-used (launch:runstep ezstep run-id test-id exit-info m tal testconfig)))
(if (and (steprun-good? logpro-used (vector-ref exit-info 2))
(not (null? tal)))
(loop (car tal) (cdr tal) stepname)))
(debug:print 4 "WARNING: a prior step failed, stopping at " ezstep))))))))
(monitorjob (lambda ()
(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)))
(begin
(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 "KILLING TEST DUE TO TIME LIMIT EXCEEDED! Runtime=" run-seconds " seconds, limit=" runtlim)
#t)
#f)))))
(tests:update-central-meta-info run-id test-id (get-cpu-load) (get-df (current-directory))(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 (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
(begin
(debug:print-info 0 "Unable to kill process with pid " pid ", possibly already killed.")
(debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn)))
(debug:print 0 "WARNING: Request received to kill job " pid) ;; " (attempt # " kill-tries ")")
(debug:print-info 0 "Signal mask=" (signal-mask))
;; (if (process:alive? pid)
;; (begin
(map (lambda (pid-num)
(process-signal pid-num signal/term))
(process:get-sub-pids pid))
(thread-sleep! 5)
;; (if (process:process-alive? pid)
(map (lambda (pid-num)
(handle-exceptions
exn
#f
(process-signal pid-num signal/kill)))
(process:get-sub-pids pid))))
;; (debug:print-info 0 "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 0 "ERROR: 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 keep-going
(begin
(thread-sleep! 3) ;; (+ 3 (random 6))) ;; add some jitter to the call home time to spread out the db accesses
(if keep-going
(loop (calc-minutes)))))))
(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
(th1 (make-thread monitorjob "monitor job"))
(th2 (make-thread runit "run job")))
(set! job-thread th2)
(thread-start! th1)
(thread-start! th2)
(thread-join! th2)
(debug:print-info 0 "Megatest exectute of test " test-name ", item path " item-path " complete. Notifying the db ...")
(set! keep-going #f)
(thread-join! th1)
(thread-sleep! 1) ;; give 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
((not (vector-ref exit-info 1)) "FAIL") ;; job failed to run
((eq? (vector-ref exit-info 3) 0)
;; if the current status is AUTO then defer to the calculated value (i.e. leave this AUTO)
(if (equal? (db:test-get-status testinfo) "AUTO") "AUTO" "PASS"))
((eq? (vector-ref exit-info 3) 1) "FAIL")
((eq? (vector-ref exit-info 3) 2)
;; if the current status is AUTO the defer to the calculated value but qualify (i.e. make this AUTO-WARN)
(if (equal? (db:test-get-status testinfo) "AUTO") "AUTO-WARN" "WARN"))
(else "FAIL")))) ;; (db:test-get-status testinfo)))
(debug:print-info 1 "Test exited in state=" (db:test-get-state testinfo) ", setting state/status based on exit code of " (vector-ref exit-info 1) " and rollup-status of " (vector-ref exit-info 3))
(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 roll-up-pass-fail-counts HERE, THIS IS DONE IN roll-up-pass-fail-counts 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
)
(mutex-unlock! m)
(debug:print 2 "Output from running " fullrunscript ", pid " (vector-ref exit-info 0) " in work area "
work-area ":\n====\n exit code " (vector-ref exit-info 2) "\n" "====\n")
(if (not (vector-ref exit-info 1))
(exit 4)))))))
;; set up the very basics needed for doing anything here.
(define (launch:setup-for-run #!key (force #f))
;; would set values for KEYS in the environment here for better support of env-override but
;; have chicken/egg scenario. need to read megatest.config then read it again. Going to
;; pass on that idea for now
;; special case
(if (or force (not (hash-table? *configdat*))) ;; no need to re-open on every call
(begin
(set! *configinfo* (or (if (get-environment-variable "MT_CMDINFO") ;; we are inside a test - do not reprocess configs
(let ((alistconfig (conc (get-environment-variable "MT_LINKTREE") "/"
(get-environment-variable "MT_TARGET") "/"
(get-environment-variable "MT_RUNNAME") "/"
".megatest.cfg-" megatest-version "-" megatest-fossil-hash)))
(if (file-exists? alistconfig)
(list (configf:read-alist alistconfig)
(get-environment-variable "MT_RUN_AREA_HOME"))
#f))
#f) ;; no config cached - give up
(let ((runname (or (args:get-arg "-runname")(args:get-arg ":runname"))))
(if runname (setenv "MT_RUNNAME" runname))
(find-and-read-config
(if (args:get-arg "-config")(args:get-arg "-config") "megatest.config")
environ-patt: "env-override"
given-toppath: (get-environment-variable "MT_RUN_AREA_HOME")
pathenvvar: "MT_RUN_AREA_HOME"))))
(set! *configdat* (if (car *configinfo*)(car *configinfo*) #f))
(set! *toppath* (if (car *configinfo*)(cadr *configinfo*) #f))
(let* ((tmptransport (configf:lookup *configdat* "server" "transport"))
(transport (if tmptransport (string->symbol tmptransport) 'http)))
(if (member transport '(http rpc nmsg))
(set! *transport-type* transport)
(begin
(debug:print 0 "ERROR: Unrecognised transport " transport)
(exit))))
(let ((linktree (configf:lookup *configdat* "setup" "linktree"))) ;; link tree is critical
(if linktree
(if (not (file-exists? linktree))
(begin
(handle-exceptions
exn
(begin
(debug:print 0 "ERROR: Something went wrong when trying to create linktree dir at " linktree)
(debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn))
(exit 1))
(create-directory linktree #t))))
(begin
(debug:print 0 "ERROR: linktree not defined in [setup] section of megatest.config")
(exit 1)))
(if linktree
(let ((dbdir (conc linktree "/.db")))
(handle-exceptions
exn
(begin
(debug:print 0 "ERROR: failed to create the " dbdir " area for your database files")
(debug:print 0 " message: " ((condition-property-accessor 'exn 'message) exn)))
(if (not (directory-exists? dbdir))(create-directory dbdir)))
(setenv "MT_LINKTREE" linktree))
(begin
(debug:print 0 "ERROR: linktree is required in your megatest.config [setup] section")
(exit 1)))
(if (and *toppath*
(directory-exists? *toppath*))
(setenv "MT_RUN_AREA_HOME" *toppath*)
(begin
(debug:print 0 "ERROR: failed to find the top path to your Megatest area.")
(exit 1)))
)))
*toppath*)
(define (launch:cache-config)
;; if we have a linktree and -runtests and -target and the directory exists dump the config
;; to megatest-(current-seconds).cfg and symlink it to megatest.cfg
(if (and *configdat*
(or (args:get-arg "-run")
(args:get-arg "-runtests")))
(let* ((linktree (get-environment-variable "MT_LINKTREE"))
(target (common:args-get-target))
(runname (or (args:get-arg "-runname")
(args:get-arg ":runname")))
(fulldir (conc linktree "/"
target "/"
runname)))
(debug:print-info 0 "Have -runtests with target=" target ", runname=" runname ", fulldir=" fulldir)
(if (file-exists? linktree) ;; can't proceed without linktree
(begin
(if (not (file-exists? fulldir))
(create-directory fulldir #t)) ;; need to protect with exception handler
(if (and target
runname
(file-exists? fulldir))
(let ((tmpfile (conc fulldir "/.megatest.cfg." (current-seconds)))
(targfile (conc fulldir "/.megatest.cfg-" megatest-version "-" megatest-fossil-hash)))
(debug:print-info 0 "Caching megatest.config in " fulldir "/.megatest.cfg")
(configf:write-alist *configdat* tmpfile)
(system (conc "ln -sf " tmpfile " " targfile))
)))))))
(define (get-best-disk confdat testconfig)
(let* ((disks (or (and testconfig (hash-table-ref/default testconfig "disks" #f))
(hash-table-ref/default confdat "disks" #f)))
(minspace (let ((m (configf:lookup confdat "setup" "minspace")))
(string->number (or m "10000")))))
(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")
(debug:print 0 "ERROR: No valid disks found in megatest.config. Please add some to your [disks] section and ensure the directory exists!"))
(exit 1)))))))
;; Desired directory structure:
;;
;; <linkdir> - <target> - <testname> -.
;; |
;; v
;; <rundir> - <target> - <testname> -|- <itempath(s)>
;;
;; dir stored in test is:
;;
;; <linkdir> - <target> - <testname> [ - <itempath> ]
;;
;; All log file links should be stored relative to the top of link path
;;
;; <target> - <testname> [ - <itempath> ]
;;
(define (create-work-area run-id run-info keyvals test-id test-src-path disk-path testname itemdat #!key (remtries 2))
(let* ((item-path (if (string? itemdat) itemdat (item-list->path itemdat))) ;; if pass in string - just use it
(runname (if (string? run-info) ;; if we pass in a string as run-info use it as run-name.
run-info
(db:get-value-by-header (db:get-rows run-info)
(db:get-header run-info)
"runname")))
;; convert back to db: from rdb: - this is always run at server end
(target (string-intersperse (map cadr keyvals) "/"))
(not-iterated (equal? "" item-path))
;; all tests are found at <rundir>/test-base or <linkdir>/test-base
(testtop-base (conc target "/" runname "/" testname))
(test-base (conc testtop-base (if not-iterated "" "/") item-path))
;; nb// if itempath is not "" then it is prefixed with "/"
(toptest-path (conc disk-path "/" testtop-base))
(test-path (conc disk-path "/" test-base))
;; ensure this exists first as links to subtests must be created there
(linktree (let ((rd (config-lookup *configdat* "setup" "linktree")))
(if rd rd (conc *toppath* "/runs"))))
(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)
(debug:print 2 "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 "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...
(if (and (not (directory-exists? lnkbase))
(not (file-exists? lnkbase)))
(handle-exceptions
exn
(begin
(debug:print "ERROR: Problem creating linktree base at " lnkbase)
(print-error-message exn (current-error-port)))
(create-directory lnkbase #t)))
;; update the toptest record with its location rundir, cache the path
;; This wass highly inefficient, one db write for every subtest, potentially
;; thousands of unnecessary updates, cache the fact it was set and don't set it
;; again.
;; Now create the link from the test path to the link tree, however
;; if the test is iterated it is necessary to create the parent path
;; to the iteration. use pathname-directory to trim the path by one
;; level
(if (not not-iterated) ;; i.e. iterated
(let ((iterated-parent (pathname-directory (conc lnkpath "/" item-path))))
(debug:print-info 2 "Creating iterated parent " iterated-parent)
(handle-exceptions
exn
(begin
(debug:print 0 "ERROR: Failed to create directory " iterated-parent ((condition-property-accessor 'exn 'message) exn) ", exiting")
(exit 1))
(create-directory iterated-parent #t))))
(if (symbolic-link? lnkpath)
(handle-exceptions
exn
(begin
(debug:print 0 "ERROR: Failed to remove symlink " lnkpath ((condition-property-accessor 'exn 'message) exn) ", exiting")
(exit 1))
(delete-file lnkpath)))
(if (not (or (file-exists? lnkpath)
(symbolic-link? lnkpath)))
(handle-exceptions
exn
(begin
(debug:print 0 "ERROR: Failed to create symlink " lnkpath ((condition-property-accessor 'exn 'message) exn) ", exiting")
(exit 1))
(create-symbolic-link toptest-path lnkpath)))
;; 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)
lnkpath)
testname "")
;; (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 "Creating " toptest-path " and link " lnkpath)
(handle-exceptions
exn
#f ;; don't care to catch and deal with errors here for now.
(create-directory toptest-path #t))
(hash-table-set! *toptest-paths* testname toptest-path)))))
;; The toptest path has been created, the link to the test in the linktree has
;; been created. Now, if this is an iterated test the real test dir must be created
(if (not not-iterated) ;; this is an iterated test
(begin ;; (let ((lnktarget (conc lnkpath "/" item-path)))
(debug:print 2 "Setting up sub test run area")
(debug:print 2 " - creating run area in " test-path)
(handle-exceptions
exn
(begin
(debug:print 0 "ERROR: Failed to create directory " test-path ((condition-property-accessor 'exn 'message) exn) ", exiting")
(exit 1))
(create-directory test-path #t))
(debug:print 2
" - creating link from: " test-path "\n"
" to: " lnktarget)
;; If there is already a symlink delete it and recreate it.
(handle-exceptions
exn
(begin
(debug:print 0 "ERROR: Failed to re-create link " lnktarget ((condition-property-accessor 'exn 'message) exn) ", exiting")
(exit))
(if (symbolic-link? lnktarget) (delete-file lnktarget))
(if (not (file-exists? lnktarget)) (create-symbolic-link test-path lnktarget)))))
(if (not (directory? test-path))
(create-directory test-path #t)) ;; this is a hack, I don't know why out of the blue this path does not exist sometimes
(if (and test-src-path (directory? test-path))
(begin
(let* ((ovrcmd (let ((cmd (config-lookup *configdat* "setup" "testcopycmd")))
(if cmd
;; substitute the TEST_SRC_PATH and TEST_TARG_PATH
(string-substitute "TEST_TARG_PATH" test-path
(string-substitute "TEST_SRC_PATH" test-src-path cmd #t) #t)
#f)))
(cmd (if ovrcmd
ovrcmd
(conc "rsync -av" (if (debug:debug-mode 1) "" "q") " " test-src-path "/ " test-path "/"
" >> " test-path "/mt_launch.log 2>> " test-path "/mt_launch.log")))
(status (system cmd)))
(if (not (eq? status 0))
(debug:print 2 "ERROR: problem with running \"" cmd "\"")))
(list lnkpathf lnkpath ))
(if (and test-src-path (> remtries 0))
(begin
(debug:print 0 "ERROR: Failed to create work area at " test-path " with link at " lnktarget ", remaining attempts " remtries)
;;
(create-work-area run-id run-info keyvals test-id test-src-path disk-path testname itemdat remtries: (- remtries 1)))
(list #f #f)))))
;; 1. look though disks list for disk with most space
;; 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)
(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 (make-hash-table))
(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)
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 "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
;;
(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")
(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 "Using work area " work-area))
(begin
(set! work-area (conc test-path "/tmp_run"))
(create-directory work-area #t)
(debug:print 0 "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 "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 "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 "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
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 ()
(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 "Launching completed, updating db")
(debug:print 2 "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*))