3 * Bacula Director -- verify.c -- responsible for running file verification
5 * Kern Sibbald, October MM
7 * Basic tasks done here:
9 * Open connection with File daemon and pass him commands
11 * When the File daemon sends the attributes, compare them to
18 Copyright (C) 2000-2005 Kern Sibbald
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,
39 #include "findlib/find.h"
41 /* Imported Global Variables */
42 extern int debug_level;
44 /* Commands sent to File daemon */
45 static char verifycmd[] = "verify level=%s\n";
46 static char storaddr[] = "storage address=%s port=%d ssl=0\n";
47 static char sessioncmd[] = "session %s %ld %ld %ld %ld %ld %ld\n";
49 /* Responses received from File daemon */
50 static char OKverify[] = "2000 OK verify\n";
51 static char OKstore[] = "2000 OK storage\n";
52 static char OKsession[] = "2000 OK session\n";
54 /* Forward referenced functions */
55 static void verify_cleanup(JCR *jcr, int TermCode);
56 static void prt_fname(JCR *jcr);
57 static int missing_handler(void *ctx, int num_fields, char **row);
60 * Do a verification of the specified files against the Catlaog
62 * Returns: false on failure
65 bool do_verify(JCR *jcr)
67 const char *level, *Name;
69 JOB_DBR jr, verify_jr;
70 JobId_t verify_jobid = 0;
73 memset(&verify_jr, 0, sizeof(verify_jr));
74 if (!jcr->verify_jr) {
75 jcr->verify_jr = &verify_jr;
78 Dmsg1(9, "bdird: created client %s record\n", jcr->client->hdr.name);
81 * Find JobId of last job that ran. E.g.
82 * for VERIFY_CATALOG we want the JobId of the last INIT.
83 * for VERIFY_VOLUME_TO_CATALOG, we want the JobId of the
86 if (jcr->JobLevel == L_VERIFY_CATALOG ||
87 jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
88 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG) {
89 memcpy(&jr, &jcr->jr, sizeof(jr));
90 if (jcr->verify_job &&
91 (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
92 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG)) {
93 Name = jcr->verify_job->hdr.name;
97 Dmsg1(100, "find last jobid for: %s\n", NPRT(Name));
98 if (!db_find_last_jobid(jcr, jcr->db, Name, &jr)) {
99 if (jcr->JobLevel == L_VERIFY_CATALOG) {
100 Jmsg(jcr, M_FATAL, 0, _(
101 "Unable to find JobId of previous InitCatalog Job.\n"
102 "Please run a Verify with Level=InitCatalog before\n"
103 "running the current Job.\n"));
105 Jmsg(jcr, M_FATAL, 0, _(
106 "Unable to find JobId of previous Job for this client.\n"));
110 verify_jobid = jr.JobId;
111 Dmsg1(100, "Last full jobid=%d\n", verify_jobid);
114 if (!db_update_job_start_record(jcr, jcr->db, &jcr->jr)) {
115 Jmsg(jcr, M_FATAL, 0, "%s", db_strerror(jcr->db));
119 /* Print Job Start message */
120 Jmsg(jcr, M_INFO, 0, _("Start Verify JobId=%d Level=%s Job=%s\n"),
121 jcr->JobId, level_to_str(jcr->JobLevel), jcr->Job);
124 * Now get the job record for the previous backup that interests
125 * us. We use the verify_jobid that we found above.
127 if (jcr->JobLevel == L_VERIFY_CATALOG ||
128 jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG ||
129 jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG) {
130 verify_jr.JobId = verify_jobid;
131 if (!db_get_job_record(jcr, jcr->db, &verify_jr)) {
132 Jmsg(jcr, M_FATAL, 0, _("Could not get job record for previous Job. ERR=%s"),
133 db_strerror(jcr->db));
136 if (verify_jr.JobStatus != 'T') {
137 Jmsg(jcr, M_FATAL, 0, _("Last Job %d did not terminate normally. JobStatus=%c\n"),
138 verify_jobid, verify_jr.JobStatus);
141 Jmsg(jcr, M_INFO, 0, _("Verifying against JobId=%d Job=%s\n"),
142 verify_jr.JobId, verify_jr.Job);
146 * If we are verifying a Volume, we need the Storage
147 * daemon, so open a connection, otherwise, just
148 * create a dummy authorization key (passed to
149 * File daemon but not used).
151 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
152 RBSR *bsr = new_bsr();
154 bsr->JobId = verify_jr.JobId;
155 ua = new_ua_context(jcr);
156 complete_bsr(ua, bsr);
157 bsr->fi = new_findex();
159 bsr->fi->findex2 = verify_jr.JobFiles;
160 jcr->ExpectedFiles = write_bsr_file(ua, bsr);
161 if (jcr->ExpectedFiles == 0) {
168 if (jcr->RestoreBootstrap) {
169 free(jcr->RestoreBootstrap);
171 POOLMEM *fname = get_pool_memory(PM_MESSAGE);
172 Mmsg(fname, "%s/restore.bsr", working_directory);
173 jcr->RestoreBootstrap = bstrdup(fname);
174 free_pool_memory(fname);
177 * Start conversation with Storage daemon
179 set_jcr_job_status(jcr, JS_Blocked);
180 if (!connect_to_storage_daemon(jcr, 10, SDConnectTimeout, 1)) {
184 * Now start a job with the Storage daemon
186 if (!start_storage_daemon_job(jcr, jcr->storage, SD_READ)) {
190 * Now start a Storage daemon message thread
192 if (!start_storage_daemon_message_thread(jcr)) {
195 Dmsg0(50, "Storage daemon connection OK\n");
197 jcr->sd_auth_key = bstrdup("dummy"); /* dummy Storage daemon key */
200 if (jcr->JobLevel == L_VERIFY_DISK_TO_CATALOG && jcr->verify_job) {
201 jcr->fileset = jcr->verify_job->fileset;
203 Dmsg2(100, "ClientId=%u JobLevel=%c\n", verify_jr.ClientId, jcr->JobLevel);
206 * OK, now connect to the File daemon
207 * and ask him for the files.
209 set_jcr_job_status(jcr, JS_Blocked);
210 if (!connect_to_file_daemon(jcr, 10, FDConnectTimeout, 1)) {
214 set_jcr_job_status(jcr, JS_Running);
215 fd = jcr->file_bsock;
218 Dmsg0(30, ">filed: Send include list\n");
219 if (!send_include_list(jcr)) {
223 Dmsg0(30, ">filed: Send exclude list\n");
224 if (!send_exclude_list(jcr)) {
229 * Send Level command to File daemon, as well
230 * as the Storage address if appropriate.
232 switch (jcr->JobLevel) {
236 case L_VERIFY_CATALOG:
239 case L_VERIFY_VOLUME_TO_CATALOG:
241 * send Storage daemon address to the File daemon
243 if (jcr->store->SDDport == 0) {
244 jcr->store->SDDport = jcr->store->SDport;
246 bnet_fsend(fd, storaddr, jcr->store->address, jcr->store->SDDport);
247 if (!response(jcr, fd, OKstore, "Storage", DISPLAY_ERROR)) {
252 * Send the bootstrap file -- what Volumes/files to restore
254 if (!send_bootstrap_file(jcr)) {
259 * The following code is deprecated
261 if (!jcr->RestoreBootstrap) {
263 * Pass the VolSessionId, VolSessionTime, Start and
264 * end File and Blocks on the session command.
266 bnet_fsend(fd, sessioncmd,
268 jr.VolSessionId, jr.VolSessionTime,
269 jr.StartFile, jr.EndFile, jr.StartBlock,
271 if (!response(jcr, fd, OKsession, "Session", DISPLAY_ERROR)) {
280 case L_VERIFY_DISK_TO_CATALOG:
281 level="disk_to_catalog";
284 Jmsg2(jcr, M_FATAL, 0, _("Unimplemented Verify level %d(%c)\n"), jcr->JobLevel,
289 if (!send_run_before_and_after_commands(jcr)) {
294 * Send verify command/level to File daemon
296 bnet_fsend(fd, verifycmd, level);
297 if (!response(jcr, fd, OKverify, "Verify", DISPLAY_ERROR)) {
302 * Now get data back from File daemon and
303 * compare it to the catalog or store it in the
304 * catalog depending on the run type.
306 /* Compare to catalog */
307 switch (jcr->JobLevel) {
308 case L_VERIFY_CATALOG:
309 Dmsg0(10, "Verify level=catalog\n");
310 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
311 jcr->SDJobStatus = JS_Terminated;
312 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
315 case L_VERIFY_VOLUME_TO_CATALOG:
316 Dmsg0(10, "Verify level=volume\n");
317 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
320 case L_VERIFY_DISK_TO_CATALOG:
321 Dmsg0(10, "Verify level=disk_to_catalog\n");
322 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
323 jcr->SDJobStatus = JS_Terminated;
324 get_attributes_and_compare_to_catalog(jcr, verify_jobid);
329 Dmsg0(10, "Verify level=init\n");
330 jcr->sd_msg_thread_done = true; /* no SD msg thread, so it is done */
331 jcr->SDJobStatus = JS_Terminated;
332 get_attributes_and_put_in_catalog(jcr);
336 Jmsg1(jcr, M_FATAL, 0, _("Unimplemented verify level %d\n"), jcr->JobLevel);
340 stat = wait_for_job_termination(jcr);
341 verify_cleanup(jcr, stat);
345 verify_cleanup(jcr, JS_ErrorTerminated);
350 * Release resources allocated during backup.
353 static void verify_cleanup(JCR *jcr, int TermCode)
355 char sdt[50], edt[50];
356 char ec1[30], ec2[30];
357 char term_code[100], fd_term_msg[100], sd_term_msg[100];
358 const char *term_msg;
363 // Dmsg1(100, "Enter verify_cleanup() TermCod=%d\n", TermCode);
364 dequeue_messages(jcr); /* display any queued messages */
366 Dmsg3(900, "JobLevel=%c Expected=%u JobFiles=%u\n", jcr->JobLevel,
367 jcr->ExpectedFiles, jcr->JobFiles);
368 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG &&
369 jcr->ExpectedFiles != jcr->JobFiles) {
370 TermCode = JS_ErrorTerminated;
373 JobId = jcr->jr.JobId;
374 set_jcr_job_status(jcr, TermCode);
376 update_job_end_record(jcr);
378 msg_type = M_INFO; /* by default INFO message */
381 term_msg = _("Verify OK");
384 case JS_ErrorTerminated:
385 term_msg = _("*** Verify Error ***");
386 msg_type = M_ERROR; /* Generate error message */
389 term_msg = _("Verify warnings");
392 term_msg = _("Verify Canceled");
395 term_msg = _("Verify Differences");
398 term_msg = term_code;
399 bsnprintf(term_code, sizeof(term_code),
400 _("Inappropriate term code: %d %c\n"), TermCode, TermCode);
403 bstrftimes(sdt, sizeof(sdt), jcr->jr.StartTime);
404 bstrftimes(edt, sizeof(edt), jcr->jr.EndTime);
405 if (jcr->verify_job) {
406 Name = jcr->verify_job->hdr.name;
411 jobstatus_to_ascii(jcr->FDJobStatus, fd_term_msg, sizeof(fd_term_msg));
412 if (jcr->JobLevel == L_VERIFY_VOLUME_TO_CATALOG) {
413 jobstatus_to_ascii(jcr->SDJobStatus, sd_term_msg, sizeof(sd_term_msg));
414 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n"
418 " Verify Level: %s\n"
420 " Verify JobId: %d\n"
424 " Files Expected: %s\n"
425 " Files Examined: %s\n"
426 " Non-fatal FD errors: %d\n"
427 " FD termination status: %s\n"
428 " SD termination status: %s\n"
429 " Termination: %s\n\n"),
433 jcr->fileset->hdr.name,
434 level_to_str(jcr->JobLevel),
435 jcr->client->hdr.name,
436 jcr->verify_jr->JobId,
440 edit_uint64_with_commas(jcr->ExpectedFiles, ec1),
441 edit_uint64_with_commas(jcr->JobFiles, ec2),
447 Jmsg(jcr, msg_type, 0, _("Bacula " VERSION " (" LSMDATE "): %s\n"
451 " Verify Level: %s\n"
453 " Verify JobId: %d\n"
457 " Files Examined: %s\n"
458 " Non-fatal FD errors: %d\n"
459 " FD termination status: %s\n"
460 " Termination: %s\n\n"),
464 jcr->fileset->hdr.name,
465 level_to_str(jcr->JobLevel),
466 jcr->client->hdr.name,
467 jcr->verify_jr->JobId,
471 edit_uint64_with_commas(jcr->JobFiles, ec1),
476 Dmsg0(100, "Leave verify_cleanup()\n");
478 free_memory(jcr->fname);
484 * This routine is called only during a Verify
486 int get_attributes_and_compare_to_catalog(JCR *jcr, JobId_t JobId)
491 struct stat statf; /* file stat */
492 struct stat statc; /* catalog stat */
493 int stat = JS_Terminated;
495 POOLMEM *fname = get_pool_memory(PM_MESSAGE);
497 int32_t file_index = 0;
499 memset(&fdbr, 0, sizeof(FILE_DBR));
500 fd = jcr->file_bsock;
504 Dmsg0(20, "bdird: waiting to receive file attributes\n");
506 * Get Attributes and Signature from File daemon
510 * Options or SIG (MD5/SHA1)
515 while ((n=bget_dirmsg(fd)) >= 0 && !job_canceled(jcr)) {
518 char Opts_SIG[MAXSTRING]; /* Verify Opts or MD5/SHA1 signature */
520 fname = check_pool_memory_size(fname, fd->msglen);
521 jcr->fname = check_pool_memory_size(jcr->fname, fd->msglen);
522 Dmsg1(200, "Atts+SIG=%s\n", fd->msg);
523 if ((len = sscanf(fd->msg, "%ld %d %100s", &file_index, &stream,
525 Jmsg3(jcr, M_FATAL, 0, _("bird<filed: bad attributes, expected 3 fields got %d\n"
526 " mslen=%d msg=%s\n"), len, fd->msglen, fd->msg);
530 * We read the Options or Signature into fname
531 * to prevent overrun, now copy it to proper location.
533 bstrncpy(Opts_SIG, fname, sizeof(Opts_SIG));
535 skip_nonspaces(&p); /* skip FileIndex */
537 skip_nonspaces(&p); /* skip Stream */
539 skip_nonspaces(&p); /* skip Opts_SIG */
540 p++; /* skip space */
543 *fn++ = *p++; /* copy filename */
545 *fn = *p++; /* term filename and point to attribs */
548 * Got attributes stream, decode it
550 if (stream == STREAM_UNIX_ATTRIBUTES || stream == STREAM_UNIX_ATTRIBUTES_EX) {
551 int32_t LinkFIf, LinkFIc;
552 Dmsg2(400, "file_index=%d attr=%s\n", file_index, attr);
554 jcr->FileIndex = file_index; /* remember attribute file_index */
555 decode_stat(attr, &statf, &LinkFIf); /* decode file stat packet */
557 jcr->fn_printed = false;
558 pm_strcpy(jcr->fname, fname); /* move filename into JCR */
560 Dmsg2(040, "dird<filed: stream=%d %s\n", stream, jcr->fname);
561 Dmsg1(020, "dird<filed: attr=%s\n", attr);
564 * Find equivalent record in the database
567 if (!db_get_file_attributes_record(jcr, jcr->db, jcr->fname,
568 jcr->verify_jr, &fdbr)) {
569 Jmsg(jcr, M_INFO, 0, _("New file: %s\n"), jcr->fname);
570 Dmsg1(020, _("File not in catalog: %s\n"), jcr->fname);
571 stat = JS_Differences;
575 * mark file record as visited by stuffing the
576 * current JobId, which is unique, into the MarkId field.
578 db_mark_file_record(jcr, jcr->db, fdbr.FileId, jcr->JobId);
581 Dmsg3(400, "Found %s in catalog. inx=%d Opts=%s\n", jcr->fname,
582 file_index, Opts_SIG);
583 decode_stat(fdbr.LStat, &statc, &LinkFIc); /* decode catalog stat */
585 * Loop over options supplied by user and verify the
586 * fields he requests.
588 for (p=Opts_SIG; *p; p++) {
589 char ed1[30], ed2[30];
591 case 'i': /* compare INODEs */
592 if (statc.st_ino != statf.st_ino) {
594 Jmsg(jcr, M_INFO, 0, _(" st_ino differ. Cat: %s File: %s\n"),
595 edit_uint64((uint64_t)statc.st_ino, ed1),
596 edit_uint64((uint64_t)statf.st_ino, ed2));
597 stat = JS_Differences;
600 case 'p': /* permissions bits */
601 if (statc.st_mode != statf.st_mode) {
603 Jmsg(jcr, M_INFO, 0, _(" st_mode differ. Cat: %x File: %x\n"),
604 (uint32_t)statc.st_mode, (uint32_t)statf.st_mode);
605 stat = JS_Differences;
608 case 'n': /* number of links */
609 if (statc.st_nlink != statf.st_nlink) {
611 Jmsg(jcr, M_INFO, 0, _(" st_nlink differ. Cat: %d File: %d\n"),
612 (uint32_t)statc.st_nlink, (uint32_t)statf.st_nlink);
613 stat = JS_Differences;
616 case 'u': /* user id */
617 if (statc.st_uid != statf.st_uid) {
619 Jmsg(jcr, M_INFO, 0, _(" st_uid differ. Cat: %u File: %u\n"),
620 (uint32_t)statc.st_uid, (uint32_t)statf.st_uid);
621 stat = JS_Differences;
624 case 'g': /* group id */
625 if (statc.st_gid != statf.st_gid) {
627 Jmsg(jcr, M_INFO, 0, _(" st_gid differ. Cat: %u File: %u\n"),
628 (uint32_t)statc.st_gid, (uint32_t)statf.st_gid);
629 stat = JS_Differences;
633 if (statc.st_size != statf.st_size) {
635 Jmsg(jcr, M_INFO, 0, _(" st_size differ. Cat: %s File: %s\n"),
636 edit_uint64((uint64_t)statc.st_size, ed1),
637 edit_uint64((uint64_t)statf.st_size, ed2));
638 stat = JS_Differences;
641 case 'a': /* access time */
642 if (statc.st_atime != statf.st_atime) {
644 Jmsg(jcr, M_INFO, 0, _(" st_atime differs\n"));
645 stat = JS_Differences;
649 if (statc.st_mtime != statf.st_mtime) {
651 Jmsg(jcr, M_INFO, 0, _(" st_mtime differs\n"));
652 stat = JS_Differences;
655 case 'c': /* ctime */
656 if (statc.st_ctime != statf.st_ctime) {
658 Jmsg(jcr, M_INFO, 0, _(" st_ctime differs\n"));
659 stat = JS_Differences;
662 case 'd': /* file size decrease */
663 if (statc.st_size > statf.st_size) {
665 Jmsg(jcr, M_INFO, 0, _(" st_size decrease. Cat: %s File: %s\n"),
666 edit_uint64((uint64_t)statc.st_size, ed1),
667 edit_uint64((uint64_t)statf.st_size, ed2));
668 stat = JS_Differences;
671 case '5': /* compare MD5 */
672 Dmsg1(500, "set Do_MD5 for %s\n", jcr->fname);
675 case '1': /* compare SHA1 */
685 * Got SIG Signature from Storage daemon
686 * It came across in the Opts_SIG field.
688 } else if (stream == STREAM_MD5_SIGNATURE || stream == STREAM_SHA1_SIGNATURE) {
689 Dmsg2(400, "stream=SIG inx=%d SIG=%s\n", file_index, Opts_SIG);
691 * When ever we get a signature is MUST have been
692 * preceded by an attributes record, which sets attr_file_index
694 if (jcr->FileIndex != (uint32_t)file_index) {
695 Jmsg2(jcr, M_FATAL, 0, _("MD5/SHA1 index %d not same as attributes %d\n"),
696 file_index, jcr->FileIndex);
700 db_escape_string(buf, Opts_SIG, strlen(Opts_SIG));
701 if (strcmp(buf, fdbr.SIG) != 0) {
703 if (debug_level >= 10) {
704 Jmsg(jcr, M_INFO, 0, _(" %s not same. File=%s Cat=%s\n"),
705 stream==STREAM_MD5_SIGNATURE?"MD5":"SHA1", buf, fdbr.SIG);
707 Jmsg(jcr, M_INFO, 0, _(" %s differs.\n"),
708 stream==STREAM_MD5_SIGNATURE?"MD5":"SHA1");
710 stat = JS_Differences;
715 jcr->JobFiles = file_index;
717 if (is_bnet_error(fd)) {
719 Jmsg2(jcr, M_FATAL, 0, _("bdird<filed: bad attributes from filed n=%d : %s\n"),
724 /* Now find all the files that are missing -- i.e. all files in
725 * the database where the MarkedId != current JobId
727 jcr->fn_printed = false;
728 bsnprintf(buf, sizeof(buf),
729 "SELECT Path.Path,Filename.Name FROM File,Path,Filename "
730 "WHERE File.JobId=%d "
731 "AND File.MarkedId!=%d AND File.PathId=Path.PathId "
732 "AND File.FilenameId=Filename.FilenameId",
734 /* missing_handler is called for each file found */
735 db_sql_query(jcr->db, buf, missing_handler, (void *)jcr);
736 if (jcr->fn_printed) {
737 stat = JS_Differences;
739 free_pool_memory(fname);
740 set_jcr_job_status(jcr, stat);
744 free_pool_memory(fname);
745 set_jcr_job_status(jcr, JS_ErrorTerminated);
750 * We are called here for each record that matches the above
751 * SQL query -- that is for each file contained in the Catalog
752 * that was not marked earlier. This means that the file in
753 * question is a missing file (in the Catalog but not on Disk).
755 static int missing_handler(void *ctx, int num_fields, char **row)
757 JCR *jcr = (JCR *)ctx;
759 if (!jcr->fn_printed) {
760 Jmsg(jcr, M_INFO, 0, "\n");
761 Jmsg(jcr, M_INFO, 0, _("The following files are missing:\n"));
762 jcr->fn_printed = true;
764 Jmsg(jcr, M_INFO, 0, " %s%s\n", row[0]?row[0]:"", row[1]?row[1]:"");
770 * Print filename for verify
772 static void prt_fname(JCR *jcr)
774 if (!jcr->fn_printed) {
775 Jmsg(jcr, M_INFO, 0, _("File: %s\n"), jcr->fname);
776 jcr->fn_printed = TRUE;