3 * Bacula Director -- backup.c -- responsible for doing backup jobs
5 * Kern Sibbald, March MM
7 * Basic tasks done here:
8 * Open DB and create records for this job.
9 * Open Message Channel with Storage daemon to tell him a job will be starting.
10 * Open connection with File daemon and pass him commands
12 * When the File daemon finishes the job, update the DB.
17 Bacula® - The Network Backup Solution
19 Copyright (C) 2000-2007 Free Software Foundation Europe e.V.
21 The main author of Bacula is Kern Sibbald, with contributions from
22 many others, a complete list can be found in the file AUTHORS.
23 This program is Free Software; you can redistribute it and/or
24 modify it under the terms of version two of the GNU General Public
25 License as published by the Free Software Foundation plus additions
26 that are listed in the file LICENSE.
28 This program is distributed in the hope that it will be useful, but
29 WITHOUT ANY WARRANTY; without even the implied warranty of
30 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
31 General Public License for more details.
33 You should have received a copy of the GNU General Public License
34 along with this program; if not, write to the Free Software
35 Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
38 Bacula® is a registered trademark of John Walker.
39 The licensor of Bacula is the Free Software Foundation Europe
40 (FSFE), Fiduciary Program, Sumatrastrasse 25, 8006 Zürich,
41 Switzerland, email:ftf@fsfeurope.org.
48 /* Commands sent to File daemon */
49 static char backupcmd[] = "backup\n";
50 static char storaddr[] = "storage address=%s port=%d ssl=%d\n";
52 /* Responses received from File daemon */
53 static char OKbackup[] = "2000 OK backup\n";
54 static char OKstore[] = "2000 OK storage\n";
55 static char EndJob[] = "2800 End Job TermCode=%d JobFiles=%u "
56 "ReadBytes=%llu JobBytes=%llu Errors=%u "
57 "VSS=%d Encrypt=%d\n";
58 /* Pre 1.39.29 (04Dec06) EndJob */
59 static char OldEndJob[] = "2800 End Job TermCode=%d JobFiles=%u "
60 "ReadBytes=%llu JobBytes=%llu Errors=%u\n";
62 * Called here before the job is run to do the job
65 bool do_backup_init(JCR *jcr)
68 free_rstorage(jcr); /* we don't read so release */
70 if (!get_or_create_fileset_record(jcr)) {
75 * Get definitive Job level and since time
77 get_level_since_time(jcr, jcr->since, sizeof(jcr->since));
79 apply_pool_overrides(jcr);
81 jcr->jr.PoolId = get_or_create_pool_record(jcr, jcr->pool->name());
82 if (jcr->jr.PoolId == 0) {
86 /* If pool storage specified, use it instead of job storage */
87 copy_wstorage(jcr, jcr->pool->storage, _("Pool resource"));
90 Jmsg(jcr, M_FATAL, 0, _("No Storage specification found in Job or Pool.\n"));
94 create_clones(jcr); /* run any clone jobs */
100 * Do a backup of the specified FileSet
102 * Returns: false on failure
105 bool do_backup(JCR *jcr)
108 int tls_need = BNET_TLS_NONE;
114 /* Print Job Start message */
115 Jmsg(jcr, M_INFO, 0, _("Start Backup JobId %s, Job=%s\n"),
116 edit_uint64(jcr->JobId, ed1), jcr->Job);
118 set_jcr_job_status(jcr, JS_Running);
119 Dmsg2(100, "JobId=%d JobLevel=%c\n", jcr->jr.JobId, jcr->jr.JobLevel);
120 if (!db_update_job_start_record(jcr, jcr->db, &jcr->jr)) {
121 Jmsg(jcr, M_FATAL, 0, "%s", db_strerror(jcr->db));
126 * Open a message channel connection with the Storage
127 * daemon. This is to let him know that our client
128 * will be contacting him for a backup session.
131 Dmsg0(110, "Open connection with storage daemon\n");
132 set_jcr_job_status(jcr, JS_WaitSD);
134 * Start conversation with Storage daemon
136 if (!connect_to_storage_daemon(jcr, 10, SDConnectTimeout, 1)) {
140 * Now start a job with the Storage daemon
142 if (!start_storage_daemon_job(jcr, NULL, jcr->wstorage)) {
147 * Start the job prior to starting the message thread below
148 * to avoid two threads from using the BSOCK structure at
151 if (!bnet_fsend(jcr->store_bsock, "run")) {
156 * Now start a Storage daemon message thread. Note,
157 * this thread is used to provide the catalog services
158 * for the backup job, including inserting the attributes
159 * into the catalog. See catalog_update() in catreq.c
161 if (!start_storage_daemon_message_thread(jcr)) {
164 Dmsg0(150, "Storage daemon connection OK\n");
166 set_jcr_job_status(jcr, JS_WaitFD);
167 if (!connect_to_file_daemon(jcr, 10, FDConnectTimeout, 1)) {
171 set_jcr_job_status(jcr, JS_Running);
172 fd = jcr->file_bsock;
174 if (!send_level_command(jcr)) {
178 if (!send_runscripts_commands(jcr)) {
182 if (!send_include_list(jcr)) {
186 if (!send_exclude_list(jcr)) {
191 * send Storage daemon address to the File daemon
194 if (store->SDDport == 0) {
195 store->SDDport = store->SDport;
198 /* TLS Requirement */
199 if (store->tls_enable) {
200 if (store->tls_require) {
201 tls_need = BNET_TLS_REQUIRED;
203 tls_need = BNET_TLS_OK;
207 bnet_fsend(fd, storaddr, store->address, store->SDDport, tls_need);
208 if (!response(jcr, fd, OKstore, "Storage", DISPLAY_ERROR)) {
213 * We re-update the job start record so that the start
214 * time is set after the run before job. This avoids
215 * that any files created by the run before job will
216 * be saved twice. They will be backed up in the current
217 * job, but not in the next one unless they are changed.
218 * Without this, they will be backed up in this job and
219 * in the next job run because in that case, their date
220 * is after the start of this run.
222 jcr->start_time = time(NULL);
223 jcr->jr.StartTime = jcr->start_time;
224 if (!db_update_job_start_record(jcr, jcr->db, &jcr->jr)) {
225 Jmsg(jcr, M_FATAL, 0, "%s", db_strerror(jcr->db));
228 /* Send backup command */
229 bnet_fsend(fd, backupcmd);
230 if (!response(jcr, fd, OKbackup, "backup", DISPLAY_ERROR)) {
234 /* Pickup Job termination data */
235 stat = wait_for_job_termination(jcr);
236 #ifdef HAVE_BATCH_FILE_INSERT
237 db_create_batch_file_record(jcr); /* used by bulk batch file insert */
239 if (stat == JS_Terminated) {
240 backup_cleanup(jcr, stat);
245 /* Come here only after starting SD thread */
247 set_jcr_job_status(jcr, JS_ErrorTerminated);
248 Dmsg1(400, "wait for sd. use=%d\n", jcr->use_count());
249 wait_for_storage_daemon_termination(jcr);
250 Dmsg1(400, "after wait for sd. use=%d\n", jcr->use_count());
256 * Here we wait for the File daemon to signal termination,
257 * then we wait for the Storage daemon. When both
258 * are done, we return the job status.
259 * Also used by restore.c
261 int wait_for_job_termination(JCR *jcr)
264 BSOCK *fd = jcr->file_bsock;
266 uint32_t JobFiles, Errors;
267 uint64_t ReadBytes = 0;
268 uint64_t JobBytes = 0;
272 set_jcr_job_status(jcr, JS_Running);
273 /* Wait for Client to terminate */
274 while ((n = bget_dirmsg(fd)) >= 0) {
276 (sscanf(fd->msg, EndJob, &jcr->FDJobStatus, &JobFiles,
277 &ReadBytes, &JobBytes, &Errors, &VSS, &Encrypt) == 7 ||
278 sscanf(fd->msg, OldEndJob, &jcr->FDJobStatus, &JobFiles,
279 &ReadBytes, &JobBytes, &Errors) == 5)) {
281 set_jcr_job_status(jcr, jcr->FDJobStatus);
282 Dmsg1(100, "FDStatus=%c\n", (char)jcr->JobStatus);
284 Jmsg(jcr, M_WARNING, 0, _("Unexpected Client Job message: %s\n"),
287 if (job_canceled(jcr)) {
292 if (is_bnet_error(fd)) {
293 Jmsg(jcr, M_FATAL, 0, _("Network error with FD during %s: ERR=%s\n"),
294 job_type_to_str(jcr->JobType), bnet_strerror(fd));
296 bnet_sig(fd, BNET_TERMINATE); /* tell Client we are terminating */
298 /* Note, the SD stores in jcr->JobFiles/ReadBytes/JobBytes/Errors */
299 wait_for_storage_daemon_termination(jcr);
302 /* Return values from FD */
304 jcr->JobFiles = JobFiles;
305 jcr->Errors = Errors;
306 jcr->ReadBytes = ReadBytes;
307 jcr->JobBytes = JobBytes;
309 jcr->Encrypt = Encrypt;
311 Jmsg(jcr, M_FATAL, 0, _("No Job status returned from FD.\n"));
314 // Dmsg4(100, "fd_ok=%d FDJS=%d JS=%d SDJS=%d\n", fd_ok, jcr->FDJobStatus,
315 // jcr->JobStatus, jcr->SDJobStatus);
317 /* Return the first error status we find Dir, FD, or SD */
318 if (!fd_ok || is_bnet_error(fd)) {
319 jcr->FDJobStatus = JS_ErrorTerminated;
321 if (jcr->JobStatus != JS_Terminated) {
322 return jcr->JobStatus;
324 if (jcr->FDJobStatus != JS_Terminated) {
325 return jcr->FDJobStatus;
327 return jcr->SDJobStatus;
331 * Release resources allocated during backup.
333 void backup_cleanup(JCR *jcr, int TermCode)
335 char sdt[50], edt[50], schedt[50];
336 char ec1[30], ec2[30], ec3[30], ec4[30], ec5[30], compress[50];
337 char ec6[30], ec7[30], ec8[30], elapsed[50];
338 char term_code[100], fd_term_msg[100], sd_term_msg[100];
339 const char *term_msg;
340 int msg_type = M_INFO;
343 double kbps, compression;
346 Dmsg2(100, "Enter backup_cleanup %d %c\n", TermCode, TermCode);
347 memset(&mr, 0, sizeof(mr));
348 memset(&cr, 0, sizeof(cr));
350 update_job_end(jcr, TermCode);
352 if (!db_get_job_record(jcr, jcr->db, &jcr->jr)) {
353 Jmsg(jcr, M_WARNING, 0, _("Error getting job record for stats: %s"),
354 db_strerror(jcr->db));
355 set_jcr_job_status(jcr, JS_ErrorTerminated);
358 bstrncpy(cr.Name, jcr->client->hdr.name, sizeof(cr.Name));
359 if (!db_get_client_record(jcr, jcr->db, &cr)) {
360 Jmsg(jcr, M_WARNING, 0, _("Error getting client record for stats: %s"),
361 db_strerror(jcr->db));
364 bstrncpy(mr.VolumeName, jcr->VolumeName, sizeof(mr.VolumeName));
365 if (!db_get_media_record(jcr, jcr->db, &mr)) {
366 Jmsg(jcr, M_WARNING, 0, _("Error getting Media record for Volume \"%s\": ERR=%s"),
367 mr.VolumeName, db_strerror(jcr->db));
368 set_jcr_job_status(jcr, JS_ErrorTerminated);
371 update_bootstrap_file(jcr);
373 switch (jcr->JobStatus) {
375 if (jcr->Errors || jcr->SDErrors) {
376 term_msg = _("Backup OK -- with warnings");
378 term_msg = _("Backup OK");
382 case JS_ErrorTerminated:
383 term_msg = _("*** Backup Error ***");
384 msg_type = M_ERROR; /* Generate error message */
385 if (jcr->store_bsock) {
386 bnet_sig(jcr->store_bsock, BNET_TERMINATE);
387 if (jcr->SD_msg_chan) {
388 pthread_cancel(jcr->SD_msg_chan);
393 term_msg = _("Backup Canceled");
394 if (jcr->store_bsock) {
395 bnet_sig(jcr->store_bsock, BNET_TERMINATE);
396 if (jcr->SD_msg_chan) {
397 pthread_cancel(jcr->SD_msg_chan);
402 term_msg = term_code;
403 sprintf(term_code, _("Inappropriate term code: %c\n"), jcr->JobStatus);
406 bstrftimes(schedt, sizeof(schedt), jcr->jr.SchedTime);
407 bstrftimes(sdt, sizeof(sdt), jcr->jr.StartTime);
408 bstrftimes(edt, sizeof(edt), jcr->jr.EndTime);
409 RunTime = jcr->jr.EndTime - jcr->jr.StartTime;
413 kbps = (double)jcr->jr.JobBytes / (1000 * RunTime);
415 if (!db_get_job_volume_names(jcr, jcr->db, jcr->jr.JobId, &jcr->VolumeName)) {
417 * Note, if the job has erred, most likely it did not write any
418 * tape, so suppress this "error" message since in that case
419 * it is normal. Or look at it the other way, only for a
420 * normal exit should we complain about this error.
422 if (jcr->JobStatus == JS_Terminated && jcr->jr.JobBytes) {
423 Jmsg(jcr, M_ERROR, 0, "%s", db_strerror(jcr->db));
425 jcr->VolumeName[0] = 0; /* none */
428 if (jcr->ReadBytes == 0) {
429 bstrncpy(compress, "None", sizeof(compress));
431 compression = (double)100 - 100.0 * ((double)jcr->JobBytes / (double)jcr->ReadBytes);
432 if (compression < 0.5) {
433 bstrncpy(compress, "None", sizeof(compress));
435 bsnprintf(compress, sizeof(compress), "%.1f %%", (float)compression);
438 jobstatus_to_ascii(jcr->FDJobStatus, fd_term_msg, sizeof(fd_term_msg));
439 jobstatus_to_ascii(jcr->SDJobStatus, sd_term_msg, sizeof(sd_term_msg));
441 // bmicrosleep(15, 0); /* for debugging SIGHUP */
443 Jmsg(jcr, msg_type, 0, _("Bacula %s (%s): %s\n"
446 " Backup Level: %s%s\n"
447 " Client: \"%s\" %s\n"
448 " FileSet: \"%s\" %s\n"
449 " Pool: \"%s\" (From %s)\n"
450 " Storage: \"%s\" (From %s)\n"
451 " Scheduled time: %s\n"
454 " Elapsed time: %s\n"
456 " FD Files Written: %s\n"
457 " SD Files Written: %s\n"
458 " FD Bytes Written: %s (%sB)\n"
459 " SD Bytes Written: %s (%sB)\n"
461 " Software Compression: %s\n"
464 " Volume name(s): %s\n"
465 " Volume Session Id: %d\n"
466 " Volume Session Time: %d\n"
467 " Last Volume Bytes: %s (%sB)\n"
468 " Non-fatal FD errors: %d\n"
470 " FD termination status: %s\n"
471 " SD termination status: %s\n"
472 " Termination: %s\n\n"),
478 level_to_str(jcr->JobLevel), jcr->since,
479 jcr->client->name(), cr.Uname,
480 jcr->fileset->name(), jcr->FSCreateTime,
481 jcr->pool->name(), jcr->pool_source,
482 jcr->wstore->name(), jcr->wstore_source,
486 edit_utime(RunTime, elapsed, sizeof(elapsed)),
488 edit_uint64_with_commas(jcr->jr.JobFiles, ec1),
489 edit_uint64_with_commas(jcr->SDJobFiles, ec2),
490 edit_uint64_with_commas(jcr->jr.JobBytes, ec3),
491 edit_uint64_with_suffix(jcr->jr.JobBytes, ec4),
492 edit_uint64_with_commas(jcr->SDJobBytes, ec5),
493 edit_uint64_with_suffix(jcr->SDJobBytes, ec6),
497 jcr->Encrypt?"yes":"no",
501 edit_uint64_with_commas(mr.VolBytes, ec7),
502 edit_uint64_with_suffix(mr.VolBytes, ec8),
509 Dmsg0(100, "Leave backup_cleanup()\n");
512 void update_bootstrap_file(JCR *jcr)
514 /* Now update the bootstrap file if any */
515 if (jcr->JobStatus == JS_Terminated && jcr->jr.JobBytes &&
516 jcr->job->WriteBootstrap) {
520 POOLMEM *fname = get_pool_memory(PM_FNAME);
521 fname = edit_job_codes(jcr, fname, jcr->job->WriteBootstrap, "");
523 VOL_PARAMS *VolParams = NULL;
529 bpipe = open_bpipe(fname+1, 0, "w"); /* skip first char "|" */
530 fd = bpipe ? bpipe->wfd : NULL;
532 /* ***FIXME*** handle BASE */
533 fd = fopen(fname, jcr->JobLevel==L_FULL?"w+b":"a+b");
536 VolCount = db_get_job_volume_parameters(jcr, jcr->db, jcr->JobId,
539 Jmsg(jcr, M_ERROR, 0, _("Could not get Job Volume Parameters to "
540 "update Bootstrap file. ERR=%s\n"), db_strerror(jcr->db));
541 if (jcr->SDJobFiles != 0) {
542 set_jcr_job_status(jcr, JS_ErrorTerminated);
546 /* Start output with when and who wrote it */
547 bstrftimes(edt, sizeof(edt), time(NULL));
548 fprintf(fd, "# %s - %s - %s%s\n", edt, jcr->jr.Job,
549 level_to_str(jcr->JobLevel), jcr->since);
550 for (int i=0; i < VolCount; i++) {
551 /* Write the record */
552 fprintf(fd, "Volume=\"%s\"\n", VolParams[i].VolumeName);
553 fprintf(fd, "MediaType=\"%s\"\n", VolParams[i].MediaType);
554 fprintf(fd, "VolSessionId=%u\n", jcr->VolSessionId);
555 fprintf(fd, "VolSessionTime=%u\n", jcr->VolSessionTime);
556 fprintf(fd, "VolFile=%u-%u\n", VolParams[i].StartFile,
557 VolParams[i].EndFile);
558 fprintf(fd, "VolBlock=%u-%u\n", VolParams[i].StartBlock,
559 VolParams[i].EndBlock);
560 fprintf(fd, "FileIndex=%d-%d\n", VolParams[i].FirstIndex,
561 VolParams[i].LastIndex);
573 Jmsg(jcr, M_ERROR, 0, _("Could not open WriteBootstrap file:\n"
574 "%s: ERR=%s\n"), fname, be.strerror());
575 set_jcr_job_status(jcr, JS_ErrorTerminated);
577 free_pool_memory(fname);