3 Bacula Projects Roadmap
4 Status updated 7 July 2007
5 After re-ordering in vote priority
8 Item: 2 Implement a Bacula GUI/management tool.
9 Item: 18 Quick release of FD-SD connection after backup.
10 Item: 23 Implement from-client and to-client on restore command line.
11 Item: 25 Implement huge exclude list support using dlist
12 Item: 41 Enable to relocate files and directories when restoring
13 Item: 42 Batch attribute inserts (ten times faster)
14 Item: 43 More concurrency in SD using micro-locking
15 Item: 44 Performance enhancements (POSIX/Win32 OS file access hints).
16 Item: 40 Include JobID in spool file name
19 Item: 1 Accurate restoration of renamed/deleted files
20 Item: 2* Implement a Bacula GUI/management tool.
21 Item: 3 Allow FD to initiate a backup
22 Item: 4 Merge multiple backups (Synthetic Backup or Consolidation).
23 Item: 5 Deletion of Disk-Based Bacula Volumes
24 Item: 6 Implement Base jobs.
25 Item: 7 Implement creation and maintenance of copy pools
26 Item: 8 Directive/mode to backup only file changes, not entire file
27 Item: 9 Implement a server-side compression feature
28 Item: 10 Improve Bacula's tape and drive usage and cleaning management.
29 Item: 11 Allow skipping execution of Jobs
30 Item: 12 Add a scheduling syntax that permits weekly rotations
31 Item: 13 Archival (removal) of User Files to Tape
32 Item: 14 Cause daemons to use a specific IP address to source communications
33 Item: 15 Multiple threads in file daemon for the same job
34 Item: 16 Add Plug-ins to the FileSet Include statements.
35 Item: 17 Restore only file attributes (permissions, ACL, owner, group...)
36 Item: 18* Quick release of FD-SD connection after backup.
37 Item: 19 Implement a Python interface to the Bacula catalog.
39 Item: 21 Split documentation
40 Item: 22 Implement support for stacking arbitrary stream filters, sinks.
41 Item: 23* Implement from-client and to-client on restore command line.
42 Item: 24 Add an override in Schedule for Pools based on backup types.
43 Item: 25* Implement huge exclude list support using hashing.
44 Item: 26 Implement more Python events in Bacula.
45 Item: 27 Incorporation of XACML2/SAML2 parsing
46 Item: 28 Filesystem watch triggered backup.
47 Item: 29 Allow inclusion/exclusion of files in a fileset by creation/mod times
48 Item: 30 Tray monitor window cleanups
49 Item: 31 Implement multiple numeric backup levels as supported by dump
50 Item: 32 Automatic promotion of backup levels
51 Item: 33 Clustered file-daemons
52 Item: 34 Commercial database support
53 Item: 35 Automatic disabling of devices
54 Item: 36 An option to operate on all pools with update vol parameters
55 Item: 37 Add an item to the restore option where you can select a pool
56 Item: 38 Include timestamp of job launch in "stat clients" output
57 Item: 39 Message mailing based on backup types
58 Item: 40* Include JobID in spool file name
59 Item: 41* Enable to relocate files and directories when restoring
60 Item: 42* Batch attribute inserts (ten times faster)
61 Item: 43* More concurrency in SD using micro-locking
62 Item: 44* Performance enhancements (POSIX/Win32 OS file access hints).
64 Item 1: Accurate restoration of renamed/deleted files
65 Date: 28 November 2005
66 Origin: Martin Simmons (martin at lispworks dot com)
67 Status: Robert Nelson will implement this
69 What: When restoring a fileset for a specified date (including "most
70 recent"), Bacula should give you exactly the files and directories
71 that existed at the time of the last backup prior to that date.
73 Currently this only works if the last backup was a Full backup.
74 When the last backup was Incremental/Differential, files and
75 directories that have been renamed or deleted since the last Full
76 backup are not currently restored correctly. Ditto for files with
77 extra/fewer hard links than at the time of the last Full backup.
79 Why: Incremental/Differential would be much more useful if this worked.
81 Notes: Merging of multiple backups into a single one seems to
82 rely on this working, otherwise the merged backups will not be
83 truly equivalent to a Full backup.
85 Kern: notes shortened. This can be done without the need for
86 inodes. It is essentially the same as the current Verify job,
87 but one additional database record must be written, which does
88 not need any database change.
90 Kern: see if we can correct restoration of directories if
91 replace=ifnewer is set. Currently, if the directory does not
92 exist, a "dummy" directory is created, then when all the files
93 are updated, the dummy directory is newer so the real values
96 Item 2: Implement a Bacula GUI/management tool.
101 What: Implement a Bacula console, and management tools
102 probably using Qt3 and C++.
104 Why: Don't we already have a wxWidgets GUI? Yes, but
105 it is written in C++ and changes to the user interface
106 must be hand tailored using C++ code. By developing
107 the user interface using Qt designer, the interface
108 can be very easily updated and most of the new Python
109 code will be automatically created. The user interface
110 changes become very simple, and only the new features
111 must be implement. In addition, the code will be in
112 Python, which will give many more users easy (or easier)
113 access to making additions or modifications.
115 Notes: There is a partial Python-GTK implementation
116 Lucas Di Pentima <lucas at lunix dot com dot ar> but
117 it is no longer being developed.
119 Item 3: Allow FD to initiate a backup
120 Origin: Frank Volf (frank at deze dot org)
121 Date: 17 November 2005
124 What: Provide some means, possibly by a restricted console that
125 allows a FD to initiate a backup, and that uses the connection
126 established by the FD to the Director for the backup so that
127 a Director that is firewalled can do the backup.
129 Why: Makes backup of laptops much easier.
132 Item 4: Merge multiple backups (Synthetic Backup or Consolidation).
133 Origin: Marc Cousin and Eric Bollengier
134 Date: 15 November 2005
135 Status: Waiting implementation. Depends on first implementing
136 project Item 2 (Migration) which is now done.
138 What: A merged backup is a backup made without connecting to the Client.
139 It would be a Merge of existing backups into a single backup.
140 In effect, it is like a restore but to the backup medium.
142 For instance, say that last Sunday we made a full backup. Then
143 all week long, we created incremental backups, in order to do
144 them fast. Now comes Sunday again, and we need another full.
145 The merged backup makes it possible to do instead an incremental
146 backup (during the night for instance), and then create a merged
147 backup during the day, by using the full and incrementals from
148 the week. The merged backup will be exactly like a full made
149 Sunday night on the tape, but the production interruption on the
150 Client will be minimal, as the Client will only have to send
153 In fact, if it's done correctly, you could merge all the
154 Incrementals into single Incremental, or all the Incrementals
155 and the last Differential into a new Differential, or the Full,
156 last differential and all the Incrementals into a new Full
157 backup. And there is no need to involve the Client.
159 Why: The benefit is that :
160 - the Client just does an incremental ;
161 - the merged backup on tape is just as a single full backup,
162 and can be restored very fast.
164 This is also a way of reducing the backup data since the old
165 data can then be pruned (or not) from the catalog, possibly
166 allowing older volumes to be recycled
168 Item 5: Deletion of Disk-Based Bacula Volumes
170 Origin: Ross Boylan <RossBoylan at stanfordalumni dot org> (edited
174 What: Provide a way for Bacula to automatically remove Volumes
175 from the filesystem, or optionally to truncate them.
176 Obviously, the Volume must be pruned prior removal.
178 Why: This would allow users more control over their Volumes and
179 prevent disk based volumes from consuming too much space.
181 Notes: The following two directives might do the trick:
183 Volume Data Retention = <time period>
184 Remove Volume After = <time period>
186 The migration project should also remove a Volume that is
187 migrated. This might also work for tape Volumes.
189 Item 6: Implement Base jobs.
190 Date: 28 October 2005
194 What: A base job is sort of like a Full save except that you
195 will want the FileSet to contain only files that are
196 unlikely to change in the future (i.e. a snapshot of
197 most of your system after installing it). After the
198 base job has been run, when you are doing a Full save,
199 you specify one or more Base jobs to be used. All
200 files that have been backed up in the Base job/jobs but
201 not modified will then be excluded from the backup.
202 During a restore, the Base jobs will be automatically
203 pulled in where necessary.
205 Why: This is something none of the competition does, as far as
206 we know (except perhaps BackupPC, which is a Perl program that
207 saves to disk only). It is big win for the user, it
208 makes Bacula stand out as offering a unique
209 optimization that immediately saves time and money.
210 Basically, imagine that you have 100 nearly identical
211 Windows or Linux machine containing the OS and user
212 files. Now for the OS part, a Base job will be backed
213 up once, and rather than making 100 copies of the OS,
214 there will be only one. If one or more of the systems
215 have some files updated, no problem, they will be
216 automatically restored.
218 Notes: Huge savings in tape usage even for a single machine.
219 Will require more resources because the DIR must send
220 FD a list of files/attribs, and the FD must search the
221 list and compare it for each file to be saved.
223 Item 7: Implement creation and maintenance of copy pools
224 Date: 27 November 2005
225 Origin: David Boyes (dboyes at sinenomine dot net)
228 What: I would like Bacula to have the capability to write copies
229 of backed-up data on multiple physical volumes selected
230 from different pools without transferring the data
231 multiple times, and to accept any of the copy volumes
232 as valid for restore.
234 Why: In many cases, businesses are required to keep offsite
235 copies of backup volumes, or just wish for simple
236 protection against a human operator dropping a storage
237 volume and damaging it. The ability to generate multiple
238 volumes in the course of a single backup job allows
239 customers to simple check out one copy and send it
240 offsite, marking it as out of changer or otherwise
241 unavailable. Currently, the library and magazine
242 management capability in Bacula does not make this process
245 Restores would use the copy of the data on the first
246 available volume, in order of copy pool chain definition.
248 This is also a major scalability issue -- as the number of
249 clients increases beyond several thousand, and the volume
250 of data increases, transferring the data multiple times to
251 produce additional copies of the backups will become
252 physically impossible due to transfer speed
253 issues. Generating multiple copies at server side will
254 become the only practical option.
256 How: I suspect that this will require adding a multiplexing
257 SD that appears to be a SD to a specific FD, but 1-n FDs
258 to the specific back end SDs managing the primary and copy
259 pools. Storage pools will also need to acquire parameters
260 to define the pools to be used for copies.
262 Notes: I would commit some of my developers' time if we can agree
263 on the design and behavior.
265 Item 8: Directive/mode to backup only file changes, not entire file
266 Date: 11 November 2005
267 Origin: Joshua Kugler <joshua dot kugler at uaf dot edu>
268 Marek Bajon <mbajon at bimsplus dot com dot pl>
271 What: Currently when a file changes, the entire file will be backed up in
272 the next incremental or full backup. To save space on the tapes
273 it would be nice to have a mode whereby only the changes to the
274 file would be backed up when it is changed.
276 Why: This would save lots of space when backing up large files such as
277 logs, mbox files, Outlook PST files and the like.
279 Notes: This would require the usage of disk-based volumes as comparing
280 files would not be feasible using a tape drive.
282 Item 9: Implement a server-side compression feature
283 Date: 18 December 2006
284 Origin: Vadim A. Umanski , e-mail umanski@ext.ru
286 What: The ability to compress backup data on server receiving data
287 instead of doing that on client sending data.
288 Why: The need is practical. I've got some machines that can send
289 data to the network 4 or 5 times faster than compressing
290 them (I've measured that). They're using fast enough SCSI/FC
291 disk subsystems but rather slow CPUs (ex. UltraSPARC II).
292 And the backup server has got a quite fast CPUs (ex. Dual P4
293 Xeons) and quite a low load. When you have 20, 50 or 100 GB
294 of raw data - running a job 4 to 5 times faster - that
295 really matters. On the other hand, the data can be
296 compressed 50% or better - so losing twice more space for
297 disk backup is not good at all. And the network is all mine
298 (I have a dedicated management/provisioning network) and I
299 can get as high bandwidth as I need - 100Mbps, 1000Mbps...
300 That's why the server-side compression feature is needed!
303 Item 10: Improve Bacula's tape and drive usage and cleaning management.
304 Date: 8 November 2005, November 11, 2005
305 Origin: Adam Thornton <athornton at sinenomine dot net>,
306 Arno Lehmann <al at its-lehmann dot de>
309 What: Make Bacula manage tape life cycle information, tape reuse
310 times and drive cleaning cycles.
312 Why: All three parts of this project are important when operating
314 We need to know which tapes need replacement, and we need to
315 make sure the drives are cleaned when necessary. While many
316 tape libraries and even autoloaders can handle all this
317 automatically, support by Bacula can be helpful for smaller
318 (older) libraries and single drives. Limiting the number of
319 times a tape is used might prevent tape errors when using
320 tapes until the drives can't read it any more. Also, checking
321 drive status during operation can prevent some failures (as I
322 [Arno] had to learn the hard way...)
324 Notes: First, Bacula could (and even does, to some limited extent)
325 record tape and drive usage. For tapes, the number of mounts,
326 the amount of data, and the time the tape has actually been
327 running could be recorded. Data fields for Read and Write
328 time and Number of mounts already exist in the catalog (I'm
329 not sure if VolBytes is the sum of all bytes ever written to
330 that volume by Bacula). This information can be important
331 when determining which media to replace. The ability to mark
332 Volumes as "used up" after a given number of write cycles
333 should also be implemented so that a tape is never actually
334 worn out. For the tape drives known to Bacula, similar
335 information is interesting to determine the device status and
336 expected life time: Time it's been Reading and Writing, number
337 of tape Loads / Unloads / Errors. This information is not yet
338 recorded as far as I [Arno] know. A new volume status would
339 be necessary for the new state, like "Used up" or "Worn out".
340 Volumes with this state could be used for restores, but not
341 for writing. These volumes should be migrated first (assuming
342 migration is implemented) and, once they are no longer needed,
343 could be moved to a Trash pool.
345 The next step would be to implement a drive cleaning setup.
346 Bacula already has knowledge about cleaning tapes. Once it
347 has some information about cleaning cycles (measured in drive
348 run time, number of tapes used, or calender days, for example)
349 it can automatically execute tape cleaning (with an
350 autochanger, obviously) or ask for operator assistance loading
353 The final step would be to implement TAPEALERT checks not only
354 when changing tapes and only sending the information to the
355 administrator, but rather checking after each tape error,
356 checking on a regular basis (for example after each tape
357 file), and also before unloading and after loading a new tape.
358 Then, depending on the drives TAPEALERT state and the known
359 drive cleaning state Bacula could automatically schedule later
360 cleaning, clean immediately, or inform the operator.
362 Implementing this would perhaps require another catalog change
363 and perhaps major changes in SD code and the DIR-SD protocol,
364 so I'd only consider this worth implementing if it would
365 actually be used or even needed by many people.
367 Implementation of these projects could happen in three distinct
368 sub-projects: Measuring Tape and Drive usage, retiring
369 volumes, and handling drive cleaning and TAPEALERTs.
371 Item 11: Allow skipping execution of Jobs
372 Date: 29 November 2005
373 Origin: Florian Schnabel <florian.schnabel at docufy dot de>
376 What: An easy option to skip a certain job on a certain date.
377 Why: You could then easily skip tape backups on holidays. Especially
378 if you got no autochanger and can only fit one backup on a tape
379 that would be really handy, other jobs could proceed normally
380 and you won't get errors that way.
382 Item 12: Add a scheduling syntax that permits weekly rotations
383 Date: 15 December 2006
384 Origin: Gregory Brauer (greg at wildbrain dot com)
387 What: Currently, Bacula only understands how to deal with weeks of the
388 month or weeks of the year in schedules. This makes it impossible
389 to do a true weekly rotation of tapes. There will always be a
390 discontinuity that will require disruptive manual intervention at
391 least monthly or yearly because week boundaries never align with
392 month or year boundaries.
394 A solution would be to add a new syntax that defines (at least)
395 a start timestamp, and repetition period.
397 Why: Rotated backups done at weekly intervals are useful, and Bacula
398 cannot currently do them without extensive hacking.
400 Notes: Here is an example syntax showing a 3-week rotation where full
401 Backups would be performed every week on Saturday, and an
402 incremental would be performed every week on Tuesday. Each
403 set of tapes could be removed from the loader for the following
404 two cycles before coming back and being reused on the third
405 week. Since the execution times are determined by intervals
406 from a given point in time, there will never be any issues with
407 having to adjust to any sort of arbitrary time boundary. In
408 the example provided, I even define the starting schedule
409 as crossing both a year and a month boundary, but the run times
410 would be based on the "Repeat" value and would therefore happen
415 Name = "Week 1 Rotation"
416 #Saturday. Would run Dec 30, Jan 20, Feb 10, etc.
420 Start = 2006-12-30 01:00
424 #Tuesday. Would run Jan 2, Jan 23, Feb 13, etc.
428 Start = 2007-01-02 01:00
435 Name = "Week 2 Rotation"
436 #Saturday. Would run Jan 6, Jan 27, Feb 17, etc.
440 Start = 2007-01-06 01:00
444 #Tuesday. Would run Jan 9, Jan 30, Feb 20, etc.
448 Start = 2007-01-09 01:00
455 Name = "Week 3 Rotation"
456 #Saturday. Would run Jan 13, Feb 3, Feb 24, etc.
460 Start = 2007-01-13 01:00
464 #Tuesday. Would run Jan 16, Feb 6, Feb 27, etc.
468 Start = 2007-01-16 01:00
474 Item 13: Archival (removal) of User Files to Tape
476 Origin: Ray Pengelly [ray at biomed dot queensu dot ca
479 What: The ability to archive data to storage based on certain parameters
480 such as age, size, or location. Once the data has been written to
481 storage and logged it is then pruned from the originating
482 filesystem. Note! We are talking about user's files and not
485 Why: This would allow fully automatic storage management which becomes
486 useful for large datastores. It would also allow for auto-staging
487 from one media type to another.
489 Example 1) Medical imaging needs to store large amounts of data.
490 They decide to keep data on their servers for 6 months and then put
491 it away for long term storage. The server then finds all files
492 older than 6 months writes them to tape. The files are then removed
495 Example 2) All data that hasn't been accessed in 2 months could be
496 moved from high-cost, fibre-channel disk storage to a low-cost
497 large-capacity SATA disk storage pool which doesn't have as quick of
498 access time. Then after another 6 months (or possibly as one
499 storage pool gets full) data is migrated to Tape.
501 Item 14: Cause daemons to use a specific IP address to source communications
502 Origin: Bill Moran <wmoran@collaborativefusion.com>
505 What: Cause Bacula daemons (dir, fd, sd) to always use the ip address
506 specified in the [DIR|DF|SD]Addr directive as the source IP
507 for initiating communication.
508 Why: On complex networks, as well as extremely secure networks, it's
509 not unusual to have multiple possible routes through the network.
510 Often, each of these routes is secured by different policies
511 (effectively, firewalls allow or deny different traffic depending
512 on the source address)
513 Unfortunately, it can sometimes be difficult or impossible to
514 represent this in a system routing table, as the result is
515 excessive subnetting that quickly exhausts available IP space.
516 The best available workaround is to provide multiple IPs to
517 a single machine that are all on the same subnet. In order
518 for this to work properly, applications must support the ability
519 to bind outgoing connections to a specified address, otherwise
520 the operating system will always choose the first IP that
521 matches the required route.
522 Notes: Many other programs support this. For example, the following
523 can be configured in BIND:
524 query-source address 10.0.0.1;
525 transfer-source 10.0.0.2;
526 Which means queries from this server will always come from
527 10.0.0.1 and zone transfers will always originate from
530 Item 15: Multiple threads in file daemon for the same job
531 Date: 27 November 2005
532 Origin: Ove Risberg (Ove.Risberg at octocode dot com)
535 What: I want the file daemon to start multiple threads for a backup
536 job so the fastest possible backup can be made.
538 The file daemon could parse the FileSet information and start
539 one thread for each File entry located on a separate
542 A confiuration option in the job section should be used to
543 enable or disable this feature. The confgutration option could
544 specify the maximum number of threads in the file daemon.
546 If the theads could spool the data to separate spool files
547 the restore process will not be much slower.
549 Why: Multiple concurrent backups of a large fileserver with many
550 disks and controllers will be much faster.
552 Item 16: Add Plug-ins to the FileSet Include statements.
553 Date: 28 October 2005
555 Status: Partially coded in 1.37 -- much more to do.
557 What: Allow users to specify wild-card and/or regular
558 expressions to be matched in both the Include and
559 Exclude directives in a FileSet. At the same time,
560 allow users to define plug-ins to be called (based on
561 regular expression/wild-card matching).
563 Why: This would give the users the ultimate ability to control
564 how files are backed up/restored. A user could write a
565 plug-in knows how to backup his Oracle database without
566 stopping/starting it, for example.
568 Item 17: Restore only file attributes (permissions, ACL, owner, group...)
569 Origin: Eric Bollengier
573 What: The goal of this project is to be able to restore only rights
574 and attributes of files without crushing them.
576 Why: Who have never had to repair a chmod -R 777, or a wild update
577 of recursive right under Windows? At this time, you must have
578 enough space to restore data, dump attributes (easy with acl,
579 more complex with unix/windows rights) and apply them to your
580 broken tree. With this options, it will be very easy to compare
581 right or ACL over the time.
583 Notes: If the file is here, we skip restore and we change rights.
584 If the file isn't here, we can create an empty one and apply
585 rights or do nothing.
587 Item 18: Quick release of FD-SD connection after backup.
588 Origin: Frank Volf (frank at deze dot org)
589 Date: 17 November 2005
590 Status: Done -- implemented by Kern -- in CVS 26Jan07
592 What: In the Bacula implementation a backup is finished after all data
593 and attributes are successfully written to storage. When using a
594 tape backup it is very annoying that a backup can take a day,
595 simply because the current tape (or whatever) is full and the
596 administrator has not put a new one in. During that time the
597 system cannot be taken off-line, because there is still an open
598 session between the storage daemon and the file daemon on the
601 Although this is a very good strategy for making "safe backups"
602 This can be annoying for e.g. laptops, that must remain
603 connected until the backup is completed.
605 Using a new feature called "migration" it will be possible to
606 spool first to harddisk (using a special 'spool' migration
607 scheme) and then migrate the backup to tape.
609 There is still the problem of getting the attributes committed.
610 If it takes a very long time to do, with the current code, the
611 job has not terminated, and the File daemon is not freed up. The
612 Storage daemon should release the File daemon as soon as all the
613 file data and all the attributes have been sent to it (the SD).
614 Currently the SD waits until everything is on tape and all the
615 attributes are transmitted to the Director before signaling
616 completion to the FD. I don't think I would have any problem
617 changing this. The reason is that even if the FD reports back to
618 the Dir that all is OK, the job will not terminate until the SD
619 has done the same thing -- so in a way keeping the SD-FD link
620 open to the very end is not really very productive ...
622 Why: Makes backup of laptops much faster.
624 Item 19: Implement a Python interface to the Bacula catalog.
625 Date: 28 October 2005
629 What: Implement an interface for Python scripts to access
630 the catalog through Bacula.
632 Why: This will permit users to customize Bacula through
635 Item 20: Archive data
637 Origin: calvin streeting calvin at absentdream dot com
640 What: The abilty to archive to media (dvd/cd) in a uncompressed format
641 for dead filing (archiving not backing up)
643 Why: At my works when jobs are finished and moved off of the main file
644 servers (raid based systems) onto a simple linux file server (ide based
645 system) so users can find old information without contacting the IT
648 So this data dosn't realy change it only gets added to,
649 But it also needs backing up. At the moment it takes
650 about 8 hours to back up our servers (working data) so
651 rather than add more time to existing backups i am trying
652 to implement a system where we backup the acrhive data to
653 cd/dvd these disks would only need to be appended to
654 (burn only new/changed files to new disks for off site
655 storage). basialy understand the differnce between
656 achive data and live data.
658 Notes: Scan the data and email me when it needs burning divide
659 into predifind chunks keep a recored of what is on what
660 disk make me a label (simple php->mysql=>pdf stuff) i
661 could do this bit ability to save data uncompresed so
662 it can be read in any other system (future proof data)
663 save the catalog with the disk as some kind of menu
666 Item 21: Split documentation
667 Origin: Maxx <maxxatworkat gmail dot com>
671 What: Split documentation in several books
673 Why: Bacula manual has now more than 600 pages, and looking for
674 implementation details is getting complicated. I think
675 it would be good to split the single volume in two or
678 1) Introduction, requirements and tutorial, typically
679 are useful only until first installation time
681 2) Basic installation and configuration, with all the
682 gory details about the directives supported 3)
683 Advanced Bacula: testing, troubleshooting, GUI and
684 ancillary programs, security managements, scripting,
688 Item 22: Implement support for stacking arbitrary stream filters, sinks.
689 Date: 23 November 2006
690 Origin: Landon Fuller <landonf@threerings.net>
691 Status: Planning. Assigned to landonf.
693 What: Implement support for the following:
694 - Stacking arbitrary stream filters (eg, encryption, compression,
695 sparse data handling))
696 - Attaching file sinks to terminate stream filters (ie, write out
697 the resultant data to a file)
698 - Refactor the restoration state machine accordingly
700 Why: The existing stream implementation suffers from the following:
701 - All state (compression, encryption, stream restoration), is
702 global across the entire restore process, for all streams. There are
703 multiple entry and exit points in the restoration state machine, and
704 thus multiple places where state must be allocated, deallocated,
705 initialized, or reinitialized. This results in exceptional complexity
706 for the author of a stream filter.
707 - The developer must enumerate all possible combinations of filters
708 and stream types (ie, win32 data with encryption, without encryption,
709 with encryption AND compression, etc).
711 Notes: This feature request only covers implementing the stream filters/
712 sinks, and refactoring the file daemon's restoration implementation
713 accordingly. If I have extra time, I will also rewrite the backup
714 implementation. My intent in implementing the restoration first is to
715 solve pressing bugs in the restoration handling, and to ensure that
716 the new restore implementation handles existing backups correctly.
718 I do not plan on changing the network or tape data structures to
719 support defining arbitrary stream filters, but supporting that
720 functionality is the ultimate goal.
722 Assistance with either code or testing would be fantastic.
724 Item 23: Implement from-client and to-client on restore command line.
725 Date: 11 December 2006
726 Origin: Discussion on Bacula-users entitled 'Scripted restores to
727 different clients', December 2006
728 Status: New feature request
730 What: While using bconsole interactively, you can specify the client
731 that a backup job is to be restored for, and then you can
732 specify later a different client to send the restored files
733 back to. However, using the 'restore' command with all options
734 on the command line, this cannot be done, due to the ambiguous
735 'client' parameter. Additionally, this parameter means different
736 things depending on if it's specified on the command line or
737 afterwards, in the Modify Job screens.
739 Why: This feature would enable restore jobs to be more completely
740 automated, for example by a web or GUI front-end.
742 Notes: client can also be implied by specifying the jobid on the command
745 Item 24: Add an override in Schedule for Pools based on backup types.
747 Origin: Chad Slater <chad.slater@clickfox.com>
750 What: Adding a FullStorage=BigTapeLibrary in the Schedule resource
751 would help those of us who use different storage devices for different
752 backup levels cope with the "auto-upgrade" of a backup.
754 Why: Assume I add several new device to be backed up, i.e. several
755 hosts with 1TB RAID. To avoid tape switching hassles, incrementals are
756 stored in a disk set on a 2TB RAID. If you add these devices in the
757 middle of the month, the incrementals are upgraded to "full" backups,
758 but they try to use the same storage device as requested in the
759 incremental job, filling up the RAID holding the differentials. If we
760 could override the Storage parameter for full and/or differential
761 backups, then the Full job would use the proper Storage device, which
762 has more capacity (i.e. a 8TB tape library.
764 Item 25: Implement huge exclude list support using hashing (dlists).
765 Date: 28 October 2005
767 Status: Done in 2.1.2 but was done with dlists (doubly linked lists
768 since hashing will not help. The huge list also supports
769 large include lists).
771 What: Allow users to specify very large exclude list (currently
772 more than about 1000 files is too many).
774 Why: This would give the users the ability to exclude all
775 files that are loaded with the OS (e.g. using rpms
776 or debs). If the user can restore the base OS from
777 CDs, there is no need to backup all those files. A
778 complete restore would be to restore the base OS, then
779 do a Bacula restore. By excluding the base OS files, the
780 backup set will be *much* smaller.
782 Item 26: Implement more Python events in Bacula.
783 Date: 28 October 2005
787 What: Allow Python scripts to be called at more places
788 within Bacula and provide additional access to Bacula
791 Why: This will permit users to customize Bacula through
799 Also add a way to get a listing of currently running
800 jobs (possibly also scheduled jobs).
803 Item 27: Incorporation of XACML2/SAML2 parsing
804 Date: 19 January 2006
805 Origin: Adam Thornton <athornton@sinenomine.net>
808 What: XACML is "eXtensible Access Control Markup Language" and
809 "SAML is the "Security Assertion Markup Language"--an XML standard
810 for making statements about identity and authorization. Having these
811 would give us a framework to approach ACLs in a generic manner, and
812 in a way flexible enough to support the four major sorts of ACLs I
813 see as a concern to Bacula at this point, as well as (probably) to
814 deal with new sorts of ACLs that may appear in the future.
816 Why: Bacula is beginning to need to back up systems with ACLs
817 that do not map cleanly onto traditional Unix permissions. I see
818 four sets of ACLs--in general, mutually incompatible with one
819 another--that we're going to need to deal with. These are: NTFS
820 ACLs, POSIX ACLs, NFSv4 ACLS, and AFS ACLS. (Some may question the
821 relevance of AFS; AFS is one of Sine Nomine's core consulting
822 businesses, and having a reputable file-level backup and restore
823 technology for it (as Tivoli is probably going to drop AFS support
824 soon since IBM no longer supports AFS) would be of huge benefit to
825 our customers; we'd most likely create the AFS support at Sine Nomine
826 for inclusion into the Bacula (and perhaps some changes to the
827 OpenAFS volserver) core code.)
829 Now, obviously, Bacula already handles NTFS just fine. However, I
830 think there's a lot of value in implementing a generic ACL model, so
831 that it's easy to support whatever particular instances of ACLs come
832 down the pike: POSIX ACLS (think SELinux) and NFSv4 are the obvious
833 things arriving in the Linux world in a big way in the near future.
834 XACML, although overcomplicated for our needs, provides this
835 framework, and we should be able to leverage other people's
836 implementations to minimize the amount of work *we* have to do to get
837 a generic ACL framework. Basically, the costs of implementation are
838 high, but they're largely both external to Bacula and already sunk.
840 Item 28: Filesystem watch triggered backup.
842 Origin: Jesper Krogh <jesper@krogh.cc>
843 Status: Unimplemented, depends probably on "client initiated backups"
845 What: With inotify and similar filesystem triggeret notification
846 systems is it possible to have the file-daemon to monitor
847 filesystem changes and initiate backup.
849 Why: There are 2 situations where this is nice to have.
850 1) It is possible to get a much finer-grained backup than
851 the fixed schedules used now.. A file created and deleted
852 a few hours later, can automatically be caught.
854 2) The introduced load on the system will probably be
855 distributed more even on the system.
857 Notes: This can be combined with configration that specifies
858 something like: "at most every 15 minutes or when changes
861 Kern Notes: I would rather see this implemented by an external program
862 that monitors the Filesystem changes, then uses the console
863 to start the appropriate job.
865 Item 29: Allow inclusion/exclusion of files in a fileset by creation/mod times
866 Origin: Evan Kaufman <evan.kaufman@gmail.com>
867 Date: January 11, 2006
870 What: In the vein of the Wild and Regex directives in a Fileset's
871 Options, it would be helpful to allow a user to include or exclude
872 files and directories by creation or modification times.
874 You could factor the Exclude=yes|no option in much the same way it
875 affects the Wild and Regex directives. For example, you could exclude
876 all files modified before a certain date:
880 Modified Before = ####
883 Or you could exclude all files created/modified since a certain date:
887 Created Modified Since = ####
890 The format of the time/date could be done several ways, say the number
891 of seconds since the epoch:
892 1137008553 = Jan 11 2006, 1:42:33PM # result of `date +%s`
894 Or a human readable date in a cryptic form:
895 20060111134233 = Jan 11 2006, 1:42:33PM # YYYYMMDDhhmmss
897 Why: I imagine a feature like this could have many uses. It would
898 allow a user to do a full backup while excluding the base operating
899 system files, so if I installed a Linux snapshot from a CD yesterday,
900 I'll *exclude* all files modified *before* today. If I need to
901 recover the system, I use the CD I already have, plus the tape backup.
902 Or if, say, a Windows client is hit by a particularly corrosive
903 virus, and I need to *exclude* any files created/modified *since* the
906 Notes: Of course, this feature would work in concert with other
907 in/exclude rules, and wouldnt override them (or each other).
909 Notes: The directives I'd imagine would be along the lines of
910 "[Created] [Modified] [Before|Since] = <date>".
911 So one could compare against 'ctime' and/or 'mtime', but ONLY 'before'
915 Item 30: Tray monitor window cleanups
916 Origin: Alan Brown ajb2 at mssl dot ucl dot ac dot uk
919 What: Resizeable and scrollable windows in the tray monitor.
921 Why: With multiple clients, or with many jobs running, the displayed
922 window often ends up larger than the available screen, making
923 the trailing items difficult to read.
926 Item 31: Implement multiple numeric backup levels as supported by dump
928 Origin: Daniel Rich <drich@employees.org>
930 What: Dump allows specification of backup levels numerically instead of just
931 "full", "incr", and "diff". In this system, at any given level, all
932 files are backed up that were were modified since the last backup of a
933 higher level (with 0 being the highest and 9 being the lowest). A
934 level 0 is therefore equivalent to a full, level 9 an incremental, and
935 the levels 1 through 8 are varying levels of differentials. For
936 bacula's sake, these could be represented as "full", "incr", and
937 "diff1", "diff2", etc.
939 Why: Support of multiple backup levels would provide for more advanced backup
940 rotation schemes such as "Towers of Hanoi". This would allow better
941 flexibility in performing backups, and can lead to shorter recover
944 Notes: Legato Networker supports a similar system with full, incr, and 1-9 as
947 Item 32: Automatic promotion of backup levels
948 Date: 19 January 2006
949 Origin: Adam Thornton <athornton@sinenomine.net>
952 What: Amanda has a feature whereby it estimates the space that a
953 differential, incremental, and full backup would take. If the
954 difference in space required between the scheduled level and the next
955 level up is beneath some user-defined critical threshold, the backup
956 level is bumped to the next type. Doing this minimizes the number of
957 volumes necessary during a restore, with a fairly minimal cost in
960 Why: I know at least one (quite sophisticated and smart) user
961 for whom the absence of this feature is a deal-breaker in terms of
962 using Bacula; if we had it it would eliminate the one cool thing
963 Amanda can do and we can't (at least, the one cool thing I know of).
965 Item 33: Clustered file-daemons
966 Origin: Alan Brown ajb2 at mssl dot ucl dot ac dot uk
969 What: A "virtual" filedaemon, which is actually a cluster of real ones.
971 Why: In the case of clustered filesystems (SAN setups, GFS, or OCFS2, etc)
972 multiple machines may have access to the same set of filesystems
974 For performance reasons, one may wish to initate backups from
975 several of these machines simultaneously, instead of just using
976 one backup source for the common clustered filesystem.
978 For obvious reasons, normally backups of $A-FD/$PATH and
979 B-FD/$PATH are treated as different backup sets. In this case
980 they are the same communal set.
982 Likewise when restoring, it would be easier to just specify
983 one of the cluster machines and let bacula decide which to use.
985 This can be faked to some extent using DNS round robin entries
986 and a virtual IP address, however it means "status client" will
987 always give bogus answers. Additionally there is no way of
988 spreading the load evenly among the servers.
990 What is required is something similar to the storage daemon
991 autochanger directives, so that Bacula can keep track of
992 operating backups/restores and direct new jobs to a "free"
997 Item 34: Commercial database support
998 Origin: Russell Howe <russell_howe dot wreckage dot org>
1002 What: It would be nice for the database backend to support more
1003 databases. I'm thinking of SQL Server at the moment, but I guess Oracle,
1004 DB2, MaxDB, etc are all candidates. SQL Server would presumably be
1005 implemented using FreeTDS or maybe an ODBC library?
1007 Why: We only really have one database server, which is MS SQL Server
1008 2000. Maintaining a second one for the backup software (we grew out of
1009 SQLite, which I liked, but which didn't work so well with our database
1010 size). We don't really have a machine with the resources to run
1011 postgres, and would rather only maintain a single DBMS. We're stuck with
1012 SQL Server because pretty much all the company's custom applications
1013 (written by consultants) are locked into SQL Server 2000. I can imagine
1014 this scenario is fairly common, and it would be nice to use the existing
1015 properly specced database server for storing Bacula's catalog, rather
1016 than having to run a second DBMS.
1018 Item 35: Automatic disabling of devices
1020 Origin: Peter Eriksson <peter at ifm.liu dot se>
1023 What: After a configurable amount of fatal errors with a tape drive
1024 Bacula should automatically disable further use of a certain
1025 tape drive. There should also be "disable"/"enable" commands in
1026 the "bconsole" tool.
1028 Why: On a multi-drive jukebox there is a possibility of tape drives
1029 going bad during large backups (needing a cleaning tape run,
1030 tapes getting stuck). It would be advantageous if Bacula would
1031 automatically disable further use of a problematic tape drive
1032 after a configurable amount of errors has occurred.
1034 An example: I have a multi-drive jukebox (6 drives, 380+ slots)
1035 where tapes occasionally get stuck inside the drive. Bacula will
1036 notice that the "mtx-changer" command will fail and then fail
1037 any backup jobs trying to use that drive. However, it will still
1038 keep on trying to run new jobs using that drive and fail -
1039 forever, and thus failing lots and lots of jobs... Since we have
1040 many drives Bacula could have just automatically disabled
1041 further use of that drive and used one of the other ones
1044 Item 36: An option to operate on all pools with update vol parameters
1045 Origin: Dmitriy Pinchukov <absh@bossdev.kiev.ua>
1046 Date: 16 August 2006
1049 What: When I do update -> Volume parameters -> All Volumes
1050 from Pool, then I have to select pools one by one. I'd like
1051 console to have an option like "0: All Pools" in the list of
1054 Why: I have many pools and therefore unhappy with manually
1055 updating each of them using update -> Volume parameters -> All
1056 Volumes from Pool -> pool #.
1058 Item 37: Add an item to the restore option where you can select a pool
1059 Origin: kshatriyak at gmail dot com
1063 What: In the restore option (Select the most recent backup for a
1064 client) it would be useful to add an option where you can limit
1065 the selection to a certain pool.
1067 Why: When using cloned jobs, most of the time you have 2 pools - a
1068 disk pool and a tape pool. People who have 2 pools would like to
1069 select the most recent backup from disk, not from tape (tape
1070 would be only needed in emergency). However, the most recent
1071 backup (which may just differ a second from the disk backup) may
1072 be on tape and would be selected. The problem becomes bigger if
1073 you have a full and differential - the most "recent" full backup
1074 may be on disk, while the most recent differential may be on tape
1075 (though the differential on disk may differ even only a second or
1076 so). Bacula will complain that the backups reside on different
1077 media then. For now the only solution now when restoring things
1078 when you have 2 pools is to manually search for the right
1079 job-id's and enter them by hand, which is a bit fault tolerant.
1081 Item 38: Include timestamp of job launch in "stat clients" output
1082 Origin: Mark Bergman <mark.bergman@uphs.upenn.edu>
1083 Date: Tue Aug 22 17:13:39 EDT 2006
1086 What: The "stat clients" command doesn't include any detail on when
1087 the active backup jobs were launched.
1089 Why: Including the timestamp would make it much easier to decide whether
1090 a job is running properly.
1092 Notes: It may be helpful to have the output from "stat clients" formatted
1093 more like that from "stat dir" (and other commands), in a column
1094 format. The per-client information that's currently shown (level,
1095 client name, JobId, Volume, pool, device, Files, etc.) is good, but
1096 somewhat hard to parse (both programmatically and visually),
1097 particularly when there are many active clients.
1100 Item 39: Message mailing based on backup types
1101 Origin: Evan Kaufman <evan.kaufman@gmail.com>
1102 Date: January 6, 2006
1105 What: In the "Messages" resource definitions, allowing messages
1106 to be mailed based on the type (backup, restore, etc.) and level
1107 (full, differential, etc) of job that created the originating
1110 Why: It would, for example, allow someone's boss to be emailed
1111 automatically only when a Full Backup job runs, so he can
1112 retrieve the tapes for offsite storage, even if the IT dept.
1113 doesn't (or can't) explicitly notify him. At the same time, his
1114 mailbox wouldnt be filled by notifications of Verifies, Restores,
1115 or Incremental/Differential Backups (which would likely be kept
1118 Notes: One way this could be done is through additional message types, for example:
1121 # email the boss only on full system backups
1122 Mail = boss@mycompany.com = full, !incremental, !differential, !restore,
1124 # email us only when something breaks
1125 MailOnError = itdept@mycompany.com = all
1129 Item 40: Include JobID in spool file name ****DONE****
1130 Origin: Mark Bergman <mark.bergman@uphs.upenn.edu>
1131 Date: Tue Aug 22 17:13:39 EDT 2006
1132 Status: Done. (patches/testing/project-include-jobid-in-spool-name.patch)
1133 No need to vote for this item.
1135 What: Change the name of the spool file to include the JobID
1137 Why: JobIDs are the common key used to refer to jobs, yet the
1138 spoolfile name doesn't include that information. The date/time
1139 stamp is useful (and should be retained).
1141 ============= New Freature Requests after vote of 26 Jan 2007 ========
1142 Item 41: Enable to relocate files and directories when restoring
1144 Origin: Eric Bollengier <eric@eb.homelinux.org>
1147 What: The where= option is not powerful enough. It will be
1148 a great feature if bacula can restore a file in the
1149 same directory, but with a different name, or in
1150 an other directory without recreating the full path.
1152 Why: When i want to restore a production environment to a
1153 development environment, i just want change the first
1154 directory. ie restore /prod/data/file.dat to /rect/data/file.dat.
1155 At this time, i have to move by hand files. You must have a big
1156 dump space to restore and move data after.
1158 When i use Linux or SAN snapshot, i mount them to /mnt/snap_xxx
1159 so, when a restore a file, i have to move by hand
1160 from /mnt/snap_xxx/file to /xxx/file. I can't replace a file
1163 When a user ask me to restore a file in its personal folder,
1164 (without replace the existing one), i can't restore from
1165 my_file.txt to my_file.txt.old witch is very practical.
1168 Notes: I think we can enhance the where= option very easily by
1169 allowing regexp expression.
1171 Since, many users think that regexp are not user friendly, i think
1172 that bat, bconsole or brestore must provide a simple way to
1173 configure where= option (i think to something like in
1174 openoffice "search and replace").
1176 Ie, if user uses where=/tmp/bacula-restore, we keep the old
1179 If user uses something like where=s!/prod!/test!, files will
1180 be restored from /prod/xxx to /test/xxx.
1182 If user uses something like where=s/$/.old/, files will
1183 be restored from /prod/xxx.txt to /prod/xxx.txt.old.
1185 If user uses something like where=s/txt$/old.txt/, files will
1186 be restored from /prod/xxx.txt to /prod/xxx.old.txt
1188 if user uses something like where=s/([a-z]+)$/old.$1/, files will
1189 be restored from /prod/xxx.ext to /prod/xxx.old.ext
1191 Item n: Implement Catalog directive for Pool resource in Director
1193 Origin: Alan Davis adavis@ruckus.com
1197 What: The current behavior is for the director to create all pools
1198 found in the configuration file in all catalogs. Add a
1199 Catalog directive to the Pool resource to specify which
1200 catalog to use for each pool definition.
1202 Why: This allows different catalogs to have different pool
1203 attributes and eliminates the side-effect of adding
1204 pools to catalogs that don't need/use them.
1209 Item n: Implement NDMP protocol support
1214 What: Network Data Management Protocol is implemented by a number of
1215 NAS filer vendors to enable backups using third-party
1218 Why: This would allow NAS filer backups in Bacula without incurring
1219 the overhead of NFS or SBM/CIFS.
1221 Notes: Further information is available:
1223 http://www.ndmp.org/wp/wp.shtml
1224 http://www.traakan.com/ndmjob/index.html
1226 There are currently no viable open-source NDMP
1227 implementations. There is a reference SDK and example
1228 app available from ndmp.org but it has problems
1229 compiling on recent Linux and Solaris OS'. The ndmjob
1230 reference implementation from Traakan is known to
1231 compile on Solaris 10.
1233 Notes (Kern): I am not at all in favor of this until NDMP becomes
1234 an Open Standard or until there are Open Source libraries
1235 that interface to it.
1237 Item n: make changing "spooldata=yes|no" possible for
1238 manual/interactive jobs
1240 Origin: Marc Schiffbauer <marc@schiffbauer.net>
1242 Date: 12 April 2007)
1246 What: Make it possible to modify the spooldata option
1247 for a job when being run from within the console.
1248 Currently it is possible to modify the backup level
1249 and the spooldata setting in a Schedule resource.
1250 It is also possible to modify the backup level when using
1251 the "run" command in the console.
1252 But it is currently not possible to to the same
1253 with "spooldata=yes|no" like:
1255 run job=MyJob level=incremental spooldata=yes
1257 Why: In some situations it would be handy to be able to switch
1258 spooldata on or off for interactive/manual jobs based on
1259 which data the admin expects or how fast the LAN/WAN
1260 connection currently is.
1264 ============= Empty Feature Request form ===========
1265 Item n: One line summary ...
1266 Date: Date submitted
1267 Origin: Name and email of originator.
1270 What: More detailed explanation ...
1272 Why: Why it is important ...
1274 Notes: Additional notes or features (omit if not used)
1275 ============== End Feature Request form ==============