2 Bacula® - The Network Backup Solution
4 Copyright (C) 2000-2011 Free Software Foundation Europe e.V.
6 The main author of Bacula is Kern Sibbald, with contributions from
7 many others, a complete list can be found in the file AUTHORS.
8 This program is Free Software; you can redistribute it and/or
9 modify it under the terms of version three of the GNU Affero General Public
10 License as published by the Free Software Foundation and included
13 This program is distributed in the hope that it will be useful, but
14 WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16 General Public License for more details.
18 You should have received a copy of the GNU Affero General Public License
19 along with this program; if not, write to the Free Software
20 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
23 Bacula® is a registered trademark of Kern Sibbald.
24 The licensor of Bacula is the Free Software Foundation Europe
25 (FSFE), Fiduciary Program, Sumatrastrasse 25, 8006 Zürich,
26 Switzerland, email:ftf@fsfeurope.org.
29 * Manipulation routines for Job Control Records and
30 * handling of last_jobs_list.
32 * Kern E. Sibbald, December 2000
34 * These routines are thread safe.
36 * The job list routines were re-written in May 2005 to
37 * eliminate the global lock while traversing the list, and
38 * to use the dlist subroutines. The locking is now done
39 * on the list each time the list is modified or traversed.
40 * That is it is "micro-locked" rather than globally locked.
41 * The result is that there is one lock/unlock for each entry
42 * in the list while traversing it rather than a single lock
43 * at the beginning of a traversal and one at the end. This
44 * incurs slightly more overhead, but effectively eliminates
45 * the possibilty of race conditions. In addition, with the
46 * exception of the global locking of the list during the
47 * re-reading of the config file, no recursion is needed.
54 const int dbglvl = 3400;
56 /* External variables we reference */
58 /* External referenced functions */
59 void free_bregexps(alist *bregexps);
61 /* Forward referenced functions */
62 extern "C" void timeout_handler(int sig);
63 static void jcr_timeout_check(watchdog_t *self);
64 #ifdef TRACE_JCR_CHAIN
65 static void b_lock_jcr_chain(const char *filen, int line);
66 static void b_unlock_jcr_chain(const char *filen, int line);
67 #define lock_jcr_chain() b_lock_jcr_chain(__FILE__, __LINE__);
68 #define unlock_jcr_chain() b_unlock_jcr_chain(__FILE__, __LINE__);
70 static void lock_jcr_chain();
71 static void unlock_jcr_chain();
76 dlist *last_jobs = NULL;
77 const int max_last_jobs = 10;
79 static dlist *jcrs = NULL; /* JCR chain */
80 static pthread_mutex_t jcr_lock = PTHREAD_MUTEX_INITIALIZER;
82 static pthread_mutex_t job_start_mutex = PTHREAD_MUTEX_INITIALIZER;
84 static pthread_mutex_t last_jobs_mutex = PTHREAD_MUTEX_INITIALIZER;
86 static pthread_key_t jcr_key; /* Pointer to jcr for each thread */
88 pthread_once_t key_once = PTHREAD_ONCE_INIT;
101 void init_last_jobs_list()
104 struct s_last_job *job_entry = NULL;
106 last_jobs = New(dlist(job_entry, &job_entry->link));
109 jcrs = New(dlist(jcr, &jcr->link));
113 void term_last_jobs_list()
116 lock_last_jobs_list();
117 while (!last_jobs->empty()) {
118 void *je = last_jobs->first();
119 last_jobs->remove(je);
124 unlock_last_jobs_list();
132 bool read_last_jobs_list(int fd, uint64_t addr)
134 struct s_last_job *je, job;
138 Dmsg1(100, "read_last_jobs seek to %d\n", (int)addr);
139 if (addr == 0 || lseek(fd, (boffset_t)addr, SEEK_SET) < 0) {
142 if (read(fd, &num, sizeof(num)) != sizeof(num)) {
145 Dmsg1(100, "Read num_items=%d\n", num);
146 if (num > 4 * max_last_jobs) { /* sanity check */
149 lock_last_jobs_list();
150 for ( ; num; num--) {
151 if (read(fd, &job, sizeof(job)) != sizeof(job)) {
153 Pmsg1(000, "Read job entry. ERR=%s\n", be.bstrerror());
158 je = (struct s_last_job *)malloc(sizeof(struct s_last_job));
159 memcpy((char *)je, (char *)&job, sizeof(job));
161 init_last_jobs_list();
163 last_jobs->append(je);
164 if (last_jobs->size() > max_last_jobs) {
165 je = (struct s_last_job *)last_jobs->first();
166 last_jobs->remove(je);
171 unlock_last_jobs_list();
175 uint64_t write_last_jobs_list(int fd, uint64_t addr)
177 struct s_last_job *je;
181 Dmsg1(100, "write_last_jobs seek to %d\n", (int)addr);
182 if (lseek(fd, (boffset_t)addr, SEEK_SET) < 0) {
186 lock_last_jobs_list();
187 /* First record is number of entires */
188 num = last_jobs->size();
189 if (write(fd, &num, sizeof(num)) != sizeof(num)) {
191 Pmsg1(000, "Error writing num_items: ERR=%s\n", be.bstrerror());
194 foreach_dlist(je, last_jobs) {
195 if (write(fd, je, sizeof(struct s_last_job)) != sizeof(struct s_last_job)) {
197 Pmsg1(000, "Error writing job: ERR=%s\n", be.bstrerror());
201 unlock_last_jobs_list();
203 /* Return current address */
204 stat = lseek(fd, 0, SEEK_CUR);
211 unlock_last_jobs_list();
215 void lock_last_jobs_list()
220 void unlock_last_jobs_list()
225 /* Get an ASCII representation of the Operation being performed as an english Noun */
226 const char *JCR::get_OperationName()
232 return _("Verifying");
234 return _("Restoring");
236 return _("Archiving");
240 return _("Migration");
242 return _("Scanning");
244 return _("Unknown operation");
248 /* Get an ASCII representation of the Action being performed either an english Verb or Adjective */
249 const char *JCR::get_ActionName(bool past)
255 return (past == true) ? _("verified") : _("verify");
257 return (past == true) ? _("restored") : _("restore");
259 return (past == true) ? _("archived") : _("archive");
261 return (past == true) ? _("copied") : _("copy");
263 return (past == true) ? _("migrated") : _("migrate");
265 return (past == true) ? _("scanned") : _("scan");
267 return _("unknown action");
280 if (m_JobLevel == L_VIRTUAL_FULL) {
291 * Push a subroutine address into the job end callback stack
293 void job_end_push(JCR *jcr, void job_end_cb(JCR *jcr,void *), void *ctx)
295 jcr->job_end_push.append((void *)job_end_cb);
296 jcr->job_end_push.append(ctx);
299 /* Pop each job_end subroutine and call it */
300 static void job_end_pop(JCR *jcr)
302 void (*job_end_cb)(JCR *jcr, void *ctx);
304 for (int i=jcr->job_end_push.size()-1; i > 0; ) {
305 ctx = jcr->job_end_push.get(i--);
306 job_end_cb = (void (*)(JCR *,void *))jcr->job_end_push.get(i--);
307 job_end_cb(jcr, ctx);
312 * Create thread key for thread specific data
314 void create_jcr_key()
316 int status = pthread_key_create(&jcr_key, NULL);
319 Jmsg1(NULL, M_ABORT, 0, _("pthread key create failed: ERR=%s\n"),
320 be.bstrerror(status));
325 * Create a Job Control Record and link it into JCR chain
326 * Returns newly allocated JCR
327 * Note, since each daemon has a different JCR, he passes
330 JCR *new_jcr(int size, JCR_free_HANDLER *daemon_free_jcr)
333 MQUEUE_ITEM *item = NULL;
334 struct sigaction sigtimer;
337 Dmsg0(dbglvl, "Enter new_jcr\n");
338 status = pthread_once(&key_once, create_jcr_key);
341 Jmsg1(NULL, M_ABORT, 0, _("pthread_once failed. ERR=%s\n"), be.bstrerror(status));
343 jcr = (JCR *)malloc(size);
344 memset(jcr, 0, size);
345 jcr->msg_queue = New(dlist(item, &item->link));
346 if ((status = pthread_mutex_init(&jcr->msg_queue_mutex, NULL)) != 0) {
348 Jmsg(NULL, M_ABORT, 0, _("Could not init msg_queue mutex. ERR=%s\n"),
349 be.bstrerror(status));
351 jcr->job_end_push.init(1, false);
352 jcr->sched_time = time(NULL);
353 jcr->daemon_free_jcr = daemon_free_jcr; /* plug daemon free routine */
355 jcr->inc_use_count();
356 jcr->VolumeName = get_pool_memory(PM_FNAME);
357 jcr->VolumeName[0] = 0;
358 jcr->errmsg = get_pool_memory(PM_MESSAGE);
360 jcr->comment = get_pool_memory(PM_FNAME);
362 /* Setup some dummy values */
363 bstrncpy(jcr->Job, "*System*", sizeof(jcr->Job));
365 jcr->setJobType(JT_SYSTEM); /* internal job until defined */
366 jcr->setJobLevel(L_NONE);
367 jcr->setJobStatus(JS_Created); /* ready to run */
368 sigtimer.sa_flags = 0;
369 sigtimer.sa_handler = timeout_handler;
370 sigfillset(&sigtimer.sa_mask);
371 sigaction(TIMEOUT_SIGNAL, &sigtimer, NULL);
374 * Locking jobs is a global lock that is needed
375 * so that the Director can stop new jobs from being
376 * added to the jcr chain while it processes a new
377 * conf file and does the job_end_push().
382 jcrs = New(dlist(jcr, &jcr->link));
393 * Remove a JCR from the chain
394 * NOTE! The chain must be locked prior to calling
397 static void remove_jcr(JCR *jcr)
399 Dmsg0(dbglvl, "Enter remove_jcr\n");
401 Emsg0(M_ABORT, 0, _("NULL jcr.\n"));
404 Dmsg0(dbglvl, "Leave remove_jcr\n");
408 * Free stuff common to all JCRs. N.B. Be careful to include only
409 * generic stuff in the common part of the jcr.
411 static void free_common_jcr(JCR *jcr)
413 /* Uses jcr lock/unlock */
414 remove_jcr_from_tsd(jcr);
415 jcr->set_killable(false);
417 jcr->destroy_mutex();
419 if (jcr->msg_queue) {
420 delete jcr->msg_queue;
421 jcr->msg_queue = NULL;
422 pthread_mutex_destroy(&jcr->msg_queue_mutex);
424 close_msg(jcr); /* close messages for this job */
426 /* do this after closing messages */
427 if (jcr->client_name) {
428 free_pool_memory(jcr->client_name);
429 jcr->client_name = NULL;
433 free_pool_memory(jcr->attr);
437 if (jcr->sd_auth_key) {
438 free(jcr->sd_auth_key);
439 jcr->sd_auth_key = NULL;
441 if (jcr->VolumeName) {
442 free_pool_memory(jcr->VolumeName);
443 jcr->VolumeName = NULL;
446 if (jcr->dir_bsock) {
447 bnet_close(jcr->dir_bsock);
448 jcr->dir_bsock = NULL;
451 free_pool_memory(jcr->errmsg);
458 if (jcr->RegexWhere) {
459 free(jcr->RegexWhere);
460 jcr->RegexWhere = NULL;
462 if (jcr->where_bregexp) {
463 free_bregexps(jcr->where_bregexp);
464 delete jcr->where_bregexp;
465 jcr->where_bregexp = NULL;
467 if (jcr->cached_path) {
468 free_pool_memory(jcr->cached_path);
469 jcr->cached_path = NULL;
473 free_guid_list(jcr->id_list);
477 free_pool_memory(jcr->comment);
484 * Global routine to free a jcr
487 void b_free_jcr(const char *file, int line, JCR *jcr)
489 struct s_last_job *je;
491 Dmsg3(dbglvl, "Enter free_jcr jid=%u from %s:%d\n", jcr->JobId, file, line);
495 void free_jcr(JCR *jcr)
497 struct s_last_job *je;
499 Dmsg3(dbglvl, "Enter free_jcr jid=%u use_count=%d Job=%s\n",
500 jcr->JobId, jcr->use_count(), jcr->Job);
505 jcr->dec_use_count(); /* decrement use count */
506 if (jcr->use_count() < 0) {
507 Jmsg2(jcr, M_ERROR, 0, _("JCR use_count=%d JobId=%d\n"),
508 jcr->use_count(), jcr->JobId);
510 if (jcr->JobId > 0) {
511 Dmsg3(dbglvl, "Dec free_jcr jid=%u use_count=%d Job=%s\n",
512 jcr->JobId, jcr->use_count(), jcr->Job);
514 if (jcr->use_count() > 0) { /* if in use */
518 if (jcr->JobId > 0) {
519 Dmsg3(dbglvl, "remove jcr jid=%u use_count=%d Job=%s\n",
520 jcr->JobId, jcr->use_count(), jcr->Job);
522 remove_jcr(jcr); /* remove Jcr from chain */
525 dequeue_messages(jcr);
526 job_end_pop(jcr); /* pop and call hooked routines */
528 Dmsg1(dbglvl, "End job=%d\n", jcr->JobId);
530 /* Keep some statistics */
531 switch (jcr->getJobType()) {
538 /* Keep list of last jobs, but not Console where JobId==0 */
539 if (jcr->JobId > 0) {
540 lock_last_jobs_list();
542 je = (struct s_last_job *)malloc(sizeof(struct s_last_job));
543 memset(je, 0, sizeof(struct s_last_job)); /* zero in case unset fields */
544 je->Errors = jcr->JobErrors;
545 je->JobType = jcr->getJobType();
546 je->JobId = jcr->JobId;
547 je->VolSessionId = jcr->VolSessionId;
548 je->VolSessionTime = jcr->VolSessionTime;
549 bstrncpy(je->Job, jcr->Job, sizeof(je->Job));
550 je->JobFiles = jcr->JobFiles;
551 je->JobBytes = jcr->JobBytes;
552 je->JobStatus = jcr->JobStatus;
553 je->JobLevel = jcr->getJobLevel();
554 je->start_time = jcr->start_time;
555 je->end_time = time(NULL);
558 init_last_jobs_list();
560 last_jobs->append(je);
561 if (last_jobs->size() > max_last_jobs) {
562 je = (struct s_last_job *)last_jobs->first();
563 last_jobs->remove(je);
566 unlock_last_jobs_list();
573 if (jcr->daemon_free_jcr) {
574 jcr->daemon_free_jcr(jcr); /* call daemon free routine */
577 free_common_jcr(jcr);
578 close_msg(NULL); /* flush any daemon messages */
579 Dmsg0(dbglvl, "Exit free_jcr\n");
583 * Remove jcr from thread specific data, but
584 * but make sure it is us who are attached.
586 void remove_jcr_from_tsd(JCR *jcr)
588 JCR *tjcr = get_jcr_from_tsd();
590 set_jcr_in_tsd(INVALID_JCR);
594 void JCR::set_killable(bool killable)
598 jcr->my_thread_killable = killable;
600 jcr->my_thread_id = pthread_self();
602 memset(&jcr->my_thread_id, 0, sizeof(jcr->my_thread_id));
608 * Put this jcr in the thread specifc data
609 * if update_thread_info is true and the jcr is valide,
610 * we update the my_thread_id in the JCR
612 void set_jcr_in_tsd(JCR *jcr)
614 int status = pthread_setspecific(jcr_key, (void *)jcr);
617 Jmsg1(jcr, M_ABORT, 0, _("pthread_setspecific failed: ERR=%s\n"),
618 be.bstrerror(status));
622 void JCR::my_thread_send_signal(int sig)
625 if (this->is_killable() &&
626 !pthread_equal(this->my_thread_id, pthread_self()))
628 Dmsg1(800, "Send kill to jid=%d\n", this->JobId);
629 pthread_kill(this->my_thread_id, sig);
631 } else if (!this->is_killable()) {
632 Dmsg1(10, "Warning, can't send kill to jid=%d\n", this->JobId);
638 * Give me the jcr that is attached to this thread
640 JCR *get_jcr_from_tsd()
642 JCR *jcr = (JCR *)pthread_getspecific(jcr_key);
643 // printf("get_jcr_from_tsd: jcr=%p\n", jcr);
644 /* set any INVALID_JCR to NULL which the rest of Bacula understands */
645 if (jcr == INVALID_JCR) {
653 * Find which JobId corresponds to the current thread
655 uint32_t get_jobid_from_tsd()
659 jcr = get_jcr_from_tsd();
660 // printf("get_jobid_from_tsr: jcr=%p\n", jcr);
662 JobId = (uint32_t)jcr->JobId;
668 * Given a JobId, find the JCR
669 * Returns: jcr on success
672 JCR *get_jcr_by_id(uint32_t JobId)
677 if (jcr->JobId == JobId) {
678 jcr->inc_use_count();
679 Dmsg3(dbglvl, "Inc get_jcr jid=%u use_count=%d Job=%s\n",
680 jcr->JobId, jcr->use_count(), jcr->Job);
689 * Given a thread id, find the JobId
690 * Returns: JobId on success
693 uint32_t get_jobid_from_tid(pthread_t tid)
699 if (pthread_equal(jcr->my_thread_id, tid)) {
713 * Given a SessionId and SessionTime, find the JCR
714 * Returns: jcr on success
717 JCR *get_jcr_by_session(uint32_t SessionId, uint32_t SessionTime)
722 if (jcr->VolSessionId == SessionId &&
723 jcr->VolSessionTime == SessionTime) {
724 jcr->inc_use_count();
725 Dmsg3(dbglvl, "Inc get_jcr jid=%u use_count=%d Job=%s\n",
726 jcr->JobId, jcr->use_count(), jcr->Job);
736 * Given a Job, find the JCR
737 * compares on the number of characters in Job
738 * thus allowing partial matches.
739 * Returns: jcr on success
742 JCR *get_jcr_by_partial_name(char *Job)
752 if (strncmp(Job, jcr->Job, len) == 0) {
753 jcr->inc_use_count();
754 Dmsg3(dbglvl, "Inc get_jcr jid=%u use_count=%d Job=%s\n",
755 jcr->JobId, jcr->use_count(), jcr->Job);
765 * Given a Job, find the JCR
766 * requires an exact match of names.
767 * Returns: jcr on success
770 JCR *get_jcr_by_full_name(char *Job)
778 if (strcmp(jcr->Job, Job) == 0) {
779 jcr->inc_use_count();
780 Dmsg3(dbglvl, "Inc get_jcr jid=%u use_count=%d Job=%s\n",
781 jcr->JobId, jcr->use_count(), jcr->Job);
789 static void update_wait_time(JCR *jcr, int newJobStatus)
791 bool enter_in_waittime;
792 int oldJobStatus = jcr->JobStatus;
794 switch (newJobStatus) {
799 case JS_WaitStoreRes:
801 case JS_WaitClientRes:
803 case JS_WaitPriority:
804 enter_in_waittime = true;
807 enter_in_waittime = false; /* not a Wait situation */
812 * If we were previously waiting and are not any more
813 * we want to update the wait_time variable, which is
814 * the start of waiting.
816 switch (oldJobStatus) {
821 case JS_WaitStoreRes:
823 case JS_WaitClientRes:
825 case JS_WaitPriority:
826 if (!enter_in_waittime) { /* we get out the wait time */
827 jcr->wait_time_sum += (time(NULL) - jcr->wait_time);
832 /* if wait state is new, we keep current time for watchdog MaxWaitTime */
834 if (enter_in_waittime) {
835 jcr->wait_time = time(NULL);
842 * Priority runs from 0 (lowest) to 10 (highest)
844 static int get_status_priority(int JobStatus)
851 case JS_ErrorTerminated:
866 void JCR::setJobStatus(int newJobStatus)
869 int priority, old_priority;
870 int oldJobStatus = jcr->JobStatus;
871 priority = get_status_priority(newJobStatus);
872 old_priority = get_status_priority(oldJobStatus);
874 Dmsg2(800, "set_jcr_job_status(%s, %c)\n", Job, newJobStatus);
876 /* Update wait_time depending on newJobStatus and oldJobStatus */
877 update_wait_time(jcr, newJobStatus);
880 * For a set of errors, ... keep the current status
881 * so it isn't lost. For all others, set it.
883 Dmsg2(800, "OnEntry JobStatus=%c newJobstatus=%c\n", oldJobStatus, newJobStatus);
885 * If status priority is > than proposed new status, change it.
886 * If status priority == new priority and both are zero, take
888 * If it is not zero, then we keep the first non-zero "error" that
891 if (priority > old_priority || (
892 priority == 0 && old_priority == 0)) {
893 Dmsg4(800, "Set new stat. old: %c,%d new: %c,%d\n",
894 jcr->JobStatus, old_priority, newJobStatus, priority);
895 jcr->JobStatus = newJobStatus; /* replace with new status */
898 if (oldJobStatus != jcr->JobStatus) {
899 Dmsg2(800, "leave setJobStatus old=%c new=%c\n", oldJobStatus, newJobStatus);
900 // generate_plugin_event(jcr, bEventStatusChange, NULL);
904 #ifdef TRACE_JCR_CHAIN
905 static int lock_count = 0;
911 #ifdef TRACE_JCR_CHAIN
912 static void b_lock_jcr_chain(const char *fname, int line)
914 static void lock_jcr_chain()
917 #ifdef TRACE_JCR_CHAIN
918 Dmsg3(dbglvl, "Lock jcr chain %d from %s:%d\n", ++lock_count, fname, line);
926 #ifdef TRACE_JCR_CHAIN
927 static void b_unlock_jcr_chain(const char *fname, int line)
929 static void unlock_jcr_chain()
932 #ifdef TRACE_JCR_CHAIN
933 Dmsg3(dbglvl, "Unlock jcr chain %d from %s:%d\n", lock_count--, fname, line);
939 * Start walk of jcr chain
940 * The proper way to walk the jcr chain is:
947 * It is possible to leave out the endeach_jcr(jcr), but
948 * in that case, the last jcr referenced must be explicitly
954 JCR *jcr_walk_start()
958 jcr = (JCR *)jcrs->first();
960 jcr->inc_use_count();
961 if (jcr->JobId > 0) {
962 Dmsg3(dbglvl, "Inc walk_start jid=%u use_count=%d Job=%s\n",
963 jcr->JobId, jcr->use_count(), jcr->Job);
971 * Get next jcr from chain, and release current one
973 JCR *jcr_walk_next(JCR *prev_jcr)
978 jcr = (JCR *)jcrs->next(prev_jcr);
980 jcr->inc_use_count();
981 if (jcr->JobId > 0) {
982 Dmsg3(dbglvl, "Inc walk_next jid=%u use_count=%d Job=%s\n",
983 jcr->JobId, jcr->use_count(), jcr->Job);
994 * Release last jcr referenced
996 void jcr_walk_end(JCR *jcr)
999 if (jcr->JobId > 0) {
1000 Dmsg3(dbglvl, "Free walk_end jid=%u use_count=%d Job=%s\n",
1001 jcr->JobId, jcr->use_count(), jcr->Job);
1008 * Return number of Jobs
1016 for (jcr = (JCR *)jcrs->first(); (jcr = (JCR *)jcrs->next(jcr)); ) {
1017 if (jcr->JobId > 0) {
1027 * Setup to call the timeout check routine every 30 seconds
1028 * This routine will check any timers that have been enabled.
1030 bool init_jcr_subsystem(void)
1032 watchdog_t *wd = new_watchdog();
1034 wd->one_shot = false;
1035 wd->interval = 30; /* FIXME: should be configurable somewhere, even
1036 if only with a #define */
1037 wd->callback = jcr_timeout_check;
1039 register_watchdog(wd);
1044 static void jcr_timeout_check(watchdog_t *self)
1050 Dmsg0(dbglvl, "Start JCR timeout checks\n");
1052 /* Walk through all JCRs checking if any one is
1053 * blocked for more than specified max time.
1056 Dmsg2(dbglvl, "jcr_timeout_check JobId=%u jcr=0x%x\n", jcr->JobId, jcr);
1057 if (jcr->JobId == 0) {
1060 bs = jcr->store_bsock;
1062 timer_start = bs->timer_start;
1063 if (timer_start && (watchdog_time - timer_start) > bs->timeout) {
1064 bs->timer_start = 0; /* turn off timer */
1065 bs->set_timed_out();
1066 Qmsg(jcr, M_ERROR, 0, _(
1067 "Watchdog sending kill after %d secs to thread stalled reading Storage daemon.\n"),
1068 watchdog_time - timer_start);
1069 jcr->my_thread_send_signal(TIMEOUT_SIGNAL);
1072 bs = jcr->file_bsock;
1074 timer_start = bs->timer_start;
1075 if (timer_start && (watchdog_time - timer_start) > bs->timeout) {
1076 bs->timer_start = 0; /* turn off timer */
1077 bs->set_timed_out();
1078 Qmsg(jcr, M_ERROR, 0, _(
1079 "Watchdog sending kill after %d secs to thread stalled reading File daemon.\n"),
1080 watchdog_time - timer_start);
1081 jcr->my_thread_send_signal(TIMEOUT_SIGNAL);
1084 bs = jcr->dir_bsock;
1086 timer_start = bs->timer_start;
1087 if (timer_start && (watchdog_time - timer_start) > bs->timeout) {
1088 bs->timer_start = 0; /* turn off timer */
1089 bs->set_timed_out();
1090 Qmsg(jcr, M_ERROR, 0, _(
1091 "Watchdog sending kill after %d secs to thread stalled reading Director.\n"),
1092 watchdog_time - timer_start);
1093 jcr->my_thread_send_signal(TIMEOUT_SIGNAL);
1099 Dmsg0(dbglvl, "Finished JCR timeout checks\n");
1103 * Return next JobId from comma separated list
1106 * 1 if next JobId returned
1107 * 0 if no more JobIds are in list
1108 * -1 there is an error
1110 int get_next_jobid_from_list(char **p, uint32_t *JobId)
1112 const int maxlen = 30;
1113 char jobid[maxlen+1];
1117 for (int i=0; i<maxlen; i++) {
1120 } else if (*q == ',') {
1127 if (jobid[0] == 0) {
1129 } else if (!is_a_number(jobid)) {
1130 return -1; /* error */
1133 *JobId = str_to_int64(jobid);
1138 * Timeout signal comes here
1140 extern "C" void timeout_handler(int sig)
1142 return; /* thus interrupting the function */
1145 /* Used to display specific daemon information after a fatal signal
1146 * (like B_DB in the director)
1148 #define MAX_DBG_HOOK 10
1149 static dbg_jcr_hook_t *dbg_jcr_hooks[MAX_DBG_HOOK];
1150 static int dbg_jcr_handler_count;
1152 void dbg_jcr_add_hook(dbg_jcr_hook_t *hook)
1154 ASSERT(dbg_jcr_handler_count < MAX_DBG_HOOK);
1155 dbg_jcr_hooks[dbg_jcr_handler_count++] = hook;
1161 * This function should be used ONLY after a fatal signal. We walk through the
1162 * JCR chain without doing any lock, Bacula should not be running.
1164 void dbg_print_jcr(FILE *fp)
1166 char buf1[128], buf2[128], buf3[128], buf4[128];
1171 fprintf(fp, "Attempt to dump current JCRs. njcrs=%d\n", jcrs->size());
1173 for (JCR *jcr = (JCR *)jcrs->first(); jcr ; jcr = (JCR *)jcrs->next(jcr)) {
1174 fprintf(fp, "threadid=%p JobId=%d JobStatus=%c jcr=%p name=%s\n",
1175 (void *)jcr->my_thread_id, (int)jcr->JobId, jcr->JobStatus, jcr, jcr->Job);
1176 fprintf(fp, "threadid=%p killable=%d JobId=%d JobStatus=%c "
1178 (void *)jcr->my_thread_id, jcr->is_killable(),
1179 (int)jcr->JobId, jcr->JobStatus, jcr, jcr->Job);
1180 fprintf(fp, "\tuse_count=%i\n", jcr->use_count());
1181 fprintf(fp, "\tJobType=%c JobLevel=%c\n",
1182 jcr->getJobType(), jcr->getJobLevel());
1183 bstrftime(buf1, sizeof(buf1), jcr->sched_time);
1184 bstrftime(buf2, sizeof(buf2), jcr->start_time);
1185 bstrftime(buf3, sizeof(buf3), jcr->end_time);
1186 bstrftime(buf4, sizeof(buf4), jcr->wait_time);
1187 fprintf(fp, "\tsched_time=%s start_time=%s\n\tend_time=%s wait_time=%s\n",
1188 buf1, buf2, buf3, buf4);
1189 fprintf(fp, "\tdb=%p db_batch=%p batch_started=%i\n",
1190 jcr->db, jcr->db_batch, jcr->batch_started);
1193 * Call all the jcr debug hooks
1195 for(int i=0; i < dbg_jcr_handler_count; i++) {
1196 dbg_jcr_hook_t *hook = dbg_jcr_hooks[i];