3 Bacula Projects Roadmap
4 Status updated 15 December 2006
8 Item 1: Accurate restoration of renamed/deleted files
9 Item 2: Implement a Bacula GUI/management tool.
10 Item 3: Implement Base jobs.
11 Item 4: Implement from-client and to-client on restore command line.
12 Item 5: Implement creation and maintenance of copy pools
13 Item 6: Merge multiple backups (Synthetic Backup or Consolidation).
14 Item 8: Deletion of Disk-Based Bacula Volumes
15 Item 9: Implement a Python interface to the Bacula catalog.
16 Item 10: Archival (removal) of User Files to Tape
17 Item 11: Add Plug-ins to the FileSet Include statements.
18 Item 12: Implement more Python events in Bacula.
19 Item 13: Quick release of FD-SD connection after backup.
20 Item 14: Implement huge exclude list support using hashing.
21 Item 15: Allow skipping execution of Jobs
22 Item 16: Tray monitor window cleanups
23 Item 17: Split documentation
24 Item 18: Automatic promotion of backup levels
25 Item 19: Add an override in Schedule for Pools based on backup types.
26 Item 20: An option to operate on all pools with update vol parameters
27 Item 21: Include JobID in spool file name
28 Item 22: Include timestamp of job launch in "stat clients" output
29 Item 23: Message mailing based on backup types
30 Item 24: Allow inclusion/exclusion of files in a fileset by creation/mod times
31 Item 25: Add a scheduling syntax that permits weekly rotations
32 Item 26: Improve Bacula's tape and drive usage and cleaning management.
33 Item 27: Implement support for stacking arbitrary stream filters, sinks.
34 Item 28: Allow FD to initiate a backup
35 Item 29: Directive/mode to backup only file changes, not entire file
36 Item 30: Automatic disabling of devices
37 Item 31: Incorporation of XACML2/SAML2 parsing
38 Item 32: Clustered file-daemons
39 Item 33: Commercial database support
41 Item 35: Filesystem watch triggered backup.
42 Item 36: Implement multiple numeric backup levels as supported by dump
45 Below, you will find more information on future projects:
47 Item 1: Accurate restoration of renamed/deleted files
48 Date: 28 November 2005
49 Origin: Martin Simmons (martin at lispworks dot com)
50 Status: Robert Nelson will implement this
52 What: When restoring a fileset for a specified date (including "most
53 recent"), Bacula should give you exactly the files and directories
54 that existed at the time of the last backup prior to that date.
56 Currently this only works if the last backup was a Full backup.
57 When the last backup was Incremental/Differential, files and
58 directories that have been renamed or deleted since the last Full
59 backup are not currently restored correctly. Ditto for files with
60 extra/fewer hard links than at the time of the last Full backup.
62 Why: Incremental/Differential would be much more useful if this worked.
64 Notes: Merging of multiple backups into a single one seems to
65 rely on this working, otherwise the merged backups will not be
66 truly equivalent to a Full backup.
68 Kern: notes shortened. This can be done without the need for
69 inodes. It is essentially the same as the current Verify job,
70 but one additional database record must be written, which does
71 not need any database change.
73 Kern: see if we can correct restoration of directories if
74 replace=ifnewer is set. Currently, if the directory does not
75 exist, a "dummy" directory is created, then when all the files
76 are updated, the dummy directory is newer so the real values
79 Item 2: Implement a Bacula GUI/management tool.
84 What: Implement a Bacula console, and management tools
85 probably using Qt3 and C++.
87 Why: Don't we already have a wxWidgets GUI? Yes, but
88 it is written in C++ and changes to the user interface
89 must be hand tailored using C++ code. By developing
90 the user interface using Qt designer, the interface
91 can be very easily updated and most of the new Python
92 code will be automatically created. The user interface
93 changes become very simple, and only the new features
94 must be implement. In addition, the code will be in
95 Python, which will give many more users easy (or easier)
96 access to making additions or modifications.
98 Notes: There is a partial Python-GTK implementation
99 Lucas Di Pentima <lucas at lunix dot com dot ar> but
100 it is no longer being developed.
103 Item 3: Implement Base jobs.
104 Date: 28 October 2005
108 What: A base job is sort of like a Full save except that you
109 will want the FileSet to contain only files that are
110 unlikely to change in the future (i.e. a snapshot of
111 most of your system after installing it). After the
112 base job has been run, when you are doing a Full save,
113 you specify one or more Base jobs to be used. All
114 files that have been backed up in the Base job/jobs but
115 not modified will then be excluded from the backup.
116 During a restore, the Base jobs will be automatically
117 pulled in where necessary.
119 Why: This is something none of the competition does, as far as
120 we know (except perhaps BackupPC, which is a Perl program that
121 saves to disk only). It is big win for the user, it
122 makes Bacula stand out as offering a unique
123 optimization that immediately saves time and money.
124 Basically, imagine that you have 100 nearly identical
125 Windows or Linux machine containing the OS and user
126 files. Now for the OS part, a Base job will be backed
127 up once, and rather than making 100 copies of the OS,
128 there will be only one. If one or more of the systems
129 have some files updated, no problem, they will be
130 automatically restored.
132 Notes: Huge savings in tape usage even for a single machine.
133 Will require more resources because the DIR must send
134 FD a list of files/attribs, and the FD must search the
135 list and compare it for each file to be saved.
137 Item 4: Implement from-client and to-client on restore command line.
138 Date: 11 December 2006
139 Origin: Discussion on Bacula-users entitled 'Scripted restores to
140 different clients', December 2006
141 Status: New feature request
143 What: While using bconsole interactively, you can specify the client
144 that a backup job is to be restored for, and then you can
145 specify later a different client to send the restored files
146 back to. However, using the 'restore' command with all options
147 on the command line, this cannot be done, due to the ambiguous
148 'client' parameter. Additionally, this parameter means different
149 things depending on if it's specified on the command line or
150 afterwards, in the Modify Job screens.
152 Why: This feature would enable restore jobs to be more completely
153 automated, for example by a web or GUI front-end.
155 Notes: client can also be implied by specifying the jobid on the command
158 Item 5: Implement creation and maintenance of copy pools
159 Date: 27 November 2005
160 Origin: David Boyes (dboyes at sinenomine dot net)
163 What: I would like Bacula to have the capability to write copies
164 of backed-up data on multiple physical volumes selected
165 from different pools without transferring the data
166 multiple times, and to accept any of the copy volumes
167 as valid for restore.
169 Why: In many cases, businesses are required to keep offsite
170 copies of backup volumes, or just wish for simple
171 protection against a human operator dropping a storage
172 volume and damaging it. The ability to generate multiple
173 volumes in the course of a single backup job allows
174 customers to simple check out one copy and send it
175 offsite, marking it as out of changer or otherwise
176 unavailable. Currently, the library and magazine
177 management capability in Bacula does not make this process
180 Restores would use the copy of the data on the first
181 available volume, in order of copy pool chain definition.
183 This is also a major scalability issue -- as the number of
184 clients increases beyond several thousand, and the volume
185 of data increases, transferring the data multiple times to
186 produce additional copies of the backups will become
187 physically impossible due to transfer speed
188 issues. Generating multiple copies at server side will
189 become the only practical option.
191 How: I suspect that this will require adding a multiplexing
192 SD that appears to be a SD to a specific FD, but 1-n FDs
193 to the specific back end SDs managing the primary and copy
194 pools. Storage pools will also need to acquire parameters
195 to define the pools to be used for copies.
197 Notes: I would commit some of my developers' time if we can agree
198 on the design and behavior.
200 Item 6: Merge multiple backups (Synthetic Backup or Consolidation).
201 Origin: Marc Cousin and Eric Bollengier
202 Date: 15 November 2005
203 Status: Waiting implementation. Depends on first implementing
204 project Item 2 (Migration) which is now done.
206 What: A merged backup is a backup made without connecting to the Client.
207 It would be a Merge of existing backups into a single backup.
208 In effect, it is like a restore but to the backup medium.
210 For instance, say that last Sunday we made a full backup. Then
211 all week long, we created incremental backups, in order to do
212 them fast. Now comes Sunday again, and we need another full.
213 The merged backup makes it possible to do instead an incremental
214 backup (during the night for instance), and then create a merged
215 backup during the day, by using the full and incrementals from
216 the week. The merged backup will be exactly like a full made
217 Sunday night on the tape, but the production interruption on the
218 Client will be minimal, as the Client will only have to send
221 In fact, if it's done correctly, you could merge all the
222 Incrementals into single Incremental, or all the Incrementals
223 and the last Differential into a new Differential, or the Full,
224 last differential and all the Incrementals into a new Full
225 backup. And there is no need to involve the Client.
227 Why: The benefit is that :
228 - the Client just does an incremental ;
229 - the merged backup on tape is just as a single full backup,
230 and can be restored very fast.
232 This is also a way of reducing the backup data since the old
233 data can then be pruned (or not) from the catalog, possibly
234 allowing older volumes to be recycled
236 Item 8: Deletion of Disk-Based Bacula Volumes
238 Origin: Ross Boylan <RossBoylan at stanfordalumni dot org> (edited
242 What: Provide a way for Bacula to automatically remove Volumes
243 from the filesystem, or optionally to truncate them.
244 Obviously, the Volume must be pruned prior removal.
246 Why: This would allow users more control over their Volumes and
247 prevent disk based volumes from consuming too much space.
249 Notes: The following two directives might do the trick:
251 Volume Data Retention = <time period>
252 Remove Volume After = <time period>
254 The migration project should also remove a Volume that is
255 migrated. This might also work for tape Volumes.
257 Item 9: Implement a Python interface to the Bacula catalog.
258 Date: 28 October 2005
262 What: Implement an interface for Python scripts to access
263 the catalog through Bacula.
265 Why: This will permit users to customize Bacula through
268 Item 10: Archival (removal) of User Files to Tape
272 Origin: Ray Pengelly [ray at biomed dot queensu dot ca
275 What: The ability to archive data to storage based on certain parameters
276 such as age, size, or location. Once the data has been written to
277 storage and logged it is then pruned from the originating
278 filesystem. Note! We are talking about user's files and not
281 Why: This would allow fully automatic storage management which becomes
282 useful for large datastores. It would also allow for auto-staging
283 from one media type to another.
285 Example 1) Medical imaging needs to store large amounts of data.
286 They decide to keep data on their servers for 6 months and then put
287 it away for long term storage. The server then finds all files
288 older than 6 months writes them to tape. The files are then removed
291 Example 2) All data that hasn't been accessed in 2 months could be
292 moved from high-cost, fibre-channel disk storage to a low-cost
293 large-capacity SATA disk storage pool which doesn't have as quick of
294 access time. Then after another 6 months (or possibly as one
295 storage pool gets full) data is migrated to Tape.
297 Item 11: Add Plug-ins to the FileSet Include statements.
298 Date: 28 October 2005
300 Status: Partially coded in 1.37 -- much more to do.
302 What: Allow users to specify wild-card and/or regular
303 expressions to be matched in both the Include and
304 Exclude directives in a FileSet. At the same time,
305 allow users to define plug-ins to be called (based on
306 regular expression/wild-card matching).
308 Why: This would give the users the ultimate ability to control
309 how files are backed up/restored. A user could write a
310 plug-in knows how to backup his Oracle database without
311 stopping/starting it, for example.
313 Item 12: Implement more Python events in Bacula.
314 Date: 28 October 2005
318 What: Allow Python scripts to be called at more places
319 within Bacula and provide additional access to Bacula
322 Why: This will permit users to customize Bacula through
330 Also add a way to get a listing of currently running
331 jobs (possibly also scheduled jobs).
334 Item 13: Quick release of FD-SD connection after backup.
335 Origin: Frank Volf (frank at deze dot org)
336 Date: 17 November 2005
339 What: In the Bacula implementation a backup is finished after all data
340 and attributes are successfully written to storage. When using a
341 tape backup it is very annoying that a backup can take a day,
342 simply because the current tape (or whatever) is full and the
343 administrator has not put a new one in. During that time the
344 system cannot be taken off-line, because there is still an open
345 session between the storage daemon and the file daemon on the
348 Although this is a very good strategy for making "safe backups"
349 This can be annoying for e.g. laptops, that must remain
350 connected until the backup is completed.
352 Using a new feature called "migration" it will be possible to
353 spool first to harddisk (using a special 'spool' migration
354 scheme) and then migrate the backup to tape.
356 There is still the problem of getting the attributes committed.
357 If it takes a very long time to do, with the current code, the
358 job has not terminated, and the File daemon is not freed up. The
359 Storage daemon should release the File daemon as soon as all the
360 file data and all the attributes have been sent to it (the SD).
361 Currently the SD waits until everything is on tape and all the
362 attributes are transmitted to the Director before signaling
363 completion to the FD. I don't think I would have any problem
364 changing this. The reason is that even if the FD reports back to
365 the Dir that all is OK, the job will not terminate until the SD
366 has done the same thing -- so in a way keeping the SD-FD link
367 open to the very end is not really very productive ...
369 Why: Makes backup of laptops much faster.
373 Item 14: Implement huge exclude list support using hashing.
374 Date: 28 October 2005
378 What: Allow users to specify very large exclude list (currently
379 more than about 1000 files is too many).
381 Why: This would give the users the ability to exclude all
382 files that are loaded with the OS (e.g. using rpms
383 or debs). If the user can restore the base OS from
384 CDs, there is no need to backup all those files. A
385 complete restore would be to restore the base OS, then
386 do a Bacula restore. By excluding the base OS files, the
387 backup set will be *much* smaller.
390 Item 15: Allow skipping execution of Jobs
391 Date: 29 November 2005
392 Origin: Florian Schnabel <florian.schnabel at docufy dot de>
395 What: An easy option to skip a certain job on a certain date.
396 Why: You could then easily skip tape backups on holidays. Especially
397 if you got no autochanger and can only fit one backup on a tape
398 that would be really handy, other jobs could proceed normally
399 and you won't get errors that way.
402 Item 16: Tray monitor window cleanups
403 Origin: Alan Brown ajb2 at mssl dot ucl dot ac dot uk
406 What: Resizeable and scrollable windows in the tray monitor.
408 Why: With multiple clients, or with many jobs running, the displayed
409 window often ends up larger than the available screen, making
410 the trailing items difficult to read.
413 Item 17: Split documentation
414 Origin: Maxx <maxxatworkat gmail dot com>
418 What: Split documentation in several books
420 Why: Bacula manual has now more than 600 pages, and looking for
421 implementation details is getting complicated. I think
422 it would be good to split the single volume in two or
425 1) Introduction, requirements and tutorial, typically
426 are useful only until first installation time
428 2) Basic installation and configuration, with all the
429 gory details about the directives supported 3)
430 Advanced Bacula: testing, troubleshooting, GUI and
431 ancillary programs, security managements, scripting,
436 Item 18: Automatic promotion of backup levels
437 Date: 19 January 2006
438 Origin: Adam Thornton <athornton@sinenomine.net>
441 What: Amanda has a feature whereby it estimates the space that a
442 differential, incremental, and full backup would take. If the
443 difference in space required between the scheduled level and the next
444 level up is beneath some user-defined critical threshold, the backup
445 level is bumped to the next type. Doing this minimizes the number of
446 volumes necessary during a restore, with a fairly minimal cost in
449 Why: I know at least one (quite sophisticated and smart) user
450 for whom the absence of this feature is a deal-breaker in terms of
451 using Bacula; if we had it it would eliminate the one cool thing
452 Amanda can do and we can't (at least, the one cool thing I know of).
455 Item 19: Add an override in Schedule for Pools based on backup types.
457 Origin: Chad Slater <chad.slater@clickfox.com>
460 What: Adding a FullStorage=BigTapeLibrary in the Schedule resource
461 would help those of us who use different storage devices for different
462 backup levels cope with the "auto-upgrade" of a backup.
464 Why: Assume I add several new device to be backed up, i.e. several
465 hosts with 1TB RAID. To avoid tape switching hassles, incrementals are
466 stored in a disk set on a 2TB RAID. If you add these devices in the
467 middle of the month, the incrementals are upgraded to "full" backups,
468 but they try to use the same storage device as requested in the
469 incremental job, filling up the RAID holding the differentials. If we
470 could override the Storage parameter for full and/or differential
471 backups, then the Full job would use the proper Storage device, which
472 has more capacity (i.e. a 8TB tape library.
474 Item 20: An option to operate on all pools with update vol parameters
475 Origin: Dmitriy Pinchukov <absh@bossdev.kiev.ua>
479 What: When I do update -> Volume parameters -> All Volumes
480 from Pool, then I have to select pools one by one. I'd like
481 console to have an option like "0: All Pools" in the list of
484 Why: I have many pools and therefore unhappy with manually
485 updating each of them using update -> Volume parameters -> All
486 Volumes from Pool -> pool #.
490 Item 21: Include JobID in spool file name
491 Origin: Mark Bergman <mark.bergman@uphs.upenn.edu>
492 Date: Tue Aug 22 17:13:39 EDT 2006
495 What: Change the name of the spool file to include the JobID
497 Why: JobIDs are the common key used to refer to jobs, yet the
498 spoolfile name doesn't include that information. The date/time
499 stamp is useful (and should be retained).
503 Item 22: Include timestamp of job launch in "stat clients" output
504 Origin: Mark Bergman <mark.bergman@uphs.upenn.edu>
505 Date: Tue Aug 22 17:13:39 EDT 2006
508 What: The "stat clients" command doesn't include any detail on when
509 the active backup jobs were launched.
511 Why: Including the timestamp would make it much easier to decide whether
512 a job is running properly.
514 Notes: It may be helpful to have the output from "stat clients" formatted
515 more like that from "stat dir" (and other commands), in a column
516 format. The per-client information that's currently shown (level,
517 client name, JobId, Volume, pool, device, Files, etc.) is good, but
518 somewhat hard to parse (both programmatically and visually),
519 particularly when there are many active clients.
523 Item 23: Message mailing based on backup types
524 Origin: Evan Kaufman <evan.kaufman@gmail.com>
525 Date: January 6, 2006
528 What: In the "Messages" resource definitions, allowing messages
529 to be mailed based on the type (backup, restore, etc.) and level
530 (full, differential, etc) of job that created the originating
533 Why: It would, for example, allow someone's boss to be emailed
534 automatically only when a Full Backup job runs, so he can
535 retrieve the tapes for offsite storage, even if the IT dept.
536 doesn't (or can't) explicitly notify him. At the same time, his
537 mailbox wouldnt be filled by notifications of Verifies, Restores,
538 or Incremental/Differential Backups (which would likely be kept
541 Notes: One way this could be done is through additional message types, for example:
544 # email the boss only on full system backups
545 Mail = boss@mycompany.com = full, !incremental, !differential, !restore,
547 # email us only when something breaks
548 MailOnError = itdept@mycompany.com = all
552 Item 24: Allow inclusion/exclusion of files in a fileset by creation/mod times
553 Origin: Evan Kaufman <evan.kaufman@gmail.com>
554 Date: January 11, 2006
557 What: In the vein of the Wild and Regex directives in a Fileset's
558 Options, it would be helpful to allow a user to include or exclude
559 files and directories by creation or modification times.
561 You could factor the Exclude=yes|no option in much the same way it
562 affects the Wild and Regex directives. For example, you could exclude
563 all files modified before a certain date:
567 Modified Before = ####
570 Or you could exclude all files created/modified since a certain date:
574 Created Modified Since = ####
577 The format of the time/date could be done several ways, say the number
578 of seconds since the epoch:
579 1137008553 = Jan 11 2006, 1:42:33PM # result of `date +%s`
581 Or a human readable date in a cryptic form:
582 20060111134233 = Jan 11 2006, 1:42:33PM # YYYYMMDDhhmmss
584 Why: I imagine a feature like this could have many uses. It would
585 allow a user to do a full backup while excluding the base operating
586 system files, so if I installed a Linux snapshot from a CD yesterday,
587 I'll *exclude* all files modified *before* today. If I need to
588 recover the system, I use the CD I already have, plus the tape backup.
589 Or if, say, a Windows client is hit by a particularly corrosive
590 virus, and I need to *exclude* any files created/modified *since* the
593 Notes: Of course, this feature would work in concert with other
594 in/exclude rules, and wouldnt override them (or each other).
596 Notes: The directives I'd imagine would be along the lines of
597 "[Created] [Modified] [Before|Since] = <date>".
598 So one could compare against 'ctime' and/or 'mtime', but ONLY 'before'
602 Item 25: Add a scheduling syntax that permits weekly rotations
603 Date: 15 December 2006
604 Origin: Gregory Brauer (greg at wildbrain dot com)
607 What: Currently, Bacula only understands how to deal with weeks of the
608 month or weeks of the year in schedules. This makes it impossible
609 to do a true weekly rotation of tapes. There will always be a
610 discontinuity that will require disruptive manual intervention at
611 least monthly or yearly because week boundaries never align with
612 month or year boundaries.
614 A solution would be to add a new syntax that defines (at least)
615 a start timestamp, and repetition period.
617 Why: Rotated backups done at weekly intervals are useful, and Bacula
618 cannot currently do them without extensive hacking.
620 Notes: Here is an example syntax showing a 3-week rotation where full
621 Backups would be performed every week on Saturday, and an
622 incremental would be performed every week on Tuesday. Each
623 set of tapes could be removed from the loader for the following
624 two cycles before coming back and being reused on the third
625 week. Since the execution times are determined by intervals
626 from a given point in time, there will never be any issues with
627 having to adjust to any sort of arbitrary time boundary. In
628 the example provided, I even define the starting schedule
629 as crossing both a year and a month boundary, but the run times
630 would be based on the "Repeat" value and would therefore happen
635 Name = "Week 1 Rotation"
636 #Saturday. Would run Dec 30, Jan 20, Feb 10, etc.
640 Start = 2006-12-30 01:00
644 #Tuesday. Would run Jan 2, Jan 23, Feb 13, etc.
648 Start = 2007-01-02 01:00
655 Name = "Week 2 Rotation"
656 #Saturday. Would run Jan 6, Jan 27, Feb 17, etc.
660 Start = 2007-01-06 01:00
664 #Tuesday. Would run Jan 9, Jan 30, Feb 20, etc.
668 Start = 2007-01-09 01:00
675 Name = "Week 3 Rotation"
676 #Saturday. Would run Jan 13, Feb 3, Feb 24, etc.
680 Start = 2007-01-13 01:00
684 #Tuesday. Would run Jan 16, Feb 6, Feb 27, etc.
688 Start = 2007-01-16 01:00
695 Item 26: Improve Bacula's tape and drive usage and cleaning management.
696 Date: 8 November 2005, November 11, 2005
697 Origin: Adam Thornton <athornton at sinenomine dot net>,
698 Arno Lehmann <al at its-lehmann dot de>
701 What: Make Bacula manage tape life cycle information, tape reuse
702 times and drive cleaning cycles.
704 Why: All three parts of this project are important when operating
706 We need to know which tapes need replacement, and we need to
707 make sure the drives are cleaned when necessary. While many
708 tape libraries and even autoloaders can handle all this
709 automatically, support by Bacula can be helpful for smaller
710 (older) libraries and single drives. Limiting the number of
711 times a tape is used might prevent tape errors when using
712 tapes until the drives can't read it any more. Also, checking
713 drive status during operation can prevent some failures (as I
714 [Arno] had to learn the hard way...)
716 Notes: First, Bacula could (and even does, to some limited extent)
717 record tape and drive usage. For tapes, the number of mounts,
718 the amount of data, and the time the tape has actually been
719 running could be recorded. Data fields for Read and Write
720 time and Number of mounts already exist in the catalog (I'm
721 not sure if VolBytes is the sum of all bytes ever written to
722 that volume by Bacula). This information can be important
723 when determining which media to replace. The ability to mark
724 Volumes as "used up" after a given number of write cycles
725 should also be implemented so that a tape is never actually
726 worn out. For the tape drives known to Bacula, similar
727 information is interesting to determine the device status and
728 expected life time: Time it's been Reading and Writing, number
729 of tape Loads / Unloads / Errors. This information is not yet
730 recorded as far as I [Arno] know. A new volume status would
731 be necessary for the new state, like "Used up" or "Worn out".
732 Volumes with this state could be used for restores, but not
733 for writing. These volumes should be migrated first (assuming
734 migration is implemented) and, once they are no longer needed,
735 could be moved to a Trash pool.
737 The next step would be to implement a drive cleaning setup.
738 Bacula already has knowledge about cleaning tapes. Once it
739 has some information about cleaning cycles (measured in drive
740 run time, number of tapes used, or calender days, for example)
741 it can automatically execute tape cleaning (with an
742 autochanger, obviously) or ask for operator assistance loading
745 The final step would be to implement TAPEALERT checks not only
746 when changing tapes and only sending the information to the
747 administrator, but rather checking after each tape error,
748 checking on a regular basis (for example after each tape
749 file), and also before unloading and after loading a new tape.
750 Then, depending on the drives TAPEALERT state and the known
751 drive cleaning state Bacula could automatically schedule later
752 cleaning, clean immediately, or inform the operator.
754 Implementing this would perhaps require another catalog change
755 and perhaps major changes in SD code and the DIR-SD protocol,
756 so I'd only consider this worth implementing if it would
757 actually be used or even needed by many people.
759 Implementation of these projects could happen in three distinct
760 sub-projects: Measuring Tape and Drive usage, retiring
761 volumes, and handling drive cleaning and TAPEALERTs.
763 Item 27: Implement support for stacking arbitrary stream filters, sinks.
764 Date: 23 November 2006
765 Origin: Landon Fuller <landonf@threerings.net>
766 Status: Planning. Assigned to landonf.
769 Implement support for the following:
770 - Stacking arbitrary stream filters (eg, encryption, compression,
771 sparse data handling))
772 - Attaching file sinks to terminate stream filters (ie, write out
773 the resultant data to a file)
774 - Refactor the restoration state machine accordingly
777 The existing stream implementation suffers from the following:
778 - All state (compression, encryption, stream restoration), is
779 global across the entire restore process, for all streams. There are
780 multiple entry and exit points in the restoration state machine, and
781 thus multiple places where state must be allocated, deallocated,
782 initialized, or reinitialized. This results in exceptional complexity
783 for the author of a stream filter.
784 - The developer must enumerate all possible combinations of filters
785 and stream types (ie, win32 data with encryption, without encryption,
786 with encryption AND compression, etc).
789 This feature request only covers implementing the stream filters/
790 sinks, and refactoring the file daemon's restoration implementation
791 accordingly. If I have extra time, I will also rewrite the backup
792 implementation. My intent in implementing the restoration first is to
793 solve pressing bugs in the restoration handling, and to ensure that
794 the new restore implementation handles existing backups correctly.
796 I do not plan on changing the network or tape data structures to
797 support defining arbitrary stream filters, but supporting that
798 functionality is the ultimate goal.
800 Assistance with either code or testing would be fantastic.
802 Item 28: Allow FD to initiate a backup
803 Origin: Frank Volf (frank at deze dot org)
804 Date: 17 November 2005
807 What: Provide some means, possibly by a restricted console that
808 allows a FD to initiate a backup, and that uses the connection
809 established by the FD to the Director for the backup so that
810 a Director that is firewalled can do the backup.
812 Why: Makes backup of laptops much easier.
814 Item 29: Directive/mode to backup only file changes, not entire file
815 Date: 11 November 2005
816 Origin: Joshua Kugler <joshua dot kugler at uaf dot edu>
817 Marek Bajon <mbajon at bimsplus dot com dot pl>
820 What: Currently when a file changes, the entire file will be backed up in
821 the next incremental or full backup. To save space on the tapes
822 it would be nice to have a mode whereby only the changes to the
823 file would be backed up when it is changed.
825 Why: This would save lots of space when backing up large files such as
826 logs, mbox files, Outlook PST files and the like.
828 Notes: This would require the usage of disk-based volumes as comparing
829 files would not be feasible using a tape drive.
831 Item 30: Automatic disabling of devices
833 Origin: Peter Eriksson <peter at ifm.liu dot se>
836 What: After a configurable amount of fatal errors with a tape drive
837 Bacula should automatically disable further use of a certain
838 tape drive. There should also be "disable"/"enable" commands in
841 Why: On a multi-drive jukebox there is a possibility of tape drives
842 going bad during large backups (needing a cleaning tape run,
843 tapes getting stuck). It would be advantageous if Bacula would
844 automatically disable further use of a problematic tape drive
845 after a configurable amount of errors has occurred.
847 An example: I have a multi-drive jukebox (6 drives, 380+ slots)
848 where tapes occasionally get stuck inside the drive. Bacula will
849 notice that the "mtx-changer" command will fail and then fail
850 any backup jobs trying to use that drive. However, it will still
851 keep on trying to run new jobs using that drive and fail -
852 forever, and thus failing lots and lots of jobs... Since we have
853 many drives Bacula could have just automatically disabled
854 further use of that drive and used one of the other ones
857 Item 31: Incorporation of XACML2/SAML2 parsing
858 Date: 19 January 2006
859 Origin: Adam Thornton <athornton@sinenomine.net>
862 What: XACML is "eXtensible Access Control Markup Language" and
863 "SAML is the "Security Assertion Markup Language"--an XML standard
864 for making statements about identity and authorization. Having these
865 would give us a framework to approach ACLs in a generic manner, and
866 in a way flexible enough to support the four major sorts of ACLs I
867 see as a concern to Bacula at this point, as well as (probably) to
868 deal with new sorts of ACLs that may appear in the future.
870 Why: Bacula is beginning to need to back up systems with ACLs
871 that do not map cleanly onto traditional Unix permissions. I see
872 four sets of ACLs--in general, mutually incompatible with one
873 another--that we're going to need to deal with. These are: NTFS
874 ACLs, POSIX ACLs, NFSv4 ACLS, and AFS ACLS. (Some may question the
875 relevance of AFS; AFS is one of Sine Nomine's core consulting
876 businesses, and having a reputable file-level backup and restore
877 technology for it (as Tivoli is probably going to drop AFS support
878 soon since IBM no longer supports AFS) would be of huge benefit to
879 our customers; we'd most likely create the AFS support at Sine Nomine
880 for inclusion into the Bacula (and perhaps some changes to the
881 OpenAFS volserver) core code.)
883 Now, obviously, Bacula already handles NTFS just fine. However, I
884 think there's a lot of value in implementing a generic ACL model, so
885 that it's easy to support whatever particular instances of ACLs come
886 down the pike: POSIX ACLS (think SELinux) and NFSv4 are the obvious
887 things arriving in the Linux world in a big way in the near future.
888 XACML, although overcomplicated for our needs, provides this
889 framework, and we should be able to leverage other people's
890 implementations to minimize the amount of work *we* have to do to get
891 a generic ACL framework. Basically, the costs of implementation are
892 high, but they're largely both external to Bacula and already sunk.
895 Item 32: Clustered file-daemons
896 Origin: Alan Brown ajb2 at mssl dot ucl dot ac dot uk
899 What: A "virtual" filedaemon, which is actually a cluster of real ones.
901 Why: In the case of clustered filesystems (SAN setups, GFS, or OCFS2, etc)
902 multiple machines may have access to the same set of filesystems
904 For performance reasons, one may wish to initate backups from
905 several of these machines simultaneously, instead of just using
906 one backup source for the common clustered filesystem.
908 For obvious reasons, normally backups of $A-FD/$PATH and
909 B-FD/$PATH are treated as different backup sets. In this case
910 they are the same communal set.
912 Likewise when restoring, it would be easier to just specify
913 one of the cluster machines and let bacula decide which to use.
915 This can be faked to some extent using DNS round robin entries
916 and a virtual IP address, however it means "status client" will
917 always give bogus answers. Additionally there is no way of
918 spreading the load evenly among the servers.
920 What is required is something similar to the storage daemon
921 autochanger directives, so that Bacula can keep track of
922 operating backups/restores and direct new jobs to a "free"
927 Item 33: Commercial database support
928 Origin: Russell Howe <russell_howe dot wreckage dot org>
932 What: It would be nice for the database backend to support more
933 databases. I'm thinking of SQL Server at the moment, but I guess Oracle,
934 DB2, MaxDB, etc are all candidates. SQL Server would presumably be
935 implemented using FreeTDS or maybe an ODBC library?
937 Why: We only really have one database server, which is MS SQL Server
938 2000. Maintaining a second one for the backup software (we grew out of
939 SQLite, which I liked, but which didn't work so well with our database
940 size). We don't really have a machine with the resources to run
941 postgres, and would rather only maintain a single DBMS. We're stuck with
942 SQL Server because pretty much all the company's custom applications
943 (written by consultants) are locked into SQL Server 2000. I can imagine
944 this scenario is fairly common, and it would be nice to use the existing
945 properly specced database server for storing Bacula's catalog, rather
946 than having to run a second DBMS.
949 Item 34: Archive data
951 Origin: calvin streeting calvin at absentdream dot com
954 What: The abilty to archive to media (dvd/cd) in a uncompressed format
955 for dead filing (archiving not backing up)
957 Why: At my works when jobs are finished and moved off of the main file
958 servers (raid based systems) onto a simple linux file server (ide based
959 system) so users can find old information without contacting the IT
962 So this data dosn't realy change it only gets added to,
963 But it also needs backing up. At the moment it takes
964 about 8 hours to back up our servers (working data) so
965 rather than add more time to existing backups i am trying
966 to implement a system where we backup the acrhive data to
967 cd/dvd these disks would only need to be appended to
968 (burn only new/changed files to new disks for off site
969 storage). basialy understand the differnce between
970 achive data and live data.
972 Notes: Scan the data and email me when it needs burning divide
973 into predifind chunks keep a recored of what is on what
974 disk make me a label (simple php->mysql=>pdf stuff) i
975 could do this bit ability to save data uncompresed so
976 it can be read in any other system (future proof data)
977 save the catalog with the disk as some kind of menu
980 Item 35: Filesystem watch triggered backup.
982 Origin: Jesper Krogh <jesper@krogh.cc>
983 Status: Unimplemented, depends probably on "client initiated backups"
985 What: With inotify and similar filesystem triggeret notification
986 systems is it possible to have the file-daemon to monitor
987 filesystem changes and initiate backup.
989 Why: There are 2 situations where this is nice to have.
990 1) It is possible to get a much finer-grained backup than
991 the fixed schedules used now.. A file created and deleted
992 a few hours later, can automatically be caught.
994 2) The introduced load on the system will probably be
995 distributed more even on the system.
997 Notes: This can be combined with configration that specifies
998 something like: "at most every 15 minutes or when changes
1001 Kern Notes: I would rather see this implemented by an external program
1002 that monitors the Filesystem changes, then uses the console
1003 to start the appropriate job.
1005 Item 36: Implement multiple numeric backup levels as supported by dump
1007 Origin: Daniel Rich <drich@employees.org>
1009 What: Dump allows specification of backup levels numerically instead of just
1010 "full", "incr", and "diff". In this system, at any given level, all
1011 files are backed up that were were modified since the last backup of a
1012 higher level (with 0 being the highest and 9 being the lowest). A
1013 level 0 is therefore equivalent to a full, level 9 an incremental, and
1014 the levels 1 through 8 are varying levels of differentials. For
1015 bacula's sake, these could be represented as "full", "incr", and
1016 "diff1", "diff2", etc.
1018 Why: Support of multiple backup levels would provide for more advanced backup
1019 rotation schemes such as "Towers of Hanoi". This would allow better
1020 flexibility in performing backups, and can lead to shorter recover
1023 Notes: Legato Networker supports a similar system with full, incr, and 1-9 as
1025 Item 1: Implement a server-side compression feature
1026 Date: 18 December 2006
1027 Origin: Vadim A. Umanski , e-mail umanski@ext.ru
1029 What: The ability to compress backup data on server receiving data
1030 instead of doing that on client sending data.
1031 Why: The need is practical. I've got some machines that can send
1032 data to the network 4 or 5 times faster than compressing
1033 them (I've measured that). They're using fast enough SCSI/FC
1034 disk subsystems but rather slow CPUs (ex. UltraSPARC II).
1035 And the backup server has got a quite fast CPUs (ex. Dual P4
1036 Xeons) and quite a low load. When you have 20, 50 or 100 GB
1037 of raw data - running a job 4 to 5 times faster - that
1038 really matters. On the other hand, the data can be
1039 compressed 50% or better - so losing twice more space for
1040 disk backup is not good at all. And the network is all mine
1041 (I have a dedicated management/provisioning network) and I
1042 can get as high bandwidth as I need - 100Mbps, 1000Mbps...
1043 That's why the server-side compression feature is needed!
1046 Item 1: Cause daemons to use a specific IP address to source communications
1047 Origin: Bill Moran <wmoran@collaborativefusion.com>
1050 What: Cause Bacula daemons (dir, fd, sd) to always use the ip address
1051 specified in the [DIR|DF|SD]Addr directive as the source IP
1052 for initiating communication.
1053 Why: On complex networks, as well as extremely secure networks, it's
1054 not unusual to have multiple possible routes through the network.
1055 Often, each of these routes is secured by different policies
1056 (effectively, firewalls allow or deny different traffic depending
1057 on the source address)
1058 Unfortunately, it can sometimes be difficult or impossible to
1059 represent this in a system routing table, as the result is
1060 excessive subnetting that quickly exhausts available IP space.
1061 The best available workaround is to provide multiple IPs to
1062 a single machine that are all on the same subnet. In order
1063 for this to work properly, applications must support the ability
1064 to bind outgoing connections to a specified address, otherwise
1065 the operating system will always choose the first IP that
1066 matches the required route.
1067 Notes: Many other programs support this. For example, the following
1068 can be configured in BIND:
1069 query-source address 10.0.0.1;
1070 transfer-source 10.0.0.2;
1071 Which means queries from this server will always come from
1072 10.0.0.1 and zone transfers will always originate from
1075 Kern notes: I think this would add very little functionality, but a *lot* of
1076 additional overhead to Bacula.
1080 ============= Empty Feature Request form ===========
1081 Item n: One line summary ...
1082 Date: Date submitted
1083 Origin: Name and email of originator.
1086 What: More detailed explanation ...
1088 Why: Why it is important ...
1090 Notes: Additional notes or features (omit if not used)
1091 ============== End Feature Request form ==============