4 Documentation to do: (any release a little bit at a time)
5 - Document running a test version.
6 - Document query file format.
7 - Document static linking
8 - Document problems with Verify and pruning.
9 - Document how to use multiple databases.
10 - For FreeBSD typical /dev/nrsa0 and for mtx /dev/pass1
11 - VXA drives have a "cleaning required"
12 indicator, but Exabyte recommends preventive cleaning after every 75
14 - Lookup HP cleaning recommendations.
15 - Lookup HP tape replacement recommendations (see trouble shooting autochanger)
16 - Document FInclude ...
17 - Document need to add "-u root" to most of MySQL script calls
18 (./create_mys... ./make_my...).
19 - Document all the status codes JobLevel, JobType, JobStatus.
20 - Document dynamic DNS
21 - Document using multiple Pools for daily rotating tapes.
22 - Fix default time in seconds without qualifier -- bacula-dir.conf
23 - From Marc Brueckner der Megauser
24 Backup OS Restore OS Result
25 -------------------------------------------------------------------------------
27 WinME WinNT Works (minor problems with rights)
28 WinME WINXP Works (minor problems with rights)
29 WinME Linux Works (minor problems with rights)
34 WinXP WinME Error: Win32 data stream not supported on this Client
35 WinXP Linux Error: Win32 data stream not supported on this Client
39 WinNT WinME Error: Win32 data stream not supported on this Client
40 WinNT Linux Error: Win32 data stream not supported on this Client
43 Linux WinNT Works (minor problems with rights)
44 Linux WinME Client hangs
45 * When restoring files on WinNT previously backed up on XP I get the following
46 Message for some files.
47 But any file was restored correctly !
49 Write error on c:/tmp/transfer/WinSCP2.exe: ERR=Die Daten sind unzulassig. (the
51 - Add Nic Bellamy's backup scheme.
53 Testing to do: (painful)
54 - that ALL console command line options work and are always implemented
55 - blocksize recognition code.
56 - Test if rewind at end of tape waits for tape to rewind.
58 - Test not zeroing Autochanger slot when it is wrong.
59 - Test of last block is correct in JobMedia when splitting file
61 - Test recycling and purging (code changed in db_find_next_volume and
63 - Figure out how to use ssh or stunnel to protect Bacula communications.
66 - Backup on havana gave:
68 The defined Storage resources are:
71 Select Storage resource (1-2): 2
72 Connecting to Storage daemon DDS-3 at havana:9103
74 havana-sd Version: 1.31 (08 Jul 2003)
75 Daemon started 10-Jul-2003 19:06, 0 Jobs run.
76 Device /tmp is not open.
77 Device /dev/nst0 is mounted with Volume "msu-8Jul03"
78 Device is being initialized.
79 Total Bytes Read=4,629,249,084,728,272,628 Blocks Read=0 Bytes/block=134,735,604
80 Positioned at File=0 Block=0
81 Full Backup job msu-8Jul03 using Volume "msu-backup.2003-07-10_19.06.42" on device /dev/nst0
82 Files=0 Bytes=0 Bytes/sec=0
83 FDReadSeqNo=6 in_msg=6 out_msg=6 fd=7
86 - Do full check the command line args in update (e.g. VolStatus ...).
87 - Maybe remove multiple simultaneous devices code in SD.
88 - Increment DB version prior to releasing.
89 - Turn off FULL_DEBUG prior to releasing.
90 - On Windows with very long path names, it may be impossible to create
91 a file (and thus restore it) because the total length is too long.
92 We must cd into the directory then create the file without the
94 - Move JobFiles and JobBytes to SD rather than FD -- more correct.
95 - lstat() is not going to work on Win32 for testing date.
96 - Implement a Recycle command
97 - Something is not right in last block of fill command.
98 - Implement List Volume Job=xxx or List scheduled volumes or Status Director
99 - Check if Incremental is working correctly when it looks for the previous Job
101 - Add FileSet to command line arguments for restore.
105 - Implement a Recycle or Prune Current Volume (as from Nic Bellamy),
106 possibly implement Prune Oldest Volume.
107 - Allow multiple Storage specifications (or multiple names on
108 a single Storage specification) in the Job record. Thus a job
109 can be backed up to a number of storage devices.
110 - Add client name to cram-md5 challenge so Director can immediately
111 verify if it is the correct client.
112 - Implement ClientRunBeforeJob and ClientRunAfterJob.
113 - Implement new alist in FileSet scanning.
114 - Add JobLevel in FD status (but make sure it is defined).
115 - Audit all UA commands to ensure that we always prompt where possible.
116 - Restrict characters permitted in a Resource name, and don't permit
118 - Prohibit backing up archive device (findlib/find_one.c:128)
119 - Make | and < work on FD side.
120 - Check Jmsg in bnet, may not work, must dup bsock.
121 - Suppress Job Name in Jmsg for console
122 - Create Pools that are referenced in a Run statement at startup if possible.
123 - Use runbeforejob to unload, then reload a volume previously used,
124 then the next job run gets an error reading the drive.
125 - Make bootstrap filename unique.
126 - Test a second language e.g. french.
127 - Start working on Base jobs.
128 - Make "make binary-release" work from any directory.
129 - Unsaved Flag in Job record (use JobMissingFiles).
130 - Base Flag in Job record.
131 - Implement UnsavedFiles DB record.
132 - Implement argc/argv for daemon command line scanning using table driven
134 - Implement table driven single argc/argv scanner to pickup all arguments.
135 Much like xxx_conf.c scan table.
136 keyword, handler(store_routine), store_address, code, flags, default.
137 - From Phil Stracchino:
138 It would probably be a per-client option, and would be called
139 something like, say, "Automatically purge obsoleted jobs". What it
140 would do is, when you successfully complete a Differential backup of a
141 client, it would automatically purge all Incremental backups for that
142 client that are rendered redundant by that Differential. Likewise,
143 when a Full backup on a client completed, it would automatically purge
144 all Differential and Incremental jobs obsoleted by that Full backup.
145 This would let people minimize the number of tapes they're keeping on
146 hand without having to master the art of retention times.
147 - Implement new serialize subroutines
148 send(socket, "string", &Vol, "uint32", &i, NULL)
149 - Scratch Pool where the volumes can be re-assigned to any Pool.
150 - Implement a M_SECURITY message class.
151 - Implement forward spacing block/file: position_device(bsr) --
152 just before read_block_from_device();
153 - When doing a Backup send all attributes back to the Director, who
154 would then figure out what files have been deleted.
155 - Currently in mount.c:236 the SD simply creates a Volume. It should have
156 explicit permission to do so. It should also mark the tape in error
157 if there is an error.
158 - Make sure all restore counters are working correctly in the FD.
159 - SD Bytes Read is wrong.
160 - Configure mtx-changer to have correct path to mtx.
161 - Look at ALL higher level routines that call block.c to be sure
162 they don't expect something in errmsg.
163 - Investigate doing RAW backup of Win32 partition.
164 - Add JobName= to VerifyToCatalog so that all verifies can be done at the end.
165 - Add thread specific data to hold the jcr -- send error messages from
166 low level routines by accessing it and using Jmsg().
167 - Cancel waiting for Client connect in SD if FD goes away.
169 - Add Progress command that periodically reports the progress of
171 - One block was orphaned in the SD probably after cancel.
172 - Add all command line arguments to "update", e.g. slot=nn volStatus=append, ...
175 - Examine Bare Metal restore problem (a FD crash exists somewhere ...).
176 - Implement timeout in response() when it should come quickly.
177 - Implement console @echo command.
178 - Implement a Slot priority (loaded/not loaded).
179 - Implement "vacation" Incremental only saves.
180 - Implement single pane restore (much like the Gftp panes).
181 - Implement Automatic Mount even in operator wait.
182 - Implement create "FileSet"?
183 - Implement Release Device in the Job resource to unmount a drive.
184 - Implement Acquire Device in the Job resource to mount a drive,
185 be sure this works with admin jobs so that the user can get
186 prompted to insert the correct tape. Possibly some way to say to
187 run the job but don't save the files.
188 - Implement all command line args on run.
189 - Implement command line "restore" args.
190 - Implement "restore current select=no"
191 - Fix watchdog pthread crash on Win32 (this is pthread_kill() Cygwin bug)
192 - Implement "scratch pool" where tapes are defined and can be
193 taken by any pool that needs them.
194 - Implement restore "current system", but take all files without
195 doing selection tree -- so that jobs without File records can
197 - Implement disk spooling. Two parts: 1. Spool to disk then
198 immediately to tape to speed up tape operations. 2. Spool to
199 disk only when the tape is full, then when a tape is hung move
201 - Implement a relocatable bacula.spec
202 - Implement dump/print label to UA
203 - Add prefixlinks to where or not where absolute links to FD.
204 - Issue message to mount a new tape before the rewind.
205 - Simplified client job initiation for portables.
206 - If SD cannot open a drive, make it periodically retry.
207 - Implement LabelTemplate (at least first cut).
208 - Add more of the config info to the tape label.
209 - Implement a Mount Command and an Unmount Command where
210 the users could specify a system command to be performed
211 to do the mount, after which Bacula could attempt to
212 read the device. This is for Removeable media such as a CDROM.
213 - Most likely, this mount command would be invoked explicitly
214 by the user using the current Console "mount" and "unmount"
215 commands -- the Storage Daemon would do the right thing
216 depending on the exact nature of the device.
217 - As with tape drives, when Bacula wanted a new removable
218 disk mounted, it would unmount the old one, and send a message
219 to the user, who would then use "mount" as described above
220 once he had actually inserted the disk.
222 - Make some way so that if a machine is skipped because it is not up
223 that Bacula will continue retrying for a specified period of time --
225 - If tape is marked read-only, then try opening it read-only rather than
226 failing, and remember that it cannot be written.
227 - Refine SD waiting output:
228 Device is being positioned
229 > Device is being positioned for append
230 > Device is being positioned to file x
232 - Figure out some way to estimate output size and to avoid splitting
233 a backup across two Volumes -- this could be useful for writing CDROMs
234 where you really prefer not to have it split -- not serious.
235 - Add RunBeforeJob and RunAfterJob to the Client program.
236 - Have SD compute MD5 or SHA1 and compare to what FD computes.
237 - Make VolumeToCatalog calculate an MD5 or SHA1 from the
238 actual data on the Volume and compare it.
239 - Implement FileOptions (see end of this document)
240 - Implement Bacula plugins -- design API
241 - Make bcopy read through bad tape records.
242 - Fix read_record to handle multiple sessions.
243 - Program files (i.e. execute a program to read/write files).
244 Pass read date of last backup, size of file last time.
245 - Add Signature type to File DB record.
246 - CD into subdirectory when open()ing files for backup to
247 speed up things. Test with testfind().
248 - Priority job to go to top of list.
249 - Why are save/restore of device different sizes (sparse?) Yup! Fix it.
250 - Implement some way for the Console to dynamically create a job.
251 - Restore to a particular time -- e.g. before date, after date.
252 - Solaris -I on tar for include list
253 - Need a verbose mode in restore, perhaps to bsr.
254 - bscan without -v is too quiet -- perhaps show jobs.
255 - Add code to reject whole blocks if not wanted on restore.
256 - Check if we can increase Bacula FD priorty in Win2000
257 - Make sure the MaxVolFiles is fully implemented in SD
258 - Check if both CatalogFiles and UseCatalog are set to SD.
259 - Need return status on read_cb() from read_records(). Need multiple
260 records -- one per Job, maybe a JCR or some other structure with
261 a block and a record.
262 - Figure out how to do a bare metal Windows restore
263 - Possibly add email to Watchdog if drive is unmounted too
264 long and a job is waiting on the drive.
265 - Use read_record.c in SD code.
266 - Restore program that errors in SD due to no tape reports
267 OK incorrectly in output.
268 - After unmount, if restore job started, ask to mount.
269 - Convert all %x substitution variables, which are hard to remember
270 and read to %(variable-name). Idea from TMDA.
271 - Remove NextId for SQLite. Optimize.
272 - Move all SQL statements into a single location.
273 - Add UA rc and history files.
274 - put termcap (used by console) in ./configure and
275 allow -with-termcap-dir.
276 - Enhance time and size scanning routines.
277 - Fix Autoprune for Volumes to respect need for full save.
278 - Fix Win32 config file definition name on /install
279 - Compare tape to Client files (attributes, or attributes and data)
280 - Make all database Ids 64 bit.
281 - Write an applet for Linux.
282 - Add estimate to Console commands
283 - Implement new daemon communications protocol.
284 - Allow console commands to detach or run in background.
285 - Fix status delay on storage daemon during rewind.
286 - Add SD message variables to control operator wait time
287 - Maximum Operator Wait
288 - Minimum Message Interval
289 - Maximum Message Interval
290 - Send Operator message when cannot read tape label.
291 - Verify level=Volume (scan only), level=Data (compare of data to file).
292 Verify level=Catalog, level=InitCatalog
294 - Add keyword search to show command in Console.
295 - Events : tape has more than xxx bytes.
296 - Complete code in Bacula Resources -- this will permit
297 reading a new config file at any time.
298 - Handle ctl-c in Console
299 - Implement script driven addition of File daemon to config files.
300 - Think about how to make Bacula work better with File (non-tape) archives.
301 - Write Unix emulator for Windows.
302 - Put memory utilization in Status output of each daemon
303 if full status requested or if some level of debug on.
304 - Make database type selectable by .conf files i.e. at runtime
305 - Set flag for uname -a. Add to Volume label.
306 - Implement throttled work queue.
307 - Restore files modified after date
308 - Restore file modified before date
309 - Restore -- do nothing but show what would happen
310 - SET LD_RUN_PATH=$HOME/mysql/lib/mysql
311 - Implement Restore FileSet=
312 - Create a protocol.h and protocol.c where all protocol messages
314 - Remove duplicate fields from jcr (e.g. jcr.level and jcr.jr.Level, ...).
315 - Timout a job or terminate if link goes down, or reopen link and query.
316 - Concept of precious tapes (cannot be reused).
317 - Make bcopy copy with a single tape drive.
318 - Permit changing ownership during restore.
321 > My suggestion: Add a feature on the systray menu-icon menu to request
322 > an immediate backup now. This would be useful for laptop users who may
323 > not be on the network when the regular scheduled backup is run.
325 > My wife's suggestion: Add a setting to the win32 client to allow it to
326 > shut down the machine after backup is complete (after, of course,
327 > displaying a "System will shut down in one minute, click here to cancel"
328 > warning dialog). This would be useful for sites that want user
329 > woorkstations to be shut down overnight to save power.
332 - Autolabel should be specified by DIR instead of SD.
335 - AutoScan (check checksum of tape)
336 - Format command = "format /dev/nst0"
340 - Seek resolution (usually corresponds to buffer size)
341 - EODErrorCode=ENOSPC or code
342 - Partial Read error code
343 - Partial write error code
344 - Nonformatted read error
345 - Nonformatted write error
346 - WriteProtected error
350 - IgnoreCloseErrors=yes
360 - FD sends unsaved file list to Director at end of job (see
362 - File daemon should build list of files skipped, and then
363 at end of save retry and report any errors.
364 - Write a Storage daemon that uses pipes and
365 standard Unix programs to write to the tape.
367 - Need something that monitors the JCR queue and
368 times out jobs by asking the deamons where they are.
369 - Enhance Jmsg code to permit buffering and saving to disk.
370 - device driver = "xxxx" for drives.
372 - Ensure that /dev/null works
373 - Need report class for messages. Perhaps
374 report resource where report=group of messages
375 - enhance scan_attrib and rename scan_jobtype, and
376 fill in code for "since" option
377 - Director needs a time after which the report status is sent
378 anyway -- or better yet, a retry time for the job.
379 - Don't reschedule a job if previous incarnation is still running.
380 - Some way to automatically backup everything is needed????
381 - Need a structure for pending actions:
383 - termination status (part of buffered msgs?)
385 Read, Write, Clean, Delete
386 - Login to Bacula; Bacula users with different permissions:
387 owner, group, user, quotas
388 - Store info on each file system type (probably in the job header on tape.
389 This could be the output of df; or perhaps some sort of /etc/mtab record.
392 - Design at hierarchial storage for Bacula. Migration and Clone.
393 - Implement FSM (File System Modules).
394 - Audit M_ error codes to ensure they are correct and consistent.
395 - Add variable break characters to lex analyzer.
396 Either a bit mask or a string of chars so that
397 the caller can change the break characters.
398 - Make a single T_BREAK to replace T_COMMA, etc.
399 - Ensure that File daemon and Storage daemon can
400 continue a save if the Director goes down (this
401 is NOT currently the case). Must detect socket error,
402 buffer messages for later.
403 - Enhance time/duration input to allow multiple qualifiers e.g. 3d2h
404 - Add ability to backup to two Storage devices (two SD sessions) at
405 the same time -- e.g. onsite, offsite.
406 - Add the ability to consolidate old backup sets (basically do a restore
407 to tape and appropriately update the catalog). Compress Volume sets.
408 Might need to spool via file is only one drive is available.
409 - Compress or consolidate Volumes of old possibly deleted files. Perhaps
410 someway to do so with every volume that has less than x% valid
414 Migration: Move a backup from one Volume to another
415 Clone: Copy a backup -- two Volumes
417 Bacula Migration is based on Jobs (apparently Networker is file by file).
419 Migration triggered by:
423 Highwater mark (keep total size)
427 Bacula Projects Roadmap
429 last update 8 May 2003
431 Item 1: Multiple simultaneous Jobs. (done)
432 Done -- Restore part needs better implementation to work correctly
433 Also, it needs considerable testing
435 What: Permit multiple simultaneous jobs in Bacula.
437 Why: An enterprise level solution needs to go fast without the
438 need for the system administrator to carefully tweak
439 timing. Based on the benchmarks, during a full
440 backup, NetWorker typically hit 10 times the bandwidth to
441 the tape compared to Bacula--largely. This is probably due to
442 running parallel jobs and multi-threaded filling of buffers
443 and writing them to tape. This should also make things work
444 better when you have a mix of fast and slow machines backing
447 Notes: Bacula was designed to run multiple simultaneous jobs. Thus
448 implementing this is a matter of some small cleanups and
452 Item 2: Make the Storage daemon use intermediate file storage to buffer data.
453 Deferred -- not necessary yet -- possibly implement with Migration.
455 What: If data is coming into the SD too fast, buffer it to
456 disk if the user has configured this option.
458 Why: This would be nice, especially if it more or less falls out
459 when implementing (1) above. If not, it probably should not
460 be given a high priority because fundamentally the backup time
461 is limited by the tape bandwidth. Even though you may finish a
462 client job quicker by spilling to disk, you still have to
463 eventually get it onto tape. If intermediate disk buffering
464 allows us to improve write bandwidth to tape, it may make
467 Notes: Whether or not this is implemented will depend upon performance
468 testing after item 1 is implemented.
471 Item 3: Write the bscan program -- also write a bcopy program.
474 What: Write a program that reads a Bacula tape and puts all the
475 appropriate data into the catalog. This allows recovery
476 from a tape that is no longer in the database, or it allows
477 re-creation of a database if lost.
479 Why: This is a fundamental robustness and disaster recovery tool
480 which will increase the comfort level of a sysadmin
481 considering adopting Bacula.
483 Notes: A skeleton of this program already exists, but much work
484 needs to be done. Implementing this will also make apparent
485 any deficiencies in the current Bacula tape format.
488 Item 4: Implement Base jobs.
490 What: A base job is sort of like a Full save except that you
491 will want the FileSet to contain only files that are unlikely
492 to change in the future (i.e. a snapshot of most of your
493 system after installing it). After the base job has been run,
494 when you are doing a Full save, you can specify to exclude
495 all files saved by the base job that have not been modified.
497 Why: This is something none of the competition does, as far as we know
498 (except BackupPC, which is a Perl program that saves to disk
499 only). It is big win for the user, it makes Bacula stand out
500 as offering a unique optimization that immediately saves time
503 Notes: Big savings in tape usage. Will require more resources because
504 the e. DIR must send FD a list of files/attribs, and the FD must
505 search the list and compare it for each file to be saved.
508 Item 5: Implement Label templates
511 What: This is a mechanism whereby Bacula can automatically create
512 a tape label for new tapes according to a detailed specification
513 provided by the user.
515 Why: It is a major convenience item for folks who use automated label
518 Notes: Bacula already has a working form of automatic tape label
519 creation, but it is very crude. The design for the complete
520 tape labeling project is already documented in the manual.
523 Item 6: Write a regression script.
524 Done -- Continue to expand its testing.
526 What: This is an automatic script that runs and tests as many features
527 of Bacula as possible. The output is compared to previous
528 versions of Bacula and any differences are reported.
530 Why: This is an enormous help in preventing introduction of new
531 errors in parts of the program that already work correctly.
533 Notes: This probably should be ranked higher, it's something the typical
534 user doesn't see. Depending on how it's implemented, it may
535 make sense to defer it until the archival tape format and
536 user interface mature.
539 Item 7: GUI for interactive restore
540 Item 8: GUI for interactive backup
542 What: The current interactive restore is implemented with a tty
543 interface. It would be much nicer to be able to "see" the
544 list of files backed up in typical GUI tree format.
545 The same mechanism could also be used for creating
546 ad-hoc backup FileSets (item 8).
548 Why: Ease of use -- especially for the end user.
550 Notes: Rather than implementing in Gtk, we probably should go directly
551 for a Browser implementation, even if doing so meant the
552 capability wouldn't be available until much later. Not only
553 is there the question of Windows sites, most
554 Solaris/HP/IRIX, etc, shops can't currently run Gtk programs
555 without installing lots of stuff admins are very wary about.
556 Real sysadmins will always use the command line anyway, and
557 the user who's doing an interactive restore or backup of his
558 own files will in most cases be on a Windows machine running
562 Item 9: Add SSL to daemon communications.
564 What: This provides for secure communications between the daemons.
566 Why: This would allow doing backup across the Internet without
567 privacy concerns (or with much less concern).
569 Notes: The vast majority of near term potential users will be backing up
570 a single site over a LAN and, correctly or not, they probably
571 won't be concerned with security, at least not enough to go to
572 the trouble to set up keys, etc. to screw things down. We suspect
573 that many users genuinely interested in multi-site backup
574 already run some form of VPN software in their internetwork
575 connections, and are willing to delegate security to that layer.
578 Item 10: Define definitive tape format.
581 What: Define that definitive tape format that will not change
582 for the next millennium.
584 Why: Stability, security.
586 Notes: See notes for item 11 below.
589 Item 11: New daemon communication protocol.
591 What: The current daemon to daemon protocol is basically an ASCII
592 printf() and sending the buffer. On the receiving end, the
593 buffer is sscanf()ed to unpack it. The new scheme would
594 be a binary format that allows quick packing and unpacking
595 of any data type with named fields.
597 Why: Using binary packing would be faster. Named fields will permit
598 error checking to ensure that what is sent is what the
599 receiver really wants.
601 Notes: These are internal improvements in the interest of the
602 long-term stability and evolution of the program. On the one
603 hand, the sooner they're done, the less code we have to rip
604 up when the time comes to install them. On the other hand, they
605 don't bring an immediately perceptible benefit to potential
606 users. Item 10 and possibly item 11 should be deferred until Bacula
607 is well established with a growing user community more or
608 less happy with the feature set. At that time, it will make a
609 good "next generation" upgrade in the interest of data
615 ======================================================
617 It is somewhat like a Full save becomes an incremental since
618 the Base job (or jobs) plus other non-base files.
620 - New BaseFile table that contains:
621 JobId, BaseJobId, FileId (from Base).
622 i.e. for each base file that exists but is not saved because
623 it has not changed, the File daemon sends the JobId, BaseId,
624 and FileId back to the Director who creates the DB entry.
625 - To initiate a Base save, the Director sends the FD
626 the FileId, and full filename for each file in the Base.
627 - When the FD finds a Base file, he requests the Director to
628 send him the full File entry (stat packet plus MD5), or
629 conversely, the FD sends it to the Director and the Director
630 says yes or no. This can be quite rapid if the FileId is kept
631 by the FD for each Base Filename.
632 - It is probably better to have the comparison done by the FD
633 despite the fact that the File entry must be sent across the
635 - An alternative would be to send the FD the whole File entry
636 from the start. The disadvantage is that it requires a lot of
637 space. The advantage is that it requires less communications
639 - The Job record must be updated to indicate that one or more
641 - At end of Job, FD returns:
642 1. Count of base files/bytes not written to tape (i.e. matches)
643 2. Count of base file that were saved i.e. had changed.
644 - No tape record would be written for a Base file that matches, in the
645 same way that no tape record is written for Incremental jobs where
646 the file is not saved because it is unchanged.
647 - On a restore, all the Base file records must explicitly be
648 found from the BaseFile tape. I.e. for each Full save that is marked
649 to have one or more Base Jobs, search the BaseFile for all occurrences
651 - An optimization might be to make the BaseFile have:
657 This would avoid the need to explicitly fetch each File record for
658 the Base job. The Base Job record will be fetched to get the
659 VolSessionId and VolSessionTime.
660 =========================================================
664 =============================================================
666 Request For Comments For File Backup Options
669 Subject: File Backup Options
672 A few days ago, a Bacula user who is backing up to file volumes and
673 using compression asked if it was possible to suppress compressing
674 all .gz files since it was a waste of CPU time. Although Bacula
675 currently permits using different options (compression, ...) on
676 a directory by directory basis, it cannot do it on a file by
677 file basis, which is clearly what was desired.
679 Proposed Implementation:
680 To solve this problem, I propose the following:
682 - Add a new Director resource type called FileOptions.
684 - The FileOptions resource will have records for all
685 options that can currently be specified on the Include record
686 (in a FileSet). Examples below.
688 - The FileOptions resource will permit an exclude option as well
689 as a number of additional options.
691 - The heart of the FileOptions resource is the ability to
692 supply any number of ApplyTo records which specify POSIX
693 regular expressions. These ApplyTo regular expressions are
694 applied to the fully qualified filename (path and all). If
695 one matches, then the FileOptions will be used.
697 - When an ApplyTo specification matches an included file, the
698 options specified in the FileOptions resource will override
699 the default options specified on the Include record.
701 - Include records will be modified to permit referencing one or
702 more FileOptions resources. The FileOptions will be used
703 in the order listed on the Include record and the first
704 one that matches will be applied.
706 - Options (or specifications) currently supplied on the Include
707 record will be deprecated (i.e. removed in a later version a
708 year or so from now).
710 - The Exclude record will be deprecated as the same functionality
711 can be obtained by using an Exclude = yes in the FileOptions.
714 The following records can appear in the FileOptions resource. An
715 asterisk preceding the name indicates a feature not currently
719 - Compression= (GZIP, ...)
720 - Signature= (MD5, SHA1, ...)
722 - OneFs= (yes/no) - remain on one filesystem
723 - Recurse= (yes/no) - recurse into subdirectories
724 - Sparse= (yes/no) - do sparse file backup
725 - *Exclude= (yes/no) - exclude file from being saved
726 - *Reader= (filename) - external read (backup) program
727 - *Plugin= (filename) - read/write plugin module
730 - verify= (ipnougsamc5) - verify options
733 - replace= (always/ifnewer/ifolder/never) - replace options currently
735 - *Writer= (filename) - external write (restore) program
739 Currently options specifying compression, MD5 signatures, recursion,
740 ... of a FileSet are supplied on the Include record. These will now
741 all be collected into a FileOptions resource, which will be
742 specified on the Include in place of the options. Multiple FileOptions
743 may be specified. Since the FileOptions contain regular expressions
744 that are applied to the full filename, this will give the ability
745 to specify backup options on a file by file basis to whatever level
754 Include = compression=GZIP signature=MD5 {
763 Include = FileOptions=Opts {
774 That's a lot more to do the same thing, but it gives the ability to
775 apply options on a file by file basis. For example, suppose you
776 want to compress all files but not any file with extensions .gz or .Z.
777 You could do so as follows:
781 Include = FileOptions=NoCompress FileOptions=Opts {
789 ApplyTo = /*.?*/ # matches all files
794 # Note multiple ApplyTos are ORed
795 ApplyTo = /*.gz/ # matches .gz files */
796 ApplyTo = /*.Z/ # matches .Z files */
799 Now, since the NoCompress FileOptions is specified first on the
800 Include line, any *.gz or *.Z file will have an MD5 signature computed,
801 but will not be compressed. For all other files, the NoCompress will not
802 match, so the Opts options will be used which will include GZIP
806 - Is it necessary to provide some means of ANDing regular expressions
807 and negation? (not currently planned)
809 e.g. ApplyTo = /*.gz/ && !/big.gz/
811 - I see that Networker has a "null" module which, if specified, does not
812 backup the file, but does make an record of the file in the catalog
813 so that the catalog will reflect an exact picture of the filesystem.
814 The result is that the file can be "seen" when "browsing" the save
815 sets, but it cannot be restored.
817 Is this really useful? Should it be implemented in Bacula?
820 After implementing the above, the user will be able to specify
821 on a file by file basis (using regular expressions) what options are
822 applied for the backup.
825 =============================================
827 ==========================================================
829 For each Incremental job that is run, there may be files that
830 were found but not saved because they were locked (this applies
831 only to Windows). Such a system could send back to the Director
832 a list of Unsaved files.
834 - New UnSavedFiles table that contains:
838 - Then in the next Incremental job, the list of Unsaved Files will be
839 feed to the FD, who will ensure that they are explicitly chosen even
840 if standard date/time check would not have selected them.
841 =============================================================
844 Done: (see kernsdone for more)
845 - Heartbeat between daemons.
846 - Fix Dir heartbeat in restore and verify vol. Be sure to make
847 bnet_recv() ignore BNET_HEARTBEAT.
848 - Implement HEART_BEAT while SD waiting for tapes.
849 - Include RunBeforeJob and RunAfterJob output in the message
851 - Change M_INFO to M_RESTORED for all restored files.
852 - Fix command prompt in gnome-console by checking on Ready.
853 - Merge SQLite, MySQL, and Rel spec into a single file.
854 - Fix config of "console"
855 - Check if cancel works with FD (fixed).
856 - Properly configure console and gconsole (currently for source not
857 configured for installation).
858 - Error labeling tape from console gets Jmsg error because of no Job.
859 - Test and implement get_pint and get_yesno.
860 - Implement global with DB name and add to btraceback.gdb
861 - Remove subsysdir from conf files (used only in autostart scripts).
863 rufus-dir: Max configured use duration exceeded. Marking Volume "MatouBackup" as Used.
864 rufus-sd: Volume "" previously written, moving to end of data.
865 rufus-sd: Matou.2003-05-10_10.39.18 Error: I canot write on this volume because:
866 The number of files mismatch! Volume=1 Catalog=0
867 rufus-sd: Matou.2003-05-10_10.39.18 Error: askdir.c:155 NULL Volume name. This shouldn't happen!!!
868 - Shell character expansion is failing occassionally.
869 - Add a section to the doc on Manual cycling of Volumes.
870 - Check if Job/File retentions apply to multivolume jobs.
871 - Fix missing casette in autoloader during read:
872 14-May-2003 14:41 undef-sd: RestoreFiles.2003-05-14_14.41.00 Warning: acquire.c:106 Volume name mismatch. Wanted TestVolume0005 got TestVolume0010
873 14-May-2003 14:41 undef-sd: 3301 Issuing autochanger "loaded" command.
874 14-May-2003 14:41 undef-sd: 3302 Issuing autochanger "unload" command.
875 14-May-2003 14:42 undef-sd: 3303 Issuing autochanger "load slot 1" command.
876 14-May-2003 14:42 undef-sd: 3304 Autochanger "load slot 1" status is OK.
877 14-May-2003 14:42 undef-sd: RestoreFiles.2003-05-14_14.41.00 Warning: acquire.c:106 Volume name mismatch. Wanted TestVolume0005 got TestVolume0009
878 14-May-2003 14:42 undef-sd: 3301 Issuing autochanger "loaded" command.
879 14-May-2003 14:42 undef-sd: RestoreFiles.2003-05-14_14.41.00 Warning: acquire.c:106 Volume name mismatch. Wanted TestVolume0005 got TestVolume0009
880 14-May-2003 14:42 undef-sd: 3301 Issuing autochanger "loaded" command.
881 14-May-2003 14:42 undef-sd: RestoreFiles.2003-05-14_14.41.00 Warning: acquire.c:106 Volume name mismatch. Wanted TestVolume0005 got TestVolume0009
882 14-May-2003 14:42 undef-sd: 3301 Issuing autochanger "loaded" command.
883 14-May-2003 14:42 undef-sd: RestoreFiles.2003-05-14_14.41.00 Warning: acquire.c:106 Volume name mismatch. Wanted TestVolume0005 got TestVolume0009
884 14-May-2003 14:42 undef-sd: 3301 Issuing autochanger "loaded" command.
885 14-May-2003 14:42 undef-sd: RestoreFiles.2003-05-14_14.41.00 Fatal error: acquire.c:129 Too many errors trying to mount device "/dev/nrsa0".
886 14-May-2003 14:42 undef-dir: Bacula 1.31 (12May03): 14-May-2003 14:42
887 - Fix problem reported by Christopher McCurdy <cmccurdy@eecis.udel.edu>
888 xeon-fd: Could not stat c:/Documents and Settings/All
889 Users/Application Data/Humc:\Documents and Settings\All User98_AIX.kbf:
890 ERR=No such file or directory
892 - The following Re-read last block at EOT failed. ERR=block.c:523 Read zero bytes on device /dev/nrsa0.
893 undef-sd: block.c:523 Read zero bytes on device /dev/nrsa0.
894 apparently masks the standard EOM message.
895 - BSD (probably) does not have strtoll()
896 - BSD does not have ioctl() MTEOM
897 - BSD defines a number of MT_xxx variables which conflict
898 with those defined by Bacula.
899 - Make default duration days if no qualifier (e.g. s) is specified.
900 - BSDI fix finding gcc version
901 - When the FD errs (e.g. disk full) have a more graceful shutdown.
902 - Make sure Bacula prunes/purges canceled and failed jobs too and all
903 jobs with zero JobFiles.
904 - Implement Volume name checking.
905 - Document what characters can go into Volume names.
906 - Getting the following on all directories on Win32
907 19-May-2003 01:14 tibs-fd: Could not access c:/cygwin/home/kern/rxvt: ERR=Permission denied
908 - Cancellation caused JobMedia error:
909 babylon5-dir: Last FULL backup time not found. Doing FULL backup.
910 babylon5-dir: Start Backup JobId 416, Job=Zocalo_Save.2003-05-19_02.15.06
911 babylon5-sd: End of media on Volume VXA-V17-Inc-001 Bytes=31,982,900,672 Blocks=495,781.
912 babylon5-sd: Job Zocalo_Save.2003-05-19_02.15.06 waiting. Cannot find any appendable volumes.
913 babylon5-sd: Someone woke me up, but I cannot find any appendable volumes
914 for Job=Zocalo_Save.2003-05-19_02.15.06.
915 babylon5-sd: Zocalo_Save.2003-05-19_02.15.06 Fatal error: Job
916 Zocalo_Save.2003-05-19_02.15.06 canceled while waiting for mount on
917 Storage Device "Ecrix_VXA-1".
918 babylon5-sd: Zocalo_Save.2003-05-19_02.15.06 Fatal error: Cannot fixup device
919 error. Job Zocalo_Save.2003-05-19_02.15.06 canceled while waiting for
920 mount on Storage Device "Ecrix_VXA-1".
922 babylon5-dir: Zocalo_Save.2003-05-19_02.15.06 Error: Catalog error creating
923 JobMedia record. sql_create.c:125 Create JobMedia failed. Record already
925 babylon5-sd: Zocalo_Save.2003-05-19_02.15.06 Error: Error creating JobMedia
926 record: 1991 Update JobMedia error
928 babylon5-sd: Zocalo_Save.2003-05-19_02.15.06 Error: askdir.c:158 NULL Volume
929 name. This shouldn't happen!!!
930 zocalo-fd: Zocalo_Save.2003-05-19_02.15.06 Error: bnet.c:310 Write error
931 sending to Storage daemon:babylon5:9103: ERR=Broken pipe
932 - Volume names with spaces get jammed into the catalog with 0x1
933 i.e. the SD bashes the Volume but they are not unbased by Dir.
934 jerom-dir: MonthlySave.2003-05-10_17.12.01 Error: Unable to get Media
935 record for Volume Tape^A1: ERR=sql_get.c:788 Media record for Volume
937 jerom-sd: MonthlySave.2003-05-10_17.12.01 Error: Error updating Volume
938 Info: 1991 Catalog Request failed: sql_get.c:788 Media record for Volume
940 - ChangeServiceConfig2A does not exist on WinNT (ADVAPI32.DLL).
941 - Fix "access not allowed" for backup of files on WinXP.
942 - Check for existence of all new Win32 API's. See LoadLibrary in
944 - Count errors during restore and print them in the Job report.
945 - Bug: fix access problems on files restored on WinXP.
946 - Put system type returned by FD into catalog.
947 - Finish WIN32_DATA stream code (bextract, check if can handle stream)
948 - Make SD keep track of Files, Bytes during restore.
949 - If you enter the userid by hand for restore, you get:
950 Enter JobId(s), comma separated, to restore: 74
951 You have selected the following JobId: 74
952 Building directory tree for JobId 74 ...
953 134645140 items inserted into the tree and marked for extraction.
954 - Add SDWriteSeqNo to SD, and probably Read on FD side.
955 - If bootstrap is non-zero for restore, do not show JobId in the
956 OK to run? (yes/mod/no): list.
957 - When all cassettes in magazine are used, got:
958 22-May-2003 18:24 undef-sd: 3304 Autochanger "load slot 1" status is OK.
959 22-May-2003 18:24 undef-sd: NightlySave.2003-05-22_14.08.16 Warning: mount.c:245 Director wanted Volume "TestVolume0009".
960 Current Volume "TestVolume0005" not acceptable because:
961 1998 Volume "TestVolume0005" not Append or Recycle.
962 22-May-2003 18:24 undef-sd: NightlySave.2003-05-22_14.08.16 Error: Autochanger Volume "TestVolume0009" not found in slot 1.
963 Setting slot to zero in catalog.
964 22-May-2003 18:24 undef-sd: Please mount Volume "TestVolume0009" on Storage Device "ARCHIVE 4586" for Job NightlySave.2003-05-22_14
966 Use "mount" command to release Job.
967 22-May-2003 19:24 undef-sd: Please mount Volume "TestVolume0009" on Storage Device "ARCHIVE 4586" for Job NightlySave.2003-05-22_14
969 Use "mount" command to release Job.
970 - Don't zero the Slot when the wrong volume is found -- simply ask
972 - Implement MTIOCERRSTAT on FreeBSD to clear tape error conditions.
973 - Shell expansion fails for working_directory in SD from time to time.
974 - File the Automatically selected: xxx
975 to say Automatically selected Pool: xxx
976 - Default duration with no qualifier is sec should be 1 day
977 - zap sd_auth_key in SD after FD connection.
978 - Find a solution for the multiple FileSet problem (when it is changed). Add date?
979 - Look at Python for a Bacula scripting language -- www.python.org
980 - When Marking a file in Restore that is a hard link, also
981 mark the link so that the data will be reloaded.
982 - Emergency restore info:
984 - Backup working directory
986 - Why don't we get an error message from Win32 FD when bootstrap
987 file cannot be created for restore command?
988 - Fix Win2000 error with no messages during startup.
989 - Make restore more robust in counting error and not immediately bailing
990 out. Also print error message once, but try to continue.
991 - Add code to check that blocks are sequential on restore.
992 - Remove "rufus" and such references from regress.
993 - No READLINE_SRC if found in alternate directory.
994 - If ./btape is called without /dev, assume argument is a Storage resource name.
995 - Find general solution for sscanf size problems (as well as sprintf. Do at run time?
996 - Bytes restored is wrong.
997 - The "List last 20 Jobs run" doesnt work correctly in restore.
998 It doesnt show the last 20 jobs , but some older ones.
999 - Fix Verify VolumeToCatalog to use BSRs -- it is broken.
1000 - Implement Release Storage=xxx
1001 - Fix restore on Win95/98
1002 - Remove the Jmsg() in sql_find.c:102 or only print on hard error.
1003 - Implement FileSet VolIndex -- done, but must update old records.
1004 - Check this below from Phil.
1005 This was SD reported data rather than FD data!
1006 > When the job was done, Bacula reported 11084 files restored:
1009 > Job: Zocalo_Restore.2003-06-05_16.42.01
1011 > Start time: 05-Jun-2003 16:42
1012 > End time: 06-Jun-2003 01:21
1013 > Files Restored: 11,084
1014 > Bytes Restored: 65,474,772
1016 > FD termination status: OK
1017 > Termination: Restore OK
1019 > when it should probably have reported 11084 files scanned, 250 restored.
1020 > The bytes restored count looks about right.
1022 - Should Bacula make an Append tape as Purged when purging?
1023 - Use switch() in backup.c and restore.c in FD instead of giant if statement.
1024 - If during a restore, a hard linked file already exists (on option), delete
1025 the file and re-link it. This is to avoid the possibility that the
1026 user had re-linked the file between the backup and the restore.
1027 Do lstat() to see if it is already properly linked.
1028 Same for symlinked file.
1029 Make sure ifnewer, ifolder, never, ... apply correctly.
1030 - Flag so that no connect does not error, and Reschedule a job.
1031 - Implement "Reschedule OnError=yes interval=nnn times=xxx"
1032 - That restoring a hard link that already exists works correctly.
1034 - Make Pool resource handle Counter resources.
1035 - Fix first block number after label to be zero instead of 1 (reset after label).
1036 - Grep for Backup OK in regression script.
1037 - Do NOT reuse same JobId if tape written.
1038 - Implement non-blocking writes and bsock->terminate in heartbeat
1039 thread, or set it in status.c cancel (used pthread_kill() instead of
1041 - Add restore to specific date.
1042 - Instrument use_count on DEVICE packets and ensure that the device is
1043 being close()ed at the appropriate time.
1044 - Test long path names (>64 chars) in Windows -- crashes FD?
1045 - Implement fast block rejection: match_bsr_block().
1046 - Complain if record dropped in bnet_recv because too long.
1047 - Test multiple simultaneous Volumes
1048 - Document recycling algorithm.
1049 - Make Restore report an error if FD or SD term codes are not OK.
1050 - To link with mysqlclient_r may require -lssl -lcrypto
1051 - Document Heart beat code
1052 - Non-fatal errors are not correct counting attribs.c:277
1053 - Check that Block number in JobMedia are correct.
1054 - The bsr for Dan's job has file indexes covering the whole range rather
1055 than only the range contained on the volume.
1056 Constrain FileIndex to be within range for Volume.
1057 - Pass prefix_links to FD.
1058 - Fix restore list of volumes if Volume not selected.
1059 - Complete (or turn off) the command line code in update.
1060 - Testing Tibs job erred and hung director on Storage resource. This was
1061 because there were a whole pile of jobs hanging around in the SD
1062 waiting for a connection from the FD that was never coming.
1063 - Possibly update all client records at startup.
1064 - Listing of Win32 restored files has ??????? for user.
1065 - In Win portable restore the directory is not create
1066 27-Jun-2003 16:52 tibs-fd: kernsrestore.2003-06-27_16.52.20 Error:
1067 create_file.c:175 Could not create
1068 /tmp/bacula-restores/cygwin/home/kern/bacula/k/src/dird/dird_conf.o: 0
1069 ERR=The system cannot find the path specified.
1070 - Finish Windows implementation (add setting of correct type on restore,
1071 add Portable Data Format flag).
1072 - Document portable=yes
1073 - Document c:/working directory better than /working directory.
1074 - Document run "yes".
1075 - Document update volume: jobid, current, before, all
1076 - Document that bscan does not work with multiple simultaneous jobs.
1077 - Update Automatic Volume Labeling in disk.wml
1078 - Figure out how to handle DHCP IP addresses -- use dynamic DNS.
1079 - Add OS, GCC version to traceback output.