4 Documentation to do: (a little bit at a time)
5 - Document running a test version.
6 - Document query file format.
7 - Document static linking
9 Testing to do: (painful)
10 - that console command line options work
11 - blocksize recognition code.
14 - Look at ua_prune.c in detail. Why did JobType work at all??????
15 - Figure out how to allow multiple simultaneous file Volumes on
17 - Why are save/restore of device different sizes (sparse?)
18 - Make sure differential handled correctly
21 - Cancel does not work for restore in FD.
22 - Restore to a particular time -- e.g. before date, after date.
24 - Implement disk spooling
25 - Implement finer multiprocessing options.
26 - Solaris -I on tar for include list
27 - Add whereabslinks to where or not where absolute links
28 - Figure out some way to automatically backup all local partitions
29 - Enable avoid backing up archive device (findlib/find_one.c:128)
30 - Implement FileOptions (see end of this document)
31 - Implement Bacula plugins -- design API
32 - Make hash table for linked files in findlib/find_one.c:161
33 - Make bcopy read through bad tape records.
34 - Need a verbose mode in restore, perhaps to bsr.
35 - bscan without -v is too quiet -- perhaps show jobs.
36 - Add code to reject whole blocks if not wanted on restore.
37 - Implement multiple simultaneous file Volumes on a single device.
38 - Start working on Base jobs.
39 - Make sure the MaxVolFiles is fully implemented in SD
40 - Flush all the daemon messages at the end of every job.
41 - Check if both CatalogFiles and UseCatalog are set to SD.
42 - Check if we can increase Bacula FD priorty in Win2000
43 - Need return status on read_cb() from read_records(). Need multiple
44 records -- one per Job, maybe a JCR or some other structure with
46 - Figure out how to do a bare metal Windows restore
47 - Fix read_record to handle multiple sessions.
48 - Program files (i.e. execute a program to read/write files).
49 Pass read date of last backup, size of file last time.
50 - Put system type returned by FD into catalog.
51 - Possibly add email to Watchdog if drive is unmounted too
52 long and a job is waiting on the drive.
53 - Strip trailing slashes from Include directory names in the FD.
54 - Use read_record.c in SD code.
55 - Why don't we get an error message from Win32 FD when bootstrap
56 file cannot be created for restore command?
57 - Need to specify MaximumConcurrentJobs in the Job resource.
58 - When Marking a file in Restore that is a hard link, also
59 mark the link so that the data will be reloaded.
60 - Restore program that errors in SD due to no tape reports
61 OK incorrectly in output.
62 - After unmount, if restore job started, ask to mount.
63 - Make Restore report an error if FD or SD term codes are not OK.
64 - Convert all %x substitution variables, which are hard to remember
65 and read to %(variable-name). Idea from TMDA.
66 - Add JobLevel in FD status (but make sure it is defined).
67 - Make Pool resource handle Counter resources.
68 - Remove NextId for SQLite. Optimize.
69 - Strip trailing / from Include
70 - Move all SQL statements into a single location.
71 - Cleanup db_update_media and db_update_pool
72 - Add UA rc and history files.
73 - put termcap (used by console) in ./configure and
74 allow -with-termcap-dir.
75 - Enhance time and size scanning routines.
76 - Fix Autoprune for Volumes to respect need for full save.
77 - Fix Win32 config file definition name on /install
78 - No READLINE_SRC if found in alternate directory.
79 - Add Client FS/OS id (Linux, Win95/98, ...).
80 - Test a second language e.g. french.
81 - Compare tape to Client files (attributes, or attributes and data)
82 - Make all database Ids 64 bit.
83 - Write an applet for Linux.
84 - Add estimate to Console commands
85 - Find solution to blank filename (i.e. path only) problem.
86 - Implement new daemon communications protocol.
87 - Remove PoolId from Job table, it exists in Media.
88 - Allow console commands to detach or run in background.
89 - Fix status delay on storage daemon during rewind.
90 - Add SD message variables to control operator wait time
91 - Maximum Operator Wait
92 - Minimum Message Interval
93 - Maximum Message Interval
94 - Send Operator message when cannot read tape label.
95 - Verify level=Volume (scan only), level=Data (compare of data to file).
96 Verify level=Catalog, level=InitCatalog
98 - Add keyword search to show command in Console.
99 - Fix Win2000 error with no messages during startup.
100 - Events : tape has more than xxx bytes.
101 - Restrict characters permitted in a Resource name.
102 - Complete code in Bacula Resources -- this will permit
103 reading a new config file at any time.
104 - Handle ctl-c in Console
105 - Implement LabelTemplate (at least first cut).
106 - Implement script driven addition of File daemon to config files.
107 - Think about how to make Bacula work better with File (non-tape) archives.
109 - Implement new serialize subroutines
110 send(socket, "string", &Vol, "uint32", &i, NULL)
111 - Audit all UA commands to ensure that we always prompt where possible.
112 - If ./btape is called without /dev, assume argument is a Storage resource name.
113 - Put memory utilization in Status output of each daemon
114 if full status requested or if some level of debug on.
115 - Make database type selectable by .conf files i.e. at runtime
116 - gethostbyname failure in bnet_connect() continues
117 generating errors -- should stop.
118 - Set flag for uname -a. Add to Volume label.
119 - Implement throttled work queue.
120 - Check for EOT at ENOSPC or EIO or ENXIO (unix Pc)
121 - Allow multiple Storage specifications (or multiple names on
122 a single Storage specification) in the Job record. Thus a job
123 can be backed up to a number of storage devices.
124 - Implement dump label to UA
125 - Concept of VolumeSet during restore which is a list
126 of Volume names needed.
127 - Restore files modified after date
128 - Restore file modified before date
129 - Emergency restore info:
131 - Backup working directory
133 - Restore -- do nothing but show what would happen
134 - SET LD_RUN_PATH=$HOME/mysql/lib/mysql
135 - Implement Restore FileSet=
136 - Create a protocol.h and protocol.c where all protocol messages
138 - If SD cannot open a drive, make it periodically retry.
139 - Remove duplicate fields from jcr (e.g. jcr.level and jcr.jr.Level, ...).
140 - Timout a job or terminate if link goes down, or reopen link and query.
141 - Find general solution for sscanf size problems (as well
142 as sprintf. Do at run time?
143 - Concept of precious tapes (cannot be reused).
144 - Make bcopy copy with a single tape drive.
145 - Permit changing ownership during restore.
147 - Autolabel should be specified by DIR instead of SD.
148 - Find out how to get the system tape block limits, e.g.:
149 Apr 22 21:22:10 polymatou kernel: st1: Block limits 1 - 245760 bytes.
150 Apr 22 21:22:10 polymatou kernel: st0: Block limits 2 - 16777214 bytes.
153 - AutoScan (check checksum of tape)
154 - Format command = "format /dev/nst0"
158 - Seek resolution (usually corresponds to buffer size)
159 - EODErrorCode=ENOSPC or code
160 - Partial Read error code
161 - Partial write error code
162 - Nonformatted read error
163 - Nonformatted write error
164 - WriteProtected error
168 - IgnoreCloseErrors=yes
178 - FD sends unsaved file list to Director at end of job (see
180 - File daemon should build list of files skipped, and then
181 at end of save retry and report any errors.
182 - Write a Storage daemon that uses pipes and
183 standard Unix programs to write to the tape.
185 - Need something that monitors the JCR queue and
186 times out jobs by asking the deamons where they are.
187 - Enhance Jmsg code to permit buffering and saving to disk.
188 - device driver = "xxxx" for drives.
189 - restart: paranoid: read label fsf to
190 eom read append block, and go
191 super-paranoid: read label, read all files
192 in between, read append block, and go
193 verify: backspace, read append block, and go
194 permissive: same as above but frees drive
195 if tape is not valid.
197 - Ensure that /dev/null works
198 - Need report class for messages. Perhaps
199 report resource where report=group of messages
200 - enhance scan_attrib and rename scan_jobtype, and
201 fill in code for "since" option
202 - Director needs a time after which the report status is sent
203 anyway -- or better yet, a retry time for the job.
204 Don't reschedule a job if previous incarnation is still running.
205 - Some way to automatically backup everything is needed????
206 - Need a structure for pending actions:
208 - termination status (part of buffered msgs?)
209 - Concept of grouping Storage devices and job can use
210 any of a number of devices
212 Read, Write, Clean, Delete
213 - Login to Bacula; Bacula users with different permissions:
214 owner, group, user, quotas
215 - Store info on each file system type (probably in the job header on tape.
216 This could be the output of df; or perhaps some sort of /etc/mtab record.
219 - Design at hierarchial storage for Bacula.
220 - Implement FSM (File System Modules).
221 - Identify unchanged or "system" files and save them to a
222 special tape thus removing them from the standard
223 backup FileSet -- BASE backup.
224 - Turn virutally all sprintfs into snprintfs.
225 - Heartbeat between daemons.
226 - Audit M_ error codes to ensure they are correct and consistent.
227 - Add variable break characters to lex analyzer.
228 Either a bit mask or a string of chars so that
229 the caller can change the break characters.
230 - Make a single T_BREAK to replace T_COMMA, etc.
231 - Ensure that File daemon and Storage daemon can
232 continue a save if the Director goes down (this
233 is NOT currently the case). Must detect socket error,
234 buffer messages for later.
235 - Enhance time/duration input to allow multiple qualifiers e.g. 3d2h
239 Bacula Projects Roadmap
241 last update 27 November 2002
243 Item 1: Multiple simultaneous Jobs. (done)
244 Done -- Restore part needs better implementation to work correctly
246 What: Permit multiple simultaneous jobs in Bacula.
248 Why: An enterprise level solution needs to go fast without the
249 need for the system administrator to carefully tweak
250 timing. Based on the benchmarks, during a full
251 backup, NetWorker typically hit 10 times the bandwidth to
252 the tape compared to Bacula--largely. This is probably due to
253 running parallel jobs and multi-threaded filling of buffers
254 and writing them to tape. This should also make things work
255 better when you have a mix of fast and slow machines backing
258 Notes: Bacula was designed to run multiple simultaneous jobs. Thus
259 implementing this is a matter of some small cleanups and
263 Item 2: Make the Storage daemon use intermediate file storage to buffer data.
264 Deferred -- not necessary yet.
266 What: If data is coming into the SD too fast, buffer it to
267 disk if the user has configured this option.
269 Why: This would be nice, especially if it more or less falls out
270 when implementing (1) above. If not, it probably should not
271 be given a high priority because fundamentally the backup time
272 is limited by the tape bandwidth. Even though you may finish a
273 client job quicker by spilling to disk, you still have to
274 eventually get it onto tape. If intermediate disk buffering
275 allows us to improve write bandwidth to tape, it may make
278 Notes: Whether or not this is implemented will depend upon performance
279 testing after item 1 is implemented.
282 Item 3: Write the bscan program -- also write a bcopy program.
285 What: Write a program that reads a Bacula tape and puts all the
286 appropriate data into the catalog. This allows recovery
287 from a tape that is no longer in the database, or it allows
288 re-creation of a database if lost.
290 Why: This is a fundamental robustness and disaster recovery tool
291 which will increase the comfort level of a sysadmin
292 considering adopting Bacula.
294 Notes: A skeleton of this program already exists, but much work
295 needs to be done. Implementing this will also make apparent
296 any deficiencies in the current Bacula tape format.
299 Item 4: Implement Base jobs.
301 What: A base job is sort of like a Full save except that you
302 will want the FileSet to contain only files that are unlikely
303 to change in the future (i.e. a snapshot of most of your
304 system after installing it). After the base job has been run,
305 when you are doing a Full save, you can specify to exclude
306 all files saved by the base job that have not been modified.
308 Why: This is something none of the competition does, as far as we know
309 (except BackupPC, which is a Perl program that saves to disk
310 only). It is big win for the user, it makes Bacula stand out
311 as offering a unique optimization that immediately saves time
314 Notes: Big savings in tape usage. Will require more resources because
315 the e. DIR must send FD a list of files/attribs, and the FD must
316 search the list and compare it for each file to be saved.
319 Item 5: Implement Label templates
321 What: This is a mechanism whereby Bacula can automatically create
322 a tape label for new tapes according to a detailed specification
323 provided by the user.
325 Why: It is a major convenience item for folks who use automated label
328 Notes: Bacula already has a working form of automatic tape label
329 creation, but it is very crude. The design for the complete
330 tape labeling project is already documented in the manual.
333 Item 6: Write a regression script.
336 What: This is an automatic script that runs and tests as many features
337 of Bacula as possible. The output is compared to previous
338 versions of Bacula and any differences are reported.
340 Why: This is an enormous help in preventing introduction of new
341 errors in parts of the program that already work correctly.
343 Notes: This probably should be ranked higher, it's something the typical
344 user doesn't see. Depending on how it's implemented, it may
345 make sense to defer it until the archival tape format and
346 user interface mature.
349 Item 7: GUI for interactive restore
350 Item 8: GUI for interactive backup
352 What: The current interactive restore is implemented with a tty
353 interface. It would be much nicer to be able to "see" the
354 list of files backed up in typical GUI tree format.
355 The same mechanism could also be used for creating
356 ad-hoc backup FileSets (item 8).
358 Why: Ease of use -- especially for the end user.
360 Notes: Rather than implementing in Gtk, we probably should go directly
361 for a Browser implementation, even if doing so meant the
362 capability wouldn't be available until much later. Not only
363 is there the question of Windows sites, most
364 Solaris/HP/IRIX, etc, shops can't currently run Gtk programs
365 without installing lots of stuff admins are very wary about.
366 Real sysadmins will always use the command line anyway, and
367 the user who's doing an interactive restore or backup of his
368 own files will in most cases be on a Windows machine running
372 Item 9: Add SSL to daemon communications.
374 What: This provides for secure communications between the daemons.
376 Why: This would allow doing backup across the Internet without
377 privacy concerns (or with much less concern).
379 Notes: The vast majority of near term potential users will be backing up
380 a single site over a LAN and, correctly or not, they probably
381 won't be concerned with security, at least not enough to go to
382 the trouble to set up keys, etc. to screw things down. We suspect
383 that many users genuinely interested in multi-site backup
384 already run some form of VPN software in their internetwork
385 connections, and are willing to delegate security to that layer.
388 Item 10: Define definitive tape format.
391 What: Define that definitive tape format that will not change
392 for the next millennium.
394 Why: Stability, security.
396 Notes: See notes for item 11 below.
399 Item 11: New daemon communication protocol.
401 What: The current daemon to daemon protocol is basically an ASCII
402 printf() and sending the buffer. On the receiving end, the
403 buffer is sscanf()ed to unpack it. The new scheme would
404 be a binary format that allows quick packing and unpacking
405 of any data type with named fields.
407 Why: Using binary packing would be faster. Named fields will permit
408 error checking to ensure that what is sent is what the
409 receiver really wants.
411 Notes: These are internal improvements in the interest of the
412 long-term stability and evolution of the program. On the one
413 hand, the sooner they're done, the less code we have to rip
414 up when the time comes to install them. On the other hand, they
415 don't bring an immediately perceptible benefit to potential
416 users. Item 10 and possibly item 11 should be deferred until Bacula
417 is well established with a growing user community more or
418 less happy with the feature set. At that time, it will make a
419 good "next generation" upgrade in the interest of data
425 ======================================================
427 It is somewhat like a Full save becomes an incremental since
428 the Base job (or jobs) plus other non-base files.
430 - New BaseFile table that contains:
431 JobId, BaseJobId, FileId (from Base).
432 i.e. for each base file that exists but is not saved because
433 it has not changed, the File daemon sends the JobId, BaseId,
434 and FileId back to the Director who creates the DB entry.
435 - To initiate a Base save, the Director sends the FD
436 the FileId, and full filename for each file in the Base.
437 - When the FD finds a Base file, he requests the Director to
438 send him the full File entry (stat packet plus MD5), or
439 conversely, the FD sends it to the Director and the Director
440 says yes or no. This can be quite rapid if the FileId is kept
441 by the FD for each Base Filename.
442 - It is probably better to have the comparison done by the FD
443 despite the fact that the File entry must be sent across the
445 - An alternative would be to send the FD the whole File entry
446 from the start. The disadvantage is that it requires a lot of
447 space. The advantage is that it requires less communications
449 - The Job record must be updated to indicate that one or more
451 - At end of Job, FD returns:
452 1. Count of base files/bytes not written to tape (i.e. matches)
453 2. Count of base file that were saved i.e. had changed.
454 - No tape record would be written for a Base file that matches, in the
455 same way that no tape record is written for Incremental jobs where
456 the file is not saved because it is unchanged.
457 - On a restore, all the Base file records must explicitly be
458 found from the BaseFile tape. I.e. for each Full save that is marked
459 to have one or more Base Jobs, search the BaseFile for all occurrences
461 - An optimization might be to make the BaseFile have:
467 This would avoid the need to explicitly fetch each File record for
468 the Base job. The Base Job record will be fetched to get the
469 VolSessionId and VolSessionTime.
470 =========================================================
472 ==========================================================
474 For each Incremental job that is run, there may be files that
475 were found but not saved because they were locked (this applies
476 only to Windows). Such a system could send back to the Director
477 a list of Unsaved files.
479 - New UnSavedFiles table that contains:
483 - Then in the next Incremental job, the list of Unsaved Files will be
484 feed to the FD, who will ensure that they are explicitly chosen even
485 if standard date/time check would not have selected them.
486 =============================================================
490 =============================================================
492 Request For Comments For File Backup Options
495 Subject: File Backup Options
498 A few days ago, a Bacula user who is backing up to file volumes and
499 using compression asked if it was possible to suppress compressing
500 all .gz files since it was a waste of CPU time. Although Bacula
501 currently permits using different options (compression, ...) on
502 a directory by directory basis, it cannot do it on a file by
503 file basis, which is clearly what was desired.
505 Proposed Implementation:
506 To solve this problem, I propose the following:
508 - Add a new Director resource type called FileOptions.
510 - The FileOptions resource will have records for all
511 options that can currently be specified on the Include record
512 (in a FileSet). Examples below.
514 - The FileOptions resource will permit an exclude option as well
515 as a number of additional options.
517 - The heart of the FileOptions resource is the ability to
518 supply any number of ApplyTo records which specify POSIX
519 regular expressions. These ApplyTo regular expressions are
520 applied to the fully qualified filename (path and all). If
521 one matches, then the FileOptions will be used.
523 - When an ApplyTo specification matches an included file, the
524 options specified in the FileOptions resource will override
525 the default options specified on the Include record.
527 - Include records will be modified to permit referencing one or
528 more FileOptions resources. The FileOptions will be used
529 in the order listed on the Include record and the first
530 one that matches will be applied.
532 - Options (or specifications) currently supplied on the Include
533 record will be deprecated (i.e. removed in a later version a
534 year or so from now).
536 - The Exclude record will be deprecated as the same functionality
537 can be obtained by using an Exclude = yes in the FileOptions.
540 The following records can appear in the FileOptions resource. An
541 asterisk preceding the name indicates a feature not currently
545 - Compression= (GZIP, ...)
546 - Signature= (MD5, SHA1, ...)
548 - OneFs= (yes/no) - remain on one filesystem
549 - Recurse= (yes/no) - recurse into subdirectories
550 - Sparse= (yes/no) - do sparse file backup
551 - *Exclude= (yes/no) - exclude file from being saved
552 - *Reader= (filename) - external read (backup) program
553 - *Plugin= (filename) - read/write plugin module
556 - verify= (ipnougsamc5) - verify options
559 - replace= (always/ifnewer/ifolder/never) - replace options currently
561 - *Writer= (filename) - external write (restore) program
565 Currently options specifying compression, MD5 signatures, recursion,
566 ... of a FileSet are supplied on the Include record. These will now
567 all be collected into a FileOptions resource, which will be
568 specified on the Include in place of the options. Multiple FileOptions
569 may be specified. Since the FileOptions contain regular expressions
570 that are applied to the full filename, this will give the ability
571 to specify backup options on a file by file basis to whatever level
580 Include = compression=GZIP signature=MD5 {
589 Include = FileOptions=Opts {
600 That's a lot more to do the same thing, but it gives the ability to
601 apply options on a file by file basis. For example, suppose you
602 want to compress all files but not any file with extensions .gz or .Z.
603 You could do so as follows:
607 Include = FileOptions=NoCompress FileOptions=Opts {
615 ApplyTo = /*.?*/ # matches all files
620 # Note multiple ApplyTos are ORed
621 ApplyTo = /*.gz/ # matches .gz files */
622 ApplyTo = /*.Z/ # matches .Z files */
625 Now, since the NoCompress FileOptions is specified first on the
626 Include line, any *.gz or *.Z file will have an MD5 signature computed,
627 but will not be compressed. For all other files, the NoCompress will not
628 match, so the Opts options will be used which will include GZIP
632 - Is it necessary to provide some means of ANDing regular expressions
633 and negation? (not currently planned)
635 e.g. ApplyTo = /*.gz/ && !/big.gz/
637 - I see that Networker has a "null" module which, if specified, does not
638 backup the file, but does make an record of the file in the catalog
639 so that the catalog will reflect an exact picture of the filesystem.
640 The result is that the file can be "seen" when "browsing" the save
641 sets, but it cannot be restored.
643 Is this really useful? Should it be implemented in Bacula?
646 After implementing the above, the user will be able to specify
647 on a file by file basis (using regular expressions) what options are
648 applied for the backup.
649 ====================================
651 Done: (see kernsdone for more)
652 - Add EOM records? No, not at this time. The current system works and
654 - Add VolumeUseDuration and MaximumVolumeJobs to Pool db record and
656 - Add VOLUME_CAT_INFO to the EOS tape record (as
657 well as to the EOD record). -- No, not at this time.
658 - Put MaximumVolumeSize in Director (MaximumVolumeJobs, MaximumVolumeFiles,
660 - Enhance schedule to have 1stSat, ...
661 - Make sure catalog doesn't keep growing.
662 - On I/O error, write EOF, then try to write again ? No, keep it simple.
663 - Figure out how compress everything except .gz,... files.
664 Implement FileOptions.
665 - Put Bacula version somewhere in Job stream, probably Start Session Labels.
666 - Fix start/end blocks for File devices
667 - Make Job err if WriteBootstrap fails.
668 - Test that mod of restore options works.
669 - Test that week position schedule code works.
670 - Make BSR accept count (total files to be restored).
671 - Add code to fast seek to proper place on tape/file when doing Restore.
672 - Replace popen() and pclose() -- fail safe and timeout, no SIG dep.
673 - Add code to put VolFile in bsr for restore command.
674 - Volumes can be listed multiple times in Restore volume list.
675 - Add watchdog timeout for child processes start_child_timer()
677 - Get rid of bscan.c:534 error message (one time only).
678 - Print some statistics when get EOF on device in bscan -- feedback
679 to let user know it is working.
680 - DateWritten field on tape may be wrong.
681 - Ensure that restore of differential jobs works (check SQL).
682 - Count number of ignored messages in bscan and print when first SOS is found.
683 - Test that EndFile/Block are correctly updated at end of tape
684 (in view of new block reading code).
685 - Test watchdog child timer code.
686 - Test new BSR code (mostly done).
687 - Work more on how to to a Bacula restore beginning with
688 just a Bacula tape and a boot floppy (bare metal recovery).
689 - Restore options (overwrite, overwrite if older,
690 overwrite if newer, never overwrite, ...)
691 - Fill all fields in Vol/Job Header -- ensure that everything
692 needed is written to tape. Think about restore to Catalog
693 from tape. Client record needs improving.
694 - Implement ./configure --with-client-only
695 - Finish up static linking
696 - that restore options work in FD
697 - Backup of raw partitions
698 - Document nofollow flag
699 - Make sure restore options are documented
700 - Should we dump a SOS when starting a new tape? (no -- too complicated)
701 - Fix gethostbyname() to use gethostbyname_r() -- no added mutex
702 - Add HOST to Volume label.
703 - Need to save contents of FileSet to tape? -- no not now
704 - Fix db_get_fileset in cats/sql_get.c for multiple records.
705 - Fix catalog filename truncation in sql_get and sql_create. Use
706 only a single filename split routine.
707 - Figure out how to save the catalog (possibly a special FileSet).
708 - Figure out how to restore the catalog.