4 Documentation to do: (a little bit at a time)
5 - Document running a test version.
6 - Make sure restore options are documented
7 - Document query file format.
9 Testing to do: (painful)
10 - that restore options work in FD (mostly done).
11 - that console command line options work
12 - blocksize recognition code.
13 - Test new BSR code (mostly done).
14 - Test watchdog child timer code.
18 - Need a verbose mode in restore, perhaps to bsr.
19 - Should we dump a SOS when starting a new tape?
20 - bscan without -v is too quiet -- perhaps show jobs.
21 - Get rid of bscan.c:534 error message (one time only).
22 - Print some statistics when get EOF on device in bscan -- feedback
23 to let user know it is working.
24 - Add code to reject whole blocks if not wanted on restore.
26 - Figure out how to allow multiple simultaneous file Volumes on
28 - Start working on Base jobs.
29 - Implement FileOptions (see end of this document)
30 - Ensure that restore of differential jobs works (check SQL).
31 - Make sure the MaxVolFiles is fully implemented in SD
32 - Flush all the daemon messages at the end of every job.
33 - Check if both CatalogFiles and UseCatalog are set to SD.
34 - Check if we can increase Bacula FD priorty in Win2000
35 - Make bcopy read through bad tape records.
36 - Need return status on read_cb() from read_records(). Need multiple
37 records -- one per Job, maybe a JCR or some other structure with
40 - Look at hierarchial storage for Bacula.
41 - Implement Bacula plugins -- design API
43 - Work more on how to to a Bacula restore beginning with
44 just a Bacula tape and a boot floppy (bare metal recovery).
45 - Try bare metal Windows restore
46 - Fix read_record to handle multiple sessions.
47 - Program files (i.e. execute a program to read/write files).
48 Pass read date of last backup, size of file last time.
49 - Put system type returned by FD into catalog.
50 - Possibly add email to Watchdog if drive is unmounted too
51 long and a job is waiting on the drive.
52 - Strip trailing slashes from Include directory names in the FD.
53 - Use read_record.c in SD code.
54 - Why don't we get an error message from Win32 FD when bootstrap
55 file cannot be created for restore command?
56 - When Marking a file in Restore that is a hard link, also
57 mark the link so that the data will be reloaded.
58 - Restore program that errors in SD due to no tape reports
59 OK incorrectly in output.
60 - After unmount, if restore job started, ask to mount.
61 - Fix db_get_fileset in cats/sql_get.c for multiple records.
62 - Fix catalog filename truncation in sql_get and sql_create. Use
63 only a single filename split routine.
64 - Make Restore report an error if FD or SD term codes are not OK.
65 - Convert all %x substitution variables, which are hard to remember
66 and read to %(variable-name). Idea from TMDA.
67 - Add JobLevel in FD status (but make sure it is defined).
68 - Make Pool resource handle Counter resources.
69 - Remove NextId for SQLite. Optimize.
70 - Fix gethostbyname() to use gethostbyname_r()
71 - Implement ./configure --with-client-only
72 - Strip trailing / from Include
73 - Move all SQL statements into a single location.
74 - Cleanup db_update_media and db_update_pool
75 - Add UA rc and history files.
76 - put termcap (used by console) in ./configure and
77 allow -with-termcap-dir.
78 - Enhance time and size scanning routines.
79 - Fix Autoprune for Volumes to respect need for full save.
80 - DateWritten field on tape may be wrong.
81 - Fix Win32 config file definition name on /install
82 - No READLINE_SRC if found in alternate directory.
83 - Add Client FS/OS id (Linux, Win95/98, ...).
84 - Test a second language e.g. french.
85 - Compare tape to Client files (attributes, or attributes and data)
86 - Restore options (overwrite, overwrite if older,
87 overwrite if newer, never overwrite, ...)
88 - Restore to a particular time -- e.g. before date, after date.
89 - Make all database Ids 64 bit.
90 - Write an applet for Linux.
91 - Add estimate to Console commands
92 - Find solution to blank filename (i.e. path only) problem.
93 - Implement new daemon communications protocol.
94 - Remove PoolId from Job table, it exists in Media.
95 - Allow console commands to detach or run in background.
96 - Fix status delay on storage daemon during rewind.
97 - Add SD message variables to control operator wait time
98 - Maximum Operator Wait
99 - Minimum Message Interval
100 - Maximum Message Interval
101 - Send Operator message when cannot read tape label.
102 - Verify level=Volume (scan only), level=Data (compare of data to file).
103 Verify level=Catalog, level=InitCatalog
105 - Add keyword search to show command in Console.
106 - Fix Win2000 error with no messages during startup.
107 - Events : tape has more than xxx bytes.
108 - Restrict characters permitted in a Resource name.
109 - Complete code in Bacula Resources -- this will permit
110 reading a new config file at any time.
111 - Handle ctl-c in Console
112 - Implement LabelTemplate (at least first cut).
113 - Implement script driven addition of File daemon to config files.
114 - Think about how to make Bacula work better with File archives.
116 - see setgroup and user for Bacula p4-5 of stunnel.c
117 - Implement new serialize subroutines
118 send(socket, "string", &Vol, "uint32", &i, NULL)
119 - Audit all UA commands to ensure that we always prompt where possible.
120 - If ./btape is called without /dev, assume argument is a Storage resource name.
121 - Put memory utilization in Status output of each daemon
122 if full status requested or if some level of debug on.
123 - Make database type selectable by .conf files i.e. at runtime
124 - gethostbyname failure in bnet_connect() continues
125 generating errors -- should stop.
126 - Add HOST to Volume label.
127 - Set flag for uname -a. Add to Volume label.
128 - Implement throttled work queue.
129 - Check for EOT at ENOSPC or EIO or ENXIO (unix Pc)
130 - Allow multiple Storage specifications (or multiple names on
131 a single Storage specification) in the Job record. Thus a job
132 can be backed up to a number of storage devices.
133 - Implement dump label to UA
134 - Copy volume using single drive.
135 - Concept of VolumeSet during restore which is a list
136 of Volume names needed.
137 - Restore files modified after date
138 - Restore file modified before date
139 - Emergency restore info:
141 - Backup working directory
143 - Restore -- do nothing but show what would happen
144 - SET LD_RUN_PATH=$HOME/mysql/lib/mysql
145 - Implement Restore FileSet=
146 - Create a protocol.h and protocol.c where all protocol messages
148 - If SD cannot open a drive, make it periodically retry.
149 - Remove duplicate fields from jcr (e.g. jcr.level and jcr.jr.Level, ...).
150 - Timout a job or terminate if link goes down, or reopen link and query.
151 - Fill all fields in Vol/Job Header -- ensure that everything
152 needed is written to tape. Think about restore to Catalog
153 from tape. Client record needs improving.
154 - Find general solution for sscanf size problems (as well
155 as sprintf. Do at run time?
156 - Concept of precious tapes (cannot be reused).
157 - Make bcopy copy with a single tape drive.
158 - Permit changing ownership during restore.
160 - Restore should get Device and Pool information from
161 job record rather than from config.
162 - Autolabel should be specified by DR instead of SD.
163 - Find out how to get the system tape block limits, e.g.:
164 Apr 22 21:22:10 polymatou kernel: st1: Block limits 1 - 245760 bytes.
165 Apr 22 21:22:10 polymatou kernel: st0: Block limits 2 - 16777214 bytes.
168 - AutoScan (check checksum of tape)
169 - Format command = "format /dev/nst0"
173 - Seek resolution (usually corresponds to buffer size)
174 - EODErrorCode=ENOSPC or code
175 - Partial Read error code
176 - Partial write error code
177 - Nonformatted read error
178 - Nonformatted write error
179 - WriteProtected error
183 - IgnoreCloseErrors=yes
193 - FD sends unsaved file list to Director at end of job.
194 - Write a Storage daemon that uses pipes and
195 standard Unix programs to write to the tape.
197 - Need something that monitors the JCR queue and
198 times out jobs by asking the deamons where they are.
200 - Enhance Jmsg code to permit buffering and saving to disk.
201 - device driver = "xxxx" for drives.
202 - restart: paranoid: read label fsf to
203 eom read append block, and go
204 super-paranoid: read label, read all files
205 in between, read append block, and go
206 verify: backspace, read append block, and go
207 permissive: same as above but frees drive
208 if tape is not valid.
210 - Ensure that /dev/null works
211 - File daemon should build list of files skipped, and then
212 at end of save retry and report any errors.
213 - Need report class for messages. Perhaps
214 report resource where report=group of messages
215 - enhance scan_attrib and rename scan_jobtype, and
216 fill in code for "since" option
217 - Need to save contents of FileSet to tape?
218 - Director needs a time after which the report status is sent
219 anyway -- or better yet, a retry time for the job.
220 Don't reschedule a job if previous incarnation is still running.
221 - Figure out how to save the catalog (possibly a special FileSet).
222 - Figure out how to restore the catalog.
223 - Some way to automatically backup everything is needed????
224 - Need a structure for pending actions:
226 - termination status (part of buffered msgs?)
227 - Concept of grouping Storage devices and job can use
228 any of a number of devices
230 Read, Write, Clean, Delete
231 - Login to Bacula; Bacula users with different permissions:
232 owner, group, user, quotas
233 - Store info on each file system type (probably in the job header on tape.
234 This could be the output of df; or perhaps some sort of /etc/mtab record.
237 - Implement FSM (File System Modules).
238 - Identify unchanged or "system" files and save them to a
239 special tape thus removing them from the standard
240 backup FileSet -- BASE backup.
241 - Turn virutally all sprintfs into snprintfs.
242 - Heartbeat between daemons.
243 - Audit M_ error codes to ensure they are correct and consistent.
244 - Add variable break characters to lex analyzer.
245 Either a bit mask or a string of chars so that
246 the caller can change the break characters.
247 - Make a single T_BREAK to replace T_COMMA, etc.
248 - Ensure that File daemon and Storage daemon can
249 continue a save if the Director goes down (this
250 is NOT currently the case). Must detect socket error,
251 buffer messages for later.
252 - Enhance time/duration input to allow multiple qualifiers e.g. 3d2h
256 Bacula Projects Roadmap
258 last update 27 November 2002
260 Item 1: Multiple simultaneous Jobs. (done)
263 What: Permit multiple simultaneous jobs in Bacula.
265 Why: An enterprise level solution needs to go fast without the
266 need for the system administrator to carefully tweak
267 timing. Based on the benchmarks, during a full
268 backup, NetWorker typically hit 10 times the bandwidth to
269 the tape compared to Bacula--largely. This is probably due to
270 running parallel jobs and multi-threaded filling of buffers
271 and writing them to tape. This should also make things work
272 better when you have a mix of fast and slow machines backing
275 Notes: Bacula was designed to run multiple simultaneous jobs. Thus
276 implementing this is a matter of some small cleanups and
280 Item 2: Make the Storage daemon use intermediate file storage to buffer data.
281 Deferred -- not necessary yet.
283 What: If data is coming into the SD too fast, buffer it to
284 disk if the user has configured this option.
286 Why: This would be nice, especially if it more or less falls out
287 when implementing (1) above. If not, it probably should not
288 be given a high priority because fundamentally the backup time
289 is limited by the tape bandwidth. Even though you may finish a
290 client job quicker by spilling to disk, you still have to
291 eventually get it onto tape. If intermediate disk buffering
292 allows us to improve write bandwidth to tape, it may make
295 Notes: Whether or not this is implemented will depend upon performance
296 testing after item 1 is implemented.
299 Item 3: Write the bscan program -- also write a bcopy program.
302 What: Write a program that reads a Bacula tape and puts all the
303 appropriate data into the catalog. This allows recovery
304 from a tape that is no longer in the database, or it allows
305 re-creation of a database if lost.
307 Why: This is a fundamental robustness and disaster recovery tool
308 which will increase the comfort level of a sysadmin
309 considering adopting Bacula.
311 Notes: A skeleton of this program already exists, but much work
312 needs to be done. Implementing this will also make apparent
313 any deficiencies in the current Bacula tape format.
316 Item 4: Implement Base jobs.
318 What: A base job is sort of like a Full save except that you
319 will want the FileSet to contain only files that are unlikely
320 to change in the future (i.e. a snapshot of most of your
321 system after installing it). After the base job has been run,
322 when you are doing a Full save, you can specify to exclude
323 all files saved by the base job that have not been modified.
325 Why: This is something none of the competition does, as far as we know
326 (except BackupPC, which is a Perl program that saves to disk
327 only). It is big win for the user, it makes Bacula stand out
328 as offering a unique optimization that immediately saves time
331 Notes: Big savings in tape usage. Will require more resources because
332 the e. DIR must send FD a list of files/attribs, and the FD must
333 search the list and compare it for each file to be saved.
336 Item 5: Implement Label templates
338 What: This is a mechanism whereby Bacula can automatically create
339 a tape label for new tapes according to a detailed specification
340 provided by the user.
342 Why: It is a major convenience item for folks who use automated label
345 Notes: Bacula already has a working form of automatic tape label
346 creation, but it is very crude. The design for the complete
347 tape labeling project is already documented in the manual.
350 Item 6: Write a regression script.
353 What: This is an automatic script that runs and tests as many features
354 of Bacula as possible. The output is compared to previous
355 versions of Bacula and any differences are reported.
357 Why: This is an enormous help in preventing introduction of new
358 errors in parts of the program that already work correctly.
360 Notes: This probably should be ranked higher, it's something the typical
361 user doesn't see. Depending on how it's implemented, it may
362 make sense to defer it until the archival tape format and
363 user interface mature.
366 Item 7: GUI for interactive restore
367 Item 8: GUI for interactive backup
369 What: The current interactive restore is implemented with a tty
370 interface. It would be much nicer to be able to "see" the
371 list of files backed up in typical GUI tree format.
372 The same mechanism could also be used for creating
373 ad-hoc backup FileSets (item 8).
375 Why: Ease of use -- especially for the end user.
377 Notes: Rather than implementing in Gtk, we probably should go directly
378 for a Browser implementation, even if doing so meant the
379 capability wouldn't be available until much later. Not only
380 is there the question of Windows sites, most
381 Solaris/HP/IRIX, etc, shops can't currently run Gtk programs
382 without installing lots of stuff admins are very wary about.
383 Real sysadmins will always use the command line anyway, and
384 the user who's doing an interactive restore or backup of his
385 own files will in most cases be on a Windows machine running
389 Item 9: Add SSL to daemon communications.
391 What: This provides for secure communications between the daemons.
393 Why: This would allow doing backup across the Internet without
394 privacy concerns (or with much less concern).
396 Notes: The vast majority of near term potential users will be backing up
397 a single site over a LAN and, correctly or not, they probably
398 won't be concerned with security, at least not enough to go to
399 the trouble to set up keys, etc. to screw things down. We suspect
400 that many users genuinely interested in multi-site backup
401 already run some form of VPN software in their internetwork
402 connections, and are willing to delegate security to that layer.
405 Item 10: Define definitive tape format.
408 What: Define that definitive tape format that will not change
409 for the next millennium.
411 Why: Stability, security.
413 Notes: See notes for item 11 below.
416 Item 11: New daemon communication protocol.
418 What: The current daemon to daemon protocol is basically an ASCII
419 printf() and sending the buffer. On the receiving end, the
420 buffer is sscanf()ed to unpack it. The new scheme would
421 be a binary format that allows quick packing and unpacking
422 of any data type with named fields.
424 Why: Using binary packing would be faster. Named fields will permit
425 error checking to ensure that what is sent is what the
426 receiver really wants.
428 Notes: These are internal improvements in the interest of the
429 long-term stability and evolution of the program. On the one
430 hand, the sooner they're done, the less code we have to rip
431 up when the time comes to install them. On the other hand, they
432 don't bring an immediately perceptible benefit to potential
433 users. Item 10 and possibly item 11 should be deferred until Bacula
434 is well established with a growing user community more or
435 less happy with the feature set. At that time, it will make a
436 good "next generation" upgrade in the interest of data
442 ====================================
447 Subject: File Backup Options
450 A few days ago, a Bacula user who is backing up to file volumes and
451 using compression asked if it was possible to suppress compressing
452 all .gz files since it was a waste of CPU time. Although Bacula
453 currently permits using different options (compression, ...) on
454 a directory by directory basis, it cannot do it on a file by
455 file basis, which is clearly what was desired.
457 Proposed Implementation:
458 To solve this problem, I propose the following:
460 - Add a new Director resource type called FileOptions.
462 - The FileOptions resource will have records for all
463 options that can currently be specified on the Include record
464 (in a FileSet). Examples below.
466 - The FileOptions resource will permit an exclude option as well
467 as a number of additional options.
469 - The heart of the FileOptions resource is the ability to
470 supply any number of ApplyTo records which specify POSIX
471 regular expressions. These ApplyTo regular expressions are
472 applied to the fully qualified filename (path and all). If
473 one matches, then the FileOptions will be used.
475 - When an ApplyTo specification matches an included file, the
476 options specified in the FileOptions resource will override
477 the default options specified on the Include record.
479 - Include records will be modified to permit referencing one or
480 more FileOptions resources. The FileOptions will be used
481 in the order listed on the Include record and the first
482 one that matches will be applied.
484 - Options (or specifications) currently supplied on the Include
485 record will be deprecated (i.e. removed in a later version a
486 year or so from now).
488 - The Exclude record will be deprecated as the same functionality
489 can be obtained by using an Exclude = yes in the FileOptions.
492 The following records can appear in the FileOptions resource. An
493 asterisk preceding the name indicates a feature not currently
497 - Compression= (GZIP, ...)
498 - Signature= (MD5, SHA1, ...)
500 - OneFs= (yes/no) - remain on one filesystem
501 - Recurse= (yes/no) - recurse into subdirectories
502 - Sparse= (yes/no) - do sparse file backup
503 - *Exclude= (yes/no) - exclude file from being saved
504 - *Reader= (filename) - external read (backup) program
507 - verify= (ipnougsamc5) - verify options
510 - replace= (always/ifnewer/ifolder/never) - replace options currently
512 - *Writer= (filename) - external write (restore) program
516 Currently options specifying compression, MD5 signatures, recursion,
517 ... of a FileSet are supplied on the Include record. These will now
518 all be collected into a FileOptions resource, which will be
519 specified on the Include in place of the options. Multiple FileOptions
520 may be specified. Since the FileOptions contain regular expressions
521 that are applied to the full filename, this will give the ability
522 to specify backup options on a file by file basis to whatever level
531 Include = compression=GZIP signature=MD5 {
540 Include = FileOptions=Opts {
551 That's a lot more to do the same thing, but it gives the ability to
552 apply options on a file by file basis. For example, suppose you
553 want to compress all files but not any file with extensions .gz or .Z.
554 You could do so as follows:
558 Include = FileOptions=NoCompress FileOptions=Opts {
566 ApplyTo = /*.?*/ # matches all files
571 # Note multiple ApplyTos are ORed
572 ApplyTo = /*.gz/ # matches .gz files */
573 ApplyTo = /*.Z/ # matches .Z files */
576 Now, since the NoCompress FileOptions is specified first on the
577 Include line, any *.gz or *.Z file will have an MD5 signature computed,
578 but will not be compressed. For all other files, the NoCompress will not
579 match, so the Opts options will be used which will include GZIP
583 - Is it necessary to provide some means of ANDing regular expressions
584 and negation? (not currently planned)
586 e.g. ApplyTo = /*.gz/ && !/big.gz/
588 - I see that Networker has a "null" module which, if specified, does not
589 backup the file, but does make an record of the file in the catalog
590 so that the catalog will reflect an exact picture of the filesystem.
591 The result is that the file can be "seen" when "browsing" the save
592 sets, but it cannot be restored.
594 Is this really useful? Should it be implemented in Bacula?
597 After implementing the above, the user will be able to specify
598 on a file by file basis (using regular expressions) what options are
599 applied for the backup.
600 ====================================
602 Done: (see kernsdone for more)
603 - Add EOM records? No, not at this time. The current system works and
605 - Add VolumeUseDuration and MaximumVolumeJobs to Pool db record and
607 - Add VOLUME_CAT_INFO to the EOS tape record (as
608 well as to the EOD record). -- No, not at this time.
609 - Put MaximumVolumeSize in Director (MaximumVolumeJobs, MaximumVolumeFiles,
611 - Enhance schedule to have 1stSat, ...
612 - Make sure catalog doesn't keep growing.
613 - On I/O error, write EOF, then try to write again ? No, keep it simple.
614 - Figure out how compress everything except .gz,... files.
615 Implement FileOptions.
616 - Put Bacula version somewhere in Job stream, probably Start Session Labels.
617 - Fix start/end blocks for File devices
618 - Make Job err if WriteBootstrap fails.
619 - Test that mod of restore options works.
620 - Test that week position schedule code works.
621 - Make BSR accept count (total files to be restored).
622 - Add code to fast seek to proper place on tape/file when doing Restore.
623 - Replace popen() and pclose() -- fail safe and timeout, no SIG dep.
624 - Add code to put VolFile in bsr for restore command.
625 - Volumes can be listed multiple times in Restore volume list.
626 - Add watchdog timeout for child processes start_child_timer()