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.
18 Copyright (C) 2000-2003 Kern Sibbald and John Walker
20 This program is free software; you can redistribute it and/or
21 modify it under the terms of the GNU General Public License as
22 published by the Free Software Foundation; either version 2 of
23 the License, or (at your option) any later version.
25 This program is distributed in the hope that it will be useful,
26 but WITHOUT ANY WARRANTY; without even the implied warranty of
27 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
28 General Public License for more details.
30 You should have received a copy of the GNU General Public
31 License along with this program; if not, write to the Free
32 Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
41 /* Commands sent to File daemon */
42 static char backupcmd[] = "backup\n";
43 static char storaddr[] = "storage address=%s port=%d ssl=%d\n";
45 /* Responses received from File daemon */
46 static char OKbackup[] = "2000 OK backup\n";
47 static char OKstore[] = "2000 OK storage\n";
48 static char EndJob[] = "2800 End Job TermCode=%d JobFiles=%u "
49 "ReadBytes=%" lld " JobBytes=%" lld " Errors=%u\n";
52 /* Forward referenced functions */
53 static void backup_cleanup(JCR *jcr, int TermCode, char *since, FILESET_DBR *fsr);
55 /* External functions */
58 * Do a backup of the specified FileSet
60 * Returns: 0 on failure
63 int do_backup(JCR *jcr)
65 char since[MAXSTRING];
73 if (!get_or_create_client_record(jcr)) {
74 Jmsg(jcr, M_ERROR, 0, _("Could not get/create Client record. ERR=%s\n"),
75 db_strerror(jcr->db));
80 * Get or Create FileSet record
82 memset(&fsr, 0, sizeof(fsr));
83 bstrncpy(fsr.FileSet, jcr->fileset->hdr.name, sizeof(fsr.FileSet));
84 if (jcr->fileset->have_MD5) {
85 struct MD5Context md5c;
86 unsigned char signature[16];
87 memcpy(&md5c, &jcr->fileset->md5c, sizeof(md5c));
88 MD5Final(signature, &md5c);
89 bin_to_base64(fsr.MD5, (char *)signature, 16); /* encode 16 bytes */
90 bstrncpy(jcr->fileset->MD5, fsr.MD5, sizeof(jcr->fileset->MD5));
92 Jmsg(jcr, M_WARNING, 0, _("FileSet MD5 signature not found.\n"));
94 if (!db_create_fileset_record(jcr, jcr->db, &fsr)) {
95 Jmsg(jcr, M_ERROR, 0, _("Could not create FileSet \"%s\" record. ERR=%s\n"),
96 fsr.FileSet, db_strerror(jcr->db));
99 jcr->jr.FileSetId = fsr.FileSetId;
101 Jmsg(jcr, M_INFO, 0, _("Created new FileSet record \"%s\" %s\n"),
102 fsr.FileSet, fsr.cCreateTime);
104 Dmsg2(119, "Created FileSet %s record %u\n", jcr->fileset->hdr.name,
107 get_level_since_time(jcr, since, sizeof(since));
109 jcr->jr.JobId = jcr->JobId;
110 jcr->jr.StartTime = jcr->start_time;
111 if (!db_update_job_start_record(jcr, jcr->db, &jcr->jr)) {
112 Jmsg(jcr, M_ERROR, 0, "%s", db_strerror(jcr->db));
116 jcr->fname = (char *) get_pool_memory(PM_FNAME);
118 /* Print Job Start message */
119 Jmsg(jcr, M_INFO, 0, _("Start Backup JobId %u, Job=%s\n"),
120 jcr->JobId, jcr->Job);
123 * Get the Pool record -- first apply any level defined pools
125 switch (jcr->JobLevel) {
127 if (jcr->full_pool) {
128 jcr->pool = jcr->full_pool;
133 jcr->pool = jcr->inc_pool;
138 jcr->pool = jcr->dif_pool;
142 memset(&pr, 0, sizeof(pr));
143 bstrncpy(pr.Name, jcr->pool->hdr.name, sizeof(pr.Name));
145 while (!db_get_pool_record(jcr, jcr->db, &pr)) { /* get by Name */
146 /* Try to create the pool */
147 if (create_pool(jcr, jcr->db, jcr->pool, POOL_OP_CREATE) < 0) {
148 Jmsg(jcr, M_FATAL, 0, _("Pool %s not in database. %s"), pr.Name,
149 db_strerror(jcr->db));
152 Jmsg(jcr, M_INFO, 0, _("Pool %s created in database.\n"), pr.Name);
155 jcr->PoolId = pr.PoolId; /****FIXME**** this can go away */
156 jcr->jr.PoolId = pr.PoolId;
159 * Open a message channel connection with the Storage
160 * daemon. This is to let him know that our client
161 * will be contacting him for a backup session.
164 Dmsg0(110, "Open connection with storage daemon\n");
165 set_jcr_job_status(jcr, JS_WaitSD);
167 * Start conversation with Storage daemon
169 if (!connect_to_storage_daemon(jcr, 10, SDConnectTimeout, 1)) {
173 * Now start a job with the Storage daemon
175 if (!start_storage_daemon_job(jcr)) {
179 * Now start a Storage daemon message thread
181 if (!start_storage_daemon_message_thread(jcr)) {
184 Dmsg0(150, "Storage daemon connection OK\n");
186 set_jcr_job_status(jcr, JS_WaitFD);
187 if (!connect_to_file_daemon(jcr, 10, FDConnectTimeout, 1)) {
191 set_jcr_job_status(jcr, JS_Running);
192 fd = jcr->file_bsock;
194 if (!send_include_list(jcr)) {
198 if (!send_exclude_list(jcr)) {
203 * send Storage daemon address to the File daemon
205 if (jcr->store->SDDport == 0) {
206 jcr->store->SDDport = jcr->store->SDport;
208 bnet_fsend(fd, storaddr, jcr->store->address, jcr->store->SDDport,
209 jcr->store->enable_ssl);
210 if (!response(jcr, fd, OKstore, "Storage", DISPLAY_ERROR)) {
215 if (!send_level_command(jcr)) {
219 if (!send_run_before_and_after_commands(jcr)) {
223 /* Send backup command */
224 bnet_fsend(fd, backupcmd);
225 if (!response(jcr, fd, OKbackup, "backup", DISPLAY_ERROR)) {
229 /* Pickup Job termination data */
230 stat = wait_for_job_termination(jcr);
231 backup_cleanup(jcr, stat, since, &fsr);
235 backup_cleanup(jcr, JS_ErrorTerminated, since, &fsr);
240 * Here we wait for the File daemon to signal termination,
241 * then we wait for the Storage daemon. When both
242 * are done, we return the job status.
243 * Also used by restore.c
245 int wait_for_job_termination(JCR *jcr)
248 BSOCK *fd = jcr->file_bsock;
250 uint32_t JobFiles, Errors;
251 uint64_t ReadBytes, JobBytes;
253 set_jcr_job_status(jcr, JS_Running);
254 /* Wait for Client to terminate */
255 while ((n = bget_dirmsg(fd)) >= 0) {
256 if (!fd_ok && sscanf(fd->msg, EndJob, &jcr->FDJobStatus, &JobFiles,
257 &ReadBytes, &JobBytes, &Errors) == 5) {
259 set_jcr_job_status(jcr, jcr->FDJobStatus);
260 Dmsg1(100, "FDStatus=%c\n", (char)jcr->JobStatus);
262 Jmsg(jcr, M_WARNING, 0, _("Unexpected Client Job message: %s\n"),
265 if (job_canceled(jcr)) {
269 if (is_bnet_error(fd)) {
270 Jmsg(jcr, M_FATAL, 0, _("Network error with FD during %s: ERR=%s\n"),
271 job_type_to_str(jcr->JobType), bnet_strerror(fd));
273 bnet_sig(fd, BNET_TERMINATE); /* tell Client we are terminating */
275 /* Note, the SD stores in jcr->JobFiles/ReadBytes/JobBytes/Errors */
276 wait_for_storage_daemon_termination(jcr);
279 /* Return values from FD */
281 jcr->JobFiles = JobFiles;
282 jcr->Errors = Errors;
283 jcr->ReadBytes = ReadBytes;
284 jcr->JobBytes = JobBytes;
286 Jmsg(jcr, M_FATAL, 0, _("No Job status returned from FD.\n"));
289 // Dmsg4(100, "fd_ok=%d FDJS=%d JS=%d SDJS=%d\n", fd_ok, jcr->FDJobStatus,
290 // jcr->JobStatus, jcr->SDJobStatus);
292 /* Return the first error status we find Dir, FD, or SD */
293 if (!fd_ok || is_bnet_error(fd)) {
294 jcr->FDJobStatus = JS_ErrorTerminated;
296 if (jcr->JobStatus != JS_Terminated) {
297 return jcr->JobStatus;
299 if (jcr->FDJobStatus != JS_Terminated) {
300 return jcr->FDJobStatus;
302 return jcr->SDJobStatus;
306 * Release resources allocated during backup.
308 static void backup_cleanup(JCR *jcr, int TermCode, char *since, FILESET_DBR *fsr)
310 char sdt[50], edt[50];
311 char ec1[30], ec2[30], ec3[30], ec4[30], ec5[30], compress[50];
312 char term_code[100], fd_term_msg[100], sd_term_msg[100];
316 double kbps, compression;
319 Dmsg2(100, "Enter backup_cleanup %d %c\n", TermCode, TermCode);
320 memset(&mr, 0, sizeof(mr));
321 set_jcr_job_status(jcr, TermCode);
323 update_job_end_record(jcr); /* update database */
325 if (!db_get_job_record(jcr, jcr->db, &jcr->jr)) {
326 Jmsg(jcr, M_WARNING, 0, _("Error getting job record for stats: %s"),
327 db_strerror(jcr->db));
328 set_jcr_job_status(jcr, JS_ErrorTerminated);
331 bstrncpy(mr.VolumeName, jcr->VolumeName, sizeof(mr.VolumeName));
332 if (!db_get_media_record(jcr, jcr->db, &mr)) {
333 Jmsg(jcr, M_WARNING, 0, _("Error getting Media record for Volume \"%s\": ERR=%s"),
334 mr.VolumeName, db_strerror(jcr->db));
335 set_jcr_job_status(jcr, JS_ErrorTerminated);
338 /* Now update the bootstrap file if any */
339 if (jcr->JobStatus == JS_Terminated && jcr->jr.JobBytes &&
340 jcr->job->WriteBootstrap) {
344 char *fname = jcr->job->WriteBootstrap;
345 VOL_PARAMS *VolParams = NULL;
351 bpipe = open_bpipe(fname, 0, "w");
352 fd = bpipe ? bpipe->wfd : NULL;
354 /* ***FIXME*** handle BASE */
355 fd = fopen(fname, jcr->JobLevel==L_FULL?"w+":"a+");
358 VolCount = db_get_job_volume_parameters(jcr, jcr->db, jcr->JobId,
361 Jmsg(jcr, M_ERROR, 0, _("Could not get Job Volume Parameters to "
362 "update Bootstrap file. ERR=%s\n"), db_strerror(jcr->db));
363 if (jcr->SDJobFiles != 0) {
364 set_jcr_job_status(jcr, JS_ErrorTerminated);
368 for (int i=0; i < VolCount; i++) {
369 /* Write the record */
370 fprintf(fd, "Volume=\"%s\"\n", VolParams[i].VolumeName);
371 fprintf(fd, "VolSessionId=%u\n", jcr->VolSessionId);
372 fprintf(fd, "VolSessionTime=%u\n", jcr->VolSessionTime);
373 fprintf(fd, "VolFile=%u-%u\n", VolParams[i].StartFile,
374 VolParams[i].EndFile);
375 fprintf(fd, "VolBlock=%u-%u\n", VolParams[i].StartBlock,
376 VolParams[i].EndBlock);
377 fprintf(fd, "FileIndex=%d-%d\n", VolParams[i].FirstIndex,
378 VolParams[i].LastIndex);
389 Jmsg(jcr, M_ERROR, 0, _("Could not open WriteBootstrap file:\n"
390 "%s: ERR=%s\n"), fname, strerror(errno));
391 set_jcr_job_status(jcr, JS_ErrorTerminated);
395 msg_type = M_INFO; /* by default INFO message */
396 switch (jcr->JobStatus) {
398 term_msg = _("Backup OK");
401 case JS_ErrorTerminated:
402 term_msg = _("*** Backup Error ***");
403 msg_type = M_ERROR; /* Generate error message */
404 if (jcr->store_bsock) {
405 bnet_sig(jcr->store_bsock, BNET_TERMINATE);
406 if (jcr->SD_msg_chan) {
407 pthread_cancel(jcr->SD_msg_chan);
412 term_msg = _("Backup Canceled");
413 if (jcr->store_bsock) {
414 bnet_sig(jcr->store_bsock, BNET_TERMINATE);
415 if (jcr->SD_msg_chan) {
416 pthread_cancel(jcr->SD_msg_chan);
421 term_msg = term_code;
422 sprintf(term_code, _("Inappropriate term code: %c\n"), jcr->JobStatus);
425 bstrftime(sdt, sizeof(sdt), jcr->jr.StartTime);
426 bstrftime(edt, sizeof(edt), jcr->jr.EndTime);
427 RunTime = jcr->jr.EndTime - jcr->jr.StartTime;
431 kbps = (double)jcr->jr.JobBytes / (1000 * RunTime);
433 if (!db_get_job_volume_names(jcr, jcr->db, jcr->jr.JobId, &jcr->VolumeName)) {
435 * Note, if the job has erred, most likely it did not write any
436 * tape, so suppress this "error" message since in that case
437 * it is normal. Or look at it the other way, only for a
438 * normal exit should we complain about this error.
440 if (jcr->JobStatus == JS_Terminated && jcr->jr.JobBytes) {
441 Jmsg(jcr, M_ERROR, 0, "%s", db_strerror(jcr->db));
443 jcr->VolumeName[0] = 0; /* none */
446 if (jcr->ReadBytes == 0) {
447 bstrncpy(compress, "None", sizeof(compress));
449 compression = (double)100 - 100.0 * ((double)jcr->JobBytes / (double)jcr->ReadBytes);
450 if (compression < 0.5) {
451 bstrncpy(compress, "None", sizeof(compress));
453 bsnprintf(compress, sizeof(compress), "%.1f %%", (float)compression);
456 jobstatus_to_ascii(jcr->FDJobStatus, fd_term_msg, sizeof(fd_term_msg));
457 jobstatus_to_ascii(jcr->SDJobStatus, sd_term_msg, sizeof(sd_term_msg));
459 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n\
462 Backup Level: %s%s\n\
464 FileSet: \"%s\" %s\n\
467 FD Files Written: %s\n\
468 SD Files Written: %s\n\
469 FD Bytes Written: %s\n\
470 SD Bytes Written: %s\n\
472 Software Compression: %s\n\
473 Volume name(s): %s\n\
474 Volume Session Id: %d\n\
475 Volume Session Time: %d\n\
476 Last Volume Bytes: %s\n\
477 Non-fatal FD errors: %d\n\
479 FD termination status: %s\n\
480 SD termination status: %s\n\
481 Termination: %s\n\n"),
485 level_to_str(jcr->JobLevel), since,
486 jcr->client->hdr.name,
487 jcr->fileset->hdr.name, fsr->cCreateTime,
490 edit_uint64_with_commas(jcr->jr.JobFiles, ec1),
491 edit_uint64_with_commas(jcr->SDJobFiles, ec4),
492 edit_uint64_with_commas(jcr->jr.JobBytes, ec2),
493 edit_uint64_with_commas(jcr->SDJobBytes, ec5),
499 edit_uint64_with_commas(mr.VolBytes, ec3),
506 Dmsg0(100, "Leave backup_cleanup()\n");