4 Documentation to do: (a little bit at a time)
5 - Document running a test version.
6 - Document query file format.
7 - Document static linking
8 - Document how to automatically backup all local partitions
9 - Document problems with Verify and pruning.
10 - Document how to use multiple databases.
11 - Document Maximum File Size
14 Testing to do: (painful)
15 - that ALL console command line options work and are always implemented
16 - blocksize recognition code.
17 - multiple simultaneous Volumes
20 - Fix "access not allowed" for backup of files on WinXP.
21 - Fix Error: bnet.c:408 gethostbyname() for lpmatou failed: ERR=Operation not permited
23 - Add code if there is no mtio.h.
24 - Add a section to the doc on Manual cycling of Volumes.
25 - Look at purge jobs volume (at least document it, and see if it is
27 - Add list volumes pool=*
28 - Add pool= to "list media" in ua_output.c
29 - Figure out some way to specify a retention period for files
30 that no longer exist on the machine -- so that we maintain
31 say backups for 30 days, but if the file is deleted, we maintain
32 the last copy for 1 year.
33 - Have Bacula "poll" the tape to see if it is there.
34 - Add more of the config info to the tape label.
35 - Implement bar code reader for autochangers
36 - Add a default File storage so that new users can do backup
37 and restores right away.
38 - Implement a Mount Command and an Unmount Command where
39 the users could specify a system command to be performed
40 to do the mount, after which Bacula could attempt to
41 read the device. This is for Removeable media such as a
43 - Most likely, this mount command would be invoked explicitly
44 by the user using the current Console "mount" and "unmount"
45 commands -- the Storage Daemon would do the right thing
46 depending on the exact nature of the device.
47 - As with tape drives, when Bacula wanted a new removable
48 disk mounted, it would unmount the old one, and send a message
49 to the user, who would then use "mount" as described above
50 once he had actually inserted the disk.
52 - Make some way so that if a machine is skipped because it is not up
53 that Bacula will continue retrying for a specified period of time --
55 - Figure out some way to estimate output size and to avoid splitting
56 a backup across two Volumes -- this could be useful for writing CDROMs
57 where you really prefer not to have it split -- not serious.
58 - Add chflags() code for FreeBSD file flags
59 - Add RunBeforeJob and RunAfterJob to the Client program.
60 - Have SD compute MD5 or SHA1 and compare to what FD computes.
61 - Make VolumeToCatalog calculate an MD5 or SHA1 from the
62 actual data on the Volume and compare it.
63 - Implement FileOptions (see end of this document)
64 - Implement Bacula plugins -- design API
65 - Make bcopy read through bad tape records.
66 - Fix read_record to handle multiple sessions.
67 - Program files (i.e. execute a program to read/write files).
68 Pass read date of last backup, size of file last time.
69 - Add Signature type to File DB record.
70 - Make Restore report an error if FD or SD term codes are not OK.
71 - CD into subdirectory when open()ing files for backup to
72 speed up things. Test with testfind().
73 - Add prefixlinks to where or not where absolute links to FD.
74 - Look at handling <> in smtp doesn't work with exim.
75 - Priority job to go to top of list.
76 - Implement Bar code handling
77 - Why is catreq.c:111 Find vol called twice for a job?
78 - Find out why Full saves run slower and slower (hashing?)
79 - Why are save/restore of device different sizes (sparse?) Yup! Fix it.
80 - Implement some way for the Console to dynamically create a job.
81 - Restore to a particular time -- e.g. before date, after date.
82 - Implement disk spooling
83 - Implement finer multiprocessing options.
84 - Solaris -I on tar for include list
85 - Enable avoid backing up archive device (findlib/find_one.c:128)
86 - Need a verbose mode in restore, perhaps to bsr.
87 - bscan without -v is too quiet -- perhaps show jobs.
88 - Add code to reject whole blocks if not wanted on restore.
89 - Start working on Base jobs.
90 - Make sure the MaxVolFiles is fully implemented in SD
91 - Check if both CatalogFiles and UseCatalog are set to SD.
92 - Check if we can increase Bacula FD priorty in Win2000
93 - Need return status on read_cb() from read_records(). Need multiple
94 records -- one per Job, maybe a JCR or some other structure with
96 - Figure out how to do a bare metal Windows restore
97 - Put system type returned by FD into catalog.
98 - Need to specify MaximumConcurrentJobs in the Job resource.
99 - Possibly add email to Watchdog if drive is unmounted too
100 long and a job is waiting on the drive.
101 - Strip trailing slashes from Include directory names in the FD.
102 - Use read_record.c in SD code.
103 - Why don't we get an error message from Win32 FD when bootstrap
104 file cannot be created for restore command?
105 - When Marking a file in Restore that is a hard link, also
106 mark the link so that the data will be reloaded.
107 - Restore program that errors in SD due to no tape reports
108 OK incorrectly in output.
109 - After unmount, if restore job started, ask to mount.
110 - Convert all %x substitution variables, which are hard to remember
111 and read to %(variable-name). Idea from TMDA.
112 - Add JobLevel in FD status (but make sure it is defined).
113 - Make Pool resource handle Counter resources.
114 - Remove NextId for SQLite. Optimize.
115 - Strip trailing / from Include
116 - Move all SQL statements into a single location.
117 - Add UA rc and history files.
118 - put termcap (used by console) in ./configure and
119 allow -with-termcap-dir.
120 - Enhance time and size scanning routines.
121 - Fix Autoprune for Volumes to respect need for full save.
122 - Fix Win32 config file definition name on /install
123 - No READLINE_SRC if found in alternate directory.
124 - Add Client FS/OS id (Linux, Win95/98, ...).
125 - Test a second language e.g. french.
126 - Compare tape to Client files (attributes, or attributes and data)
127 - Make all database Ids 64 bit.
128 - Write an applet for Linux.
129 - Add estimate to Console commands
130 - Find solution to blank filename (i.e. path only) problem.
131 - Implement new daemon communications protocol.
132 - Remove PoolId from Job table, it exists in Media.
133 - Allow console commands to detach or run in background.
134 - Fix status delay on storage daemon during rewind.
135 - Add SD message variables to control operator wait time
136 - Maximum Operator Wait
137 - Minimum Message Interval
138 - Maximum Message Interval
139 - Send Operator message when cannot read tape label.
140 - Verify level=Volume (scan only), level=Data (compare of data to file).
141 Verify level=Catalog, level=InitCatalog
143 - Add keyword search to show command in Console.
144 - Fix Win2000 error with no messages during startup.
145 - Events : tape has more than xxx bytes.
146 - Restrict characters permitted in a Resource name.
147 - Complete code in Bacula Resources -- this will permit
148 reading a new config file at any time.
149 - Handle ctl-c in Console
150 - Implement LabelTemplate (at least first cut).
151 - Implement script driven addition of File daemon to config files.
152 - Think about how to make Bacula work better with File (non-tape) archives.
153 - Write Unix emulator for Windows.
155 - Implement new serialize subroutines
156 send(socket, "string", &Vol, "uint32", &i, NULL)
157 - Audit all UA commands to ensure that we always prompt where possible.
158 - If ./btape is called without /dev, assume argument is a Storage resource name.
159 - Put memory utilization in Status output of each daemon
160 if full status requested or if some level of debug on.
161 - Make database type selectable by .conf files i.e. at runtime
162 - gethostbyname failure in bnet_connect() continues
163 generating errors -- should stop.
164 - Set flag for uname -a. Add to Volume label.
165 - Implement throttled work queue.
166 - Check for EOT at ENOSPC or EIO or ENXIO (unix Pc)
167 - Allow multiple Storage specifications (or multiple names on
168 a single Storage specification) in the Job record. Thus a job
169 can be backed up to a number of storage devices.
170 - Implement dump label to UA
171 - Concept of VolumeSet during restore which is a list
172 of Volume names needed.
173 - Restore files modified after date
174 - Restore file modified before date
175 - Emergency restore info:
177 - Backup working directory
179 - Restore -- do nothing but show what would happen
180 - SET LD_RUN_PATH=$HOME/mysql/lib/mysql
181 - Implement Restore FileSet=
182 - Create a protocol.h and protocol.c where all protocol messages
184 - If SD cannot open a drive, make it periodically retry.
185 - Remove duplicate fields from jcr (e.g. jcr.level and jcr.jr.Level, ...).
186 - Timout a job or terminate if link goes down, or reopen link and query.
187 - Find general solution for sscanf size problems (as well
188 as sprintf. Do at run time?
189 - Concept of precious tapes (cannot be reused).
190 - Make bcopy copy with a single tape drive.
191 - Permit changing ownership during restore.
193 - Autolabel should be specified by DIR instead of SD.
194 - Find out how to get the system tape block limits, e.g.:
195 Apr 22 21:22:10 polymatou kernel: st1: Block limits 1 - 245760 bytes.
196 Apr 22 21:22:10 polymatou kernel: st0: Block limits 2 - 16777214 bytes.
199 - AutoScan (check checksum of tape)
200 - Format command = "format /dev/nst0"
204 - Seek resolution (usually corresponds to buffer size)
205 - EODErrorCode=ENOSPC or code
206 - Partial Read error code
207 - Partial write error code
208 - Nonformatted read error
209 - Nonformatted write error
210 - WriteProtected error
214 - IgnoreCloseErrors=yes
224 - FD sends unsaved file list to Director at end of job (see
226 - File daemon should build list of files skipped, and then
227 at end of save retry and report any errors.
228 - Write a Storage daemon that uses pipes and
229 standard Unix programs to write to the tape.
231 - Need something that monitors the JCR queue and
232 times out jobs by asking the deamons where they are.
233 - Enhance Jmsg code to permit buffering and saving to disk.
234 - device driver = "xxxx" for drives.
235 - restart: paranoid: read label fsf to
236 eom read append block, and go
237 super-paranoid: read label, read all files
238 in between, read append block, and go
239 verify: backspace, read append block, and go
240 permissive: same as above but frees drive
241 if tape is not valid.
243 - Ensure that /dev/null works
244 - Need report class for messages. Perhaps
245 report resource where report=group of messages
246 - enhance scan_attrib and rename scan_jobtype, and
247 fill in code for "since" option
248 - Director needs a time after which the report status is sent
249 anyway -- or better yet, a retry time for the job.
250 Don't reschedule a job if previous incarnation is still running.
251 - Some way to automatically backup everything is needed????
252 - Need a structure for pending actions:
254 - termination status (part of buffered msgs?)
255 - Concept of grouping Storage devices and job can use
256 any of a number of devices
258 Read, Write, Clean, Delete
259 - Login to Bacula; Bacula users with different permissions:
260 owner, group, user, quotas
261 - Store info on each file system type (probably in the job header on tape.
262 This could be the output of df; or perhaps some sort of /etc/mtab record.
265 - Design at hierarchial storage for Bacula.
266 - Implement FSM (File System Modules).
267 - Identify unchanged or "system" files and save them to a
268 special tape thus removing them from the standard
269 backup FileSet -- BASE backup.
270 - Turn virutally all sprintfs into snprintfs.
271 - Heartbeat between daemons.
272 - Audit M_ error codes to ensure they are correct and consistent.
273 - Add variable break characters to lex analyzer.
274 Either a bit mask or a string of chars so that
275 the caller can change the break characters.
276 - Make a single T_BREAK to replace T_COMMA, etc.
277 - Ensure that File daemon and Storage daemon can
278 continue a save if the Director goes down (this
279 is NOT currently the case). Must detect socket error,
280 buffer messages for later.
281 - Enhance time/duration input to allow multiple qualifiers e.g. 3d2h
282 - Add ability to backup to two Storage devices (two SD sessions) at
283 the same time -- e.g. onsite, offsite.
284 - Add the ability to consolidate old backup sets (basically do a restore
285 to tape and appropriately update the catalog). Compress Volume sets.
286 Might need to spool via file is only one drive is available.
287 - Compress or consolidate Volumes of old possibly deleted files. Perhaps
288 someway to do so with every volume that has less than x% valid
293 Bacula Projects Roadmap
295 last update 5 January 2003
297 Item 1: Multiple simultaneous Jobs. (done)
298 Done -- Restore part needs better implementation to work correctly
300 What: Permit multiple simultaneous jobs in Bacula.
302 Why: An enterprise level solution needs to go fast without the
303 need for the system administrator to carefully tweak
304 timing. Based on the benchmarks, during a full
305 backup, NetWorker typically hit 10 times the bandwidth to
306 the tape compared to Bacula--largely. This is probably due to
307 running parallel jobs and multi-threaded filling of buffers
308 and writing them to tape. This should also make things work
309 better when you have a mix of fast and slow machines backing
312 Notes: Bacula was designed to run multiple simultaneous jobs. Thus
313 implementing this is a matter of some small cleanups and
317 Item 2: Make the Storage daemon use intermediate file storage to buffer data.
318 Deferred -- not necessary yet.
320 What: If data is coming into the SD too fast, buffer it to
321 disk if the user has configured this option.
323 Why: This would be nice, especially if it more or less falls out
324 when implementing (1) above. If not, it probably should not
325 be given a high priority because fundamentally the backup time
326 is limited by the tape bandwidth. Even though you may finish a
327 client job quicker by spilling to disk, you still have to
328 eventually get it onto tape. If intermediate disk buffering
329 allows us to improve write bandwidth to tape, it may make
332 Notes: Whether or not this is implemented will depend upon performance
333 testing after item 1 is implemented.
336 Item 3: Write the bscan program -- also write a bcopy program.
339 What: Write a program that reads a Bacula tape and puts all the
340 appropriate data into the catalog. This allows recovery
341 from a tape that is no longer in the database, or it allows
342 re-creation of a database if lost.
344 Why: This is a fundamental robustness and disaster recovery tool
345 which will increase the comfort level of a sysadmin
346 considering adopting Bacula.
348 Notes: A skeleton of this program already exists, but much work
349 needs to be done. Implementing this will also make apparent
350 any deficiencies in the current Bacula tape format.
353 Item 4: Implement Base jobs.
355 What: A base job is sort of like a Full save except that you
356 will want the FileSet to contain only files that are unlikely
357 to change in the future (i.e. a snapshot of most of your
358 system after installing it). After the base job has been run,
359 when you are doing a Full save, you can specify to exclude
360 all files saved by the base job that have not been modified.
362 Why: This is something none of the competition does, as far as we know
363 (except BackupPC, which is a Perl program that saves to disk
364 only). It is big win for the user, it makes Bacula stand out
365 as offering a unique optimization that immediately saves time
368 Notes: Big savings in tape usage. Will require more resources because
369 the e. DIR must send FD a list of files/attribs, and the FD must
370 search the list and compare it for each file to be saved.
373 Item 5: Implement Label templates
375 What: This is a mechanism whereby Bacula can automatically create
376 a tape label for new tapes according to a detailed specification
377 provided by the user.
379 Why: It is a major convenience item for folks who use automated label
382 Notes: Bacula already has a working form of automatic tape label
383 creation, but it is very crude. The design for the complete
384 tape labeling project is already documented in the manual.
387 Item 6: Write a regression script.
390 What: This is an automatic script that runs and tests as many features
391 of Bacula as possible. The output is compared to previous
392 versions of Bacula and any differences are reported.
394 Why: This is an enormous help in preventing introduction of new
395 errors in parts of the program that already work correctly.
397 Notes: This probably should be ranked higher, it's something the typical
398 user doesn't see. Depending on how it's implemented, it may
399 make sense to defer it until the archival tape format and
400 user interface mature.
403 Item 7: GUI for interactive restore
404 Item 8: GUI for interactive backup
406 What: The current interactive restore is implemented with a tty
407 interface. It would be much nicer to be able to "see" the
408 list of files backed up in typical GUI tree format.
409 The same mechanism could also be used for creating
410 ad-hoc backup FileSets (item 8).
412 Why: Ease of use -- especially for the end user.
414 Notes: Rather than implementing in Gtk, we probably should go directly
415 for a Browser implementation, even if doing so meant the
416 capability wouldn't be available until much later. Not only
417 is there the question of Windows sites, most
418 Solaris/HP/IRIX, etc, shops can't currently run Gtk programs
419 without installing lots of stuff admins are very wary about.
420 Real sysadmins will always use the command line anyway, and
421 the user who's doing an interactive restore or backup of his
422 own files will in most cases be on a Windows machine running
426 Item 9: Add SSL to daemon communications.
428 What: This provides for secure communications between the daemons.
430 Why: This would allow doing backup across the Internet without
431 privacy concerns (or with much less concern).
433 Notes: The vast majority of near term potential users will be backing up
434 a single site over a LAN and, correctly or not, they probably
435 won't be concerned with security, at least not enough to go to
436 the trouble to set up keys, etc. to screw things down. We suspect
437 that many users genuinely interested in multi-site backup
438 already run some form of VPN software in their internetwork
439 connections, and are willing to delegate security to that layer.
442 Item 10: Define definitive tape format.
445 What: Define that definitive tape format that will not change
446 for the next millennium.
448 Why: Stability, security.
450 Notes: See notes for item 11 below.
453 Item 11: New daemon communication protocol.
455 What: The current daemon to daemon protocol is basically an ASCII
456 printf() and sending the buffer. On the receiving end, the
457 buffer is sscanf()ed to unpack it. The new scheme would
458 be a binary format that allows quick packing and unpacking
459 of any data type with named fields.
461 Why: Using binary packing would be faster. Named fields will permit
462 error checking to ensure that what is sent is what the
463 receiver really wants.
465 Notes: These are internal improvements in the interest of the
466 long-term stability and evolution of the program. On the one
467 hand, the sooner they're done, the less code we have to rip
468 up when the time comes to install them. On the other hand, they
469 don't bring an immediately perceptible benefit to potential
470 users. Item 10 and possibly item 11 should be deferred until Bacula
471 is well established with a growing user community more or
472 less happy with the feature set. At that time, it will make a
473 good "next generation" upgrade in the interest of data
479 ======================================================
481 It is somewhat like a Full save becomes an incremental since
482 the Base job (or jobs) plus other non-base files.
484 - New BaseFile table that contains:
485 JobId, BaseJobId, FileId (from Base).
486 i.e. for each base file that exists but is not saved because
487 it has not changed, the File daemon sends the JobId, BaseId,
488 and FileId back to the Director who creates the DB entry.
489 - To initiate a Base save, the Director sends the FD
490 the FileId, and full filename for each file in the Base.
491 - When the FD finds a Base file, he requests the Director to
492 send him the full File entry (stat packet plus MD5), or
493 conversely, the FD sends it to the Director and the Director
494 says yes or no. This can be quite rapid if the FileId is kept
495 by the FD for each Base Filename.
496 - It is probably better to have the comparison done by the FD
497 despite the fact that the File entry must be sent across the
499 - An alternative would be to send the FD the whole File entry
500 from the start. The disadvantage is that it requires a lot of
501 space. The advantage is that it requires less communications
503 - The Job record must be updated to indicate that one or more
505 - At end of Job, FD returns:
506 1. Count of base files/bytes not written to tape (i.e. matches)
507 2. Count of base file that were saved i.e. had changed.
508 - No tape record would be written for a Base file that matches, in the
509 same way that no tape record is written for Incremental jobs where
510 the file is not saved because it is unchanged.
511 - On a restore, all the Base file records must explicitly be
512 found from the BaseFile tape. I.e. for each Full save that is marked
513 to have one or more Base Jobs, search the BaseFile for all occurrences
515 - An optimization might be to make the BaseFile have:
521 This would avoid the need to explicitly fetch each File record for
522 the Base job. The Base Job record will be fetched to get the
523 VolSessionId and VolSessionTime.
524 =========================================================
526 ==========================================================
528 For each Incremental job that is run, there may be files that
529 were found but not saved because they were locked (this applies
530 only to Windows). Such a system could send back to the Director
531 a list of Unsaved files.
533 - New UnSavedFiles table that contains:
537 - Then in the next Incremental job, the list of Unsaved Files will be
538 feed to the FD, who will ensure that they are explicitly chosen even
539 if standard date/time check would not have selected them.
540 =============================================================
544 =============================================================
546 Request For Comments For File Backup Options
549 Subject: File Backup Options
552 A few days ago, a Bacula user who is backing up to file volumes and
553 using compression asked if it was possible to suppress compressing
554 all .gz files since it was a waste of CPU time. Although Bacula
555 currently permits using different options (compression, ...) on
556 a directory by directory basis, it cannot do it on a file by
557 file basis, which is clearly what was desired.
559 Proposed Implementation:
560 To solve this problem, I propose the following:
562 - Add a new Director resource type called FileOptions.
564 - The FileOptions resource will have records for all
565 options that can currently be specified on the Include record
566 (in a FileSet). Examples below.
568 - The FileOptions resource will permit an exclude option as well
569 as a number of additional options.
571 - The heart of the FileOptions resource is the ability to
572 supply any number of ApplyTo records which specify POSIX
573 regular expressions. These ApplyTo regular expressions are
574 applied to the fully qualified filename (path and all). If
575 one matches, then the FileOptions will be used.
577 - When an ApplyTo specification matches an included file, the
578 options specified in the FileOptions resource will override
579 the default options specified on the Include record.
581 - Include records will be modified to permit referencing one or
582 more FileOptions resources. The FileOptions will be used
583 in the order listed on the Include record and the first
584 one that matches will be applied.
586 - Options (or specifications) currently supplied on the Include
587 record will be deprecated (i.e. removed in a later version a
588 year or so from now).
590 - The Exclude record will be deprecated as the same functionality
591 can be obtained by using an Exclude = yes in the FileOptions.
594 The following records can appear in the FileOptions resource. An
595 asterisk preceding the name indicates a feature not currently
599 - Compression= (GZIP, ...)
600 - Signature= (MD5, SHA1, ...)
602 - OneFs= (yes/no) - remain on one filesystem
603 - Recurse= (yes/no) - recurse into subdirectories
604 - Sparse= (yes/no) - do sparse file backup
605 - *Exclude= (yes/no) - exclude file from being saved
606 - *Reader= (filename) - external read (backup) program
607 - *Plugin= (filename) - read/write plugin module
610 - verify= (ipnougsamc5) - verify options
613 - replace= (always/ifnewer/ifolder/never) - replace options currently
615 - *Writer= (filename) - external write (restore) program
619 Currently options specifying compression, MD5 signatures, recursion,
620 ... of a FileSet are supplied on the Include record. These will now
621 all be collected into a FileOptions resource, which will be
622 specified on the Include in place of the options. Multiple FileOptions
623 may be specified. Since the FileOptions contain regular expressions
624 that are applied to the full filename, this will give the ability
625 to specify backup options on a file by file basis to whatever level
634 Include = compression=GZIP signature=MD5 {
643 Include = FileOptions=Opts {
654 That's a lot more to do the same thing, but it gives the ability to
655 apply options on a file by file basis. For example, suppose you
656 want to compress all files but not any file with extensions .gz or .Z.
657 You could do so as follows:
661 Include = FileOptions=NoCompress FileOptions=Opts {
669 ApplyTo = /*.?*/ # matches all files
674 # Note multiple ApplyTos are ORed
675 ApplyTo = /*.gz/ # matches .gz files */
676 ApplyTo = /*.Z/ # matches .Z files */
679 Now, since the NoCompress FileOptions is specified first on the
680 Include line, any *.gz or *.Z file will have an MD5 signature computed,
681 but will not be compressed. For all other files, the NoCompress will not
682 match, so the Opts options will be used which will include GZIP
686 - Is it necessary to provide some means of ANDing regular expressions
687 and negation? (not currently planned)
689 e.g. ApplyTo = /*.gz/ && !/big.gz/
691 - I see that Networker has a "null" module which, if specified, does not
692 backup the file, but does make an record of the file in the catalog
693 so that the catalog will reflect an exact picture of the filesystem.
694 The result is that the file can be "seen" when "browsing" the save
695 sets, but it cannot be restored.
697 Is this really useful? Should it be implemented in Bacula?
700 After implementing the above, the user will be able to specify
701 on a file by file basis (using regular expressions) what options are
702 applied for the backup.
703 ====================================
705 =========================================
706 Proposal by Bill Sellers
708 Return-Path: <w.a.sellers@larc.nasa.gov>
709 Received: from post.larc.nasa.gov (post.larc.nasa.gov [128.155.4.45]) by matou.sibbald.com (8.11.6/8.11.6) with ESMTP id h0ELUIm07622 for <kern@sibbald.com>; Tue, 14 Jan 2003 22:30:18 +0100
710 Received: from Baron.larc.nasa.gov (baron.larc.nasa.gov [128.155.40.132]) by post.larc.nasa.gov (pohub4.6) with ESMTP id QAA09768 for <kern@sibbald.com>; Tue, 14 Jan 2003 16:30:14 -0500 (EST)
711 Message-Id: <5.1.0.14.2.20030114153452.028dbae8@pop.larc.nasa.gov>
712 X-Sender: w.a.sellers@pop.larc.nasa.gov
713 X-Mailer: QUALCOMM Windows Eudora Version 5.1
714 Date: Tue, 14 Jan 2003 16:30:18 -0500
715 To: Kern Sibbald <kern@sibbald.com>
716 From: Bill Sellers <w.a.sellers@larc.nasa.gov>
717 Subject: Re: [Bacula-users] Bacula remote storage?
718 In-Reply-To: <1042565382.1845.177.camel@rufus>
719 References: <5.1.0.14.2.20030114113004.0293a210@pop.larc.nasa.gov> <5.1.0.14.2.20030113170650.028dad88@pop.larc.nasa.gov> <5.1.0.14.2.20030113170650.028dad88@pop.larc.nasa.gov> <5.1.0.14.2.20030114113004.0293a210@pop.larc.nasa.gov>
721 Content-Type: text/plain; charset="us-ascii"; format=flowed
722 X-Annoyance-Filter-Junk-Probability: 0
723 X-Annoyance-Filter-Classification: Mail
724 At 06:29 PM 1/14/2003 +0100, you wrote:
727 >Well, if you cannot put a Bacula client on the machine,
728 >then it is a big problem. If you know of some software
729 >that can do what you want, let me know, because I
730 >really just don't know how to do it -- at least not
736 We have been able to get Amanda to use the HSM as a storage
737 device. Someone here wrote a driver for Amanda. BUT, Amanda doesn't
738 handle Windows systems very well (or at all without Samba). So I am
739 looking for a backup system that has a Windows client. I really like the
740 Windows integration of Bacula.
742 From the command line, its rather trivial to move the data around. We use
745 tar cf - ./files | gzip -c | rsh hsm dd of=path/file.tgz
747 or if you use GNU tar:
749 tar czf hsm:path/file.tgz ./files
751 One idea for you to consider; Sendmail offers pipes in the aliases file;
752 (mailpipe: "|/usr/bin/vacation root") and Perl supports pipes in the
753 "open" statement (open FILE, "|/bin/nroff -man";) Could you could make a
754 pipe available, as a storage device? Then we could use any command that
755 handles stdin as a storage destination.
762 #Password is not used in rsh, but might be used in ftp.
763 Device = "| gzip -c | rsh hsm dd of=path/file.tgz"
770 Password = "foobar&-"
771 Device = "| ncftpput -c hsm /path/file.bacula"
775 >If you have some local storage available, you could
776 >use Bacula to backup to disk volumes, then use some
777 >other software (ftp, scp) to move them to the HSM
778 >machine. However, this is a bit kludgy.
781 It is, but maybe worth a try. Is there some function in Bacula to put
782 variables in filenames? i.e. backup.2003-01-15.root
789 w.a.sellers@larc.nasa.gov
791 ==============================================
792 The Project for the above
794 I finally realized that this is not at all
795 the same as reader/writer programs or plugins,
796 which are alternate ways of accessing the
797 files to be backed up. Rather, it is an alternate
798 form of storage device, and I have always planned
799 that Bacula should be able to handle all sorts
802 So, I propose the following phases:
804 1. OK from you to invest some time in testing
805 this as I implement it (requires that you
806 know how to download from the SourceForge
807 cvs -- which I imagine is a piece of cake
810 2. Dumb implementation by allowing a device to
811 be a fifo for write only.
812 Reason: easy to implement, proof of concept.
814 3. Try reading from fifo but with fixed block
816 Reason: proof of concept, easy to implement.
818 4. Extend reading from fifo (restores) to handle
820 Reason: requires some delicate low level coding
821 which could destabilize all of Bacula.
823 5. Implementation of above but to a program. E.g.
824 Device = "|program" (not full pipeline).
825 Reason: routines already exist, and program can
826 be a shell script which contains anything.
828 6. Add full pipeline as a possibility. E.g.
829 Device = "| gzip -c | rsh hsm dd of=path/file.tgz"
830 Reason: needs additional coding to implement full
831 pipeline (must fire off either a shell or all
832 programs and connect their pipes).
834 There are a good number of details in each step
835 that I have left out, but I will specify them at
836 every stage, and there may be a few changes as things
837 evolve. I expect that to get to stage 5 will take a
838 few weeks, and at that point, you will have
839 everything you need (just inside a script).
840 Stage 6 will probably take longer, but if this
841 project pleases you, what we do for 5 should
842 be adequate for some time.
846 =============================================
848 Done: (see kernsdone for more)
849 - Look into Pruning/purging problems or why there seem to
850 be so many files listed each night.
851 - Fix cancel in find_one -- need jcr.
852 - Cancel does not work for restore in FD.
853 - Write SetJobStatus() function so cancel status not lost.
854 - Add include list to end of chain in findlib
855 - Zap sd_auth_key after use
856 - Add Bar code reading capabilities (new mtx-changer)
857 - Figure out some way to automatically backup all local partitions
858 - Make hash table for linked files in findlib/find_one.c:161
860 - Rewrite find_one.c to use only pool_memory instead of
861 alloca and malloc (probably not necessary).
862 - Make sure btraceback goes into /sbin not sysconf directory.
863 - InitVerify is getting pruned and it shouldn't (document it)
864 - Make 1.28c release ??? NO do 1.29 directly
865 - Set timeout on opening fifo for save or restore (findlib)
866 - Document FIFO storage device.
867 - Document fifo and | and <
870 - Get correct error status from run_program or open_bpipe().
871 - Restrict permissions on File Volumes (now 0640).
872 - Umasked 022 daemons
873 - Fix restore of hard linked file.
874 - Figure out how to allow multiple simultaneous file Volumes on a single device.
875 - Implement multiple simultaneous file Volumes on a single device.
876 - Cleanup db_update_media and db_update_pool
877 - Flush all the daemon messages at the end of every job.
878 - Change stat1= fgets()!=NULL to stat1=fgest()==NULL; in
879 run_program -- bpipe.c
880 - Apparently cancel does not work for jobs waiting to be
882 - Implement TCP/IP connection for MySQL
883 - Pull a canceled job from the Scheduling queue.
884 - Implement max_file_size in block.c (already done, just tweaked).