Megatest

(define *wdnum* 0)
(define *wdnum*mutex (make-mutex))
;; currently the primary job of the watchdog is to run the sync back to megatest.db from the db in /tmp
;; if we are on the homehost and we are a server (by definition we are on the homehost if we are a server)
;;
(define (common:watchdog)

  (thread-sleep! 0.05) ;; delay for startup
  (let ((legacy-sync (common:run-sync?))
	(debug-mode  (debug:debug-mode 1))
	(last-time   (current-seconds))
        (this-wd-num     (begin (mutex-lock! *wdnum*mutex) (let ((x *wdnum*)) (set! *wdnum* (add1 *wdnum*)) (mutex-unlock! *wdnum*mutex) x))))

    (debug:print-info 0 *default-log-port* "watchdog starting. legacy-sync is " legacy-sync" pid="(current-process-id)" this-wd-num="this-wd-num)
    (if (and legacy-sync (not *time-to-exit*))
	(let* ((dbstruct (db:setup))
	       (mtdb     (dbr:dbstruct-mtdb dbstruct))
	       (mtpath   (db:dbdat-get-path mtdb)))
	  (debug:print-info 0 *default-log-port* "Server running, periodic sync started.")
	  (let loop ()

	    ;; sync for filesystem local db writes
	    ;;
	    (mutex-lock! *db-multi-sync-mutex*)
	    (let* ((need-sync        (>= *db-last-access* *db-last-sync*)) ;; no sync since last write
		   (sync-in-progress *db-sync-in-progress*)
		   (should-sync      (and (not *time-to-exit*)
                                          (> (- (current-seconds) *db-last-sync*) 5))) ;; sync every five seconds minimum
		   (start-time       (current-seconds))
		   (mt-mod-time      (file-modification-time mtpath))
		   (recently-synced  (> (- start-time mt-mod-time) 4))
		   (will-sync        (and (or need-sync should-sync)
					  (not sync-in-progress)
					  (not recently-synced))))
	      ;; (if recently-synced (debug:print-info 0 *default-log-port* "Skipping sync due to recently-synced flag=" recently-synced))
	      ;; (debug:print-info 0 *default-log-port* "need-sync: " need-sync " sync-in-progress: " sync-in-progress " should-sync: " should-sync " will-sync: " will-sync)
	      (if will-sync (set! *db-sync-in-progress* #t))
	      (mutex-unlock! *db-multi-sync-mutex*)
	      (if will-sync
		  (let ((res (common:sync-to-megatest.db dbstruct))) ;; did we sync any data? If so need to set the db touched flag to keep the server alive
		    (if (> res 0) ;; some records were transferred, keep the db alive
			(begin

;;======================================================================
;; Database access
;;======================================================================

;; dbstruct vector containing all the relevant dbs like main.db, megatest.db, run.db etc

(use (srfi 18) extras tcp stack)
(use sqlite3 srfi-1 posix regex regex-case srfi-69 csv-xml s11n md5 message-digest base64 format dot-locking z3 typed-records matchable)
(import (prefix sqlite3 sqlite3:))
(import (prefix base64 base64:))

(declare (unit db))
(declare (uses common))
(declare (uses keys))
(declare (uses ods))
(declare (uses client))
(declare (uses mt))

	(lambda (run-id)
	  (set! run-ids (cons run-id run-ids)))
	db
	"SELECT id FROM runs WHERE state != 'deleted' ORDER BY event_time DESC;")
    (reverse run-ids)))))

;; get some basic run stats
;;
;; data structure:
;;
;; ( (runname (( state  count ) ... ))
;;   (   ...
;;
(define (db:get-run-stats dbstruct)
  (let* ((totals       (make-hash-table))
	 (curr         (make-hash-table))
	 (res          '())
	 (runs-info    '()))
    ;; First get all the runname/run-ids
    (db:with-db

	       (start-time     (current-milliseconds)))
      ;; Use this opportunity to sync the tmp db to megatest.db
      (if (not server-going) ;; *dbstruct-db* 
	  (begin
	    (debug:print 0 *default-log-port* "SERVER: dbprep")
	    (set! *dbstruct-db*  (db:setup)) ;;  run-id))
	    (set! server-going #t)
	    (debug:print 0 *default-log-port* "SERVER: running, megatest version: " (common:get-full-version)) ;; NOTE: the server is NOT yet marked as running in the log. We do that in the keep-running routine.
	    (thread-start! *watchdog*)))
      
      ;; when things go wrong we don't want to be doing the various queries too often
      ;; so we strive to run this stuff only every four seconds or so.
      (let* ((sync-time (- (current-milliseconds) start-time))
	    (rem-time  (quotient (- 4000 sync-time) 1000)))
	(if (and (<= rem-time 4)

	     (adjusted-timeout (if (> hrs-since-start 1)
				   (- server-timeout (inexact->exact (round (* hrs-since-start 60))))  ;; subtract 60 seconds per hour
				   server-timeout)))
	(if (common:low-noise-print 120 "server timeout")
	    (debug:print-info 0 *default-log-port* "Adjusted server timeout: " adjusted-timeout))
	(cond
         ((and *server-run*
	       (> (+ last-access server-timeout)
		  (current-seconds))
	       (< (- (current-seconds) server-start-time) 3600)) ;; do not update log or touch log if we've been running for more than one hour.
          (if (common:low-noise-print 120 "server continuing")
              (debug:print-info 0 *default-log-port* "Server continuing, seconds since last db access: " (- (current-seconds) last-access))
	      (let ((curr-time (current-seconds)))
		(change-file-times server-log-file curr-time curr-time)))
          (loop 0 server-state bad-sync-count (current-milliseconds)))
         (else
          (debug:print-info 0 *default-log-port* "Server timed out. seconds since last db access: " (- (current-seconds) last-access))
          (http-transport:server-shutdown port)))))))

(define (http-transport:server-shutdown port)
  (let ((tdbdat (tasks:open-db)))
    ;;(BB> "http-transport:server-shutdown called")
    (debug:print-info 0 *default-log-port* "Starting to shutdown the server. pid="(current-process-id))
    ;;