9 - Document cleaning up the spool files:
10 db, pid, state, bsr, mail, conmsg, spool
11 - Document the multiple-drive-changer.txt script.
12 - Pruning with Admin job.
13 - Does WildFile match against full name? Doc.
14 - %d and %v only valid on Director, not for ClientRunBefore/After.
15 - During tests with the 260 char fix code, I found one problem:
16 if the system "sees" a long path once, it seems to forget it's
17 working drive (e.g. c:\), which will lead to a problem during
18 the next job (create bootstrap file will fail). Here is the
19 workaround: specify absolute working and pid directory in
20 bacula-fd.conf (e.g. c:\bacula\working instead of
22 - Document techniques for restoring large numbers of files.
23 - Document setting my.cnf to big file usage.
24 - Add example of proper index output to doc.
26 - Correct the Include syntax in the m4.xxx files in examples/conf
27 - Document JobStatus and Termination codes.
32 - Look at why SIGPIPE during connection can cause seg fault in
33 writing the daemon message, when Dir dropped to bacula:bacula
34 - Look at zlib 32 => 64 problems.
35 - Ensure that connection to daemon failure always indicates what
36 daemon it was trying to connect to.
37 - Try turning on disk seek code.
38 - Possibly turn on St. Bernard code.
39 - Fix bscan to report the JobType when restoring a job.
40 - Fix bextract to restore ACLs, or better yet, use common
42 - Do we migrate appendable Volumes?
43 - Remove queue.c code.
44 - Add bconsole option to use stdin/out instead of conio.
45 - Fix ClientRunBefore/AfterJob compatibility.
46 - Fix re-read of last block to check if job has actually written
47 a block, and check if block was written by a different job
48 (i.e. multiple simultaneous jobs writing).
49 - Some users claim that they must do two prune commands to get a
50 Volume marked as purged.
51 - Print warning message if LANG environment variable does not specify
53 - New dot commands from Arno.
54 .update volume [enabled|disabled|*see below]
55 > However, I could easily imagine an option to "update slots" that says
56 > "enable=yes|no" that would automatically enable or disable all the Volumes
57 > found in the autochanger. This will permit the user to optionally mark all
58 > the Volumes in the magazine disabled prior to taking them offsite, and mark
59 > them all enabled when bringing them back on site. Coupled with the options
60 > to the slots keyword, you can apply the enable/disable to any or all volumes.
61 .show device=xxx lists information from one storage device, including
62 devices (I'm not even sure that information exists in the DIR...)
63 .move eject device=xxx mostly the same as 'unmount xxx' but perhaps with
64 better machine-readable output like "Ok" or "Error busy"
65 .move eject device=xxx toslot=yyy the same as above, but with a new
66 target slot. The catalog should be updated accordingly.
67 .move transfer device=xxx fromslot=yyy toslot=zzz
70 - Get Perl replacement for bregex.c
71 - Given all the problems with FIFOs, I think the solution is to do something a
72 little different, though I will look at the code and see if there is not some
73 simple solution (i.e. some bug that was introduced). What might be a better
74 solution would be to use a FIFO as a sort of "key" to tell Bacula to read and
75 write data to a program rather than the FIFO. For example, suppose you
80 Then, I could imagine if you backup and restore this file with a direct
81 reference as is currently done for fifos, instead, during backup Bacula will
84 /home/kern/my-fifo.backup
86 and read the data that my-fifo.backup writes to stdout. For restore, Bacula
89 /home/kern/my-fifo.restore
91 and send the data backed up to stdout. These programs can either be an
92 executable or a shell script and they need only read/write to stdin/stdout.
94 I think this would give a lot of flexibility to the user without making any
95 significant changes to Bacula.
100 select FilenameId from Filename where Name='';
101 # Get list of all directories referenced in a Backup.
102 select Path.Path from Path,File where File.JobId=nnn and
103 File.FilenameId=(FilenameId-from-above) and File.PathId=Path.PathId
104 order by Path.Path ASC;
106 - Look into using Dart for testing
107 http://public.kitware.com/Dart/HTML/Index.shtml
109 - Look into replacing autotools with cmake
110 http://www.cmake.org/HTML/Index.html
112 === Migration from David ===
113 What I'd like to see:
116 Name = "<poolname>-migrate"
120 Migration Selection Type = LowestUtil | OldestVol | PoolOccupancy |
121 Client | PoolResidence | Volume | JobName | SQLquery
122 Migration Selection Pattern = "regexp"
123 Next Pool = <override>
126 There should be no need for a Level (migration is always Full, since you
127 don't calculate differential/incremental differences for migration),
128 Storage should be determined by the volume types in the pool, and Client
129 is really a selection issue. Migration should always occur to the
130 NextPool defined in the pool definition. If no nextpool is defined, the
131 job should end with a reason of "no place to go". If Next Pool statement
132 is present, we override the check in the pool definition and use the
135 Here's how I'd define Migration Selection Types:
138 Client -- Migrate data from selected client only. Migration Selection
139 Pattern regexp provides pattern to select client names, eg ^FS00* makes
140 all client names starting with FS00 eligible for migration.
142 Jobname -- Migration all jobs matching name. Migration Selection Pattern
143 regexp provides pattern to select jobnames existing in pool.
145 Volume -- Migrate all data on specified volumes. Migration Selection
146 Pattern regexp provides selection criteria for volumes to be migrated.
147 Volumes must exist in pool to be eligible for migration.
151 LowestUtil -- Identify the volume in the pool with the least data on it
152 and empty it. No Migration Selection Pattern required.
154 OldestVol -- Identify the LRU volume with data written, and empty it. No
155 Migration Selection Pattern required.
157 PoolOccupancy -- if pool occupancy exceeds <highmig>, migrate volumes
158 (starting with most full volumes) until pool occupancy drops below
159 <lowmig>. Pool highmig and lowmig values are in pool definition, no
160 Migration Selection Pattern required.
164 SQLQuery -- Migrate all jobuids returned by the supplied SQL query.
165 Migration Selection Pattern contains SQL query to execute; should return
166 a list of 1 or more jobuids to migrate.
168 PoolResidence -- Migrate data sitting in pool for longer than
169 PoolResidence value in pool definition. Migration Selection Pattern
170 optional; if specified, override value in pool definition (value in
174 [ possibly a Python event -- kes ]
176 - Mount on an Autochanger with no tape in the drive causes:
177 Automatically selected Storage: LTO-changer
178 Enter autochanger drive[0]: 0
179 3301 Issuing autochanger "loaded drive 0" command.
180 3302 Autochanger "loaded drive 0", result: nothing loaded.
181 3301 Issuing autochanger "loaded drive 0" command.
182 3302 Autochanger "loaded drive 0", result: nothing loaded.
183 3902 Cannot mount Volume on Storage Device "LTO-Drive1" (/dev/nst0) because:
184 Couldn't rewind device "LTO-Drive1" (/dev/nst0): ERR=dev.c:678 Rewind error on "LTO-Drive1" (/dev/nst0). ERR=No medium found.
185 3905 Device "LTO-Drive1" (/dev/nst0) open but no Bacula volume is mounted.
186 If this is not a blank tape, try unmounting and remounting the Volume.
187 - If Drive 0 is blocked, and drive 1 is set "Autoselect=no", drive 1 will
189 - Autochanger did not change volumes.
190 select * from Storage;
191 +-----------+-------------+-------------+
192 | StorageId | Name | AutoChanger |
193 +-----------+-------------+-------------+
194 | 1 | LTO-changer | 0 |
195 +-----------+-------------+-------------+
196 05-May 03:50 roxie-sd: 3302 Autochanger "loaded drive 0", result is Slot 11.
197 05-May 03:50 roxie-sd: Tibs.2006-05-05_03.05.02 Warning: Director wanted Volume "LT
198 Current Volume "LT0-002" not acceptable because:
199 1997 Volume "LT0-002" not in catalog.
200 05-May 03:50 roxie-sd: Tibs.2006-05-05_03.05.02 Error: Autochanger Volume "LT0-002"
201 Setting InChanger to zero in catalog.
202 05-May 03:50 roxie-dir: Tibs.2006-05-05_03.05.02 Error: Unable to get Media record
204 05-May 03:50 roxie-sd: Tibs.2006-05-05_03.05.02 Fatal error: Error getting Volume i
205 05-May 03:50 roxie-sd: Tibs.2006-05-05_03.05.02 Fatal error: Job 530 canceled.
206 05-May 03:50 roxie-sd: Tibs.2006-05-05_03.05.02 Fatal error: spool.c:249 Fatal appe
207 05-May 03:49 Tibs: Tibs.2006-05-05_03.05.02 Fatal error: c:\cygwin\home\kern\bacula
216 FirstWritten: 2006-05-05 03:11:54
217 LastWritten: 2006-05-05 03:50:23
218 LabelDate: 2005-12-26 16:52:40
229 VolRetention: 31,536,000
241 Note VolStatus is blank!!!!!
248 FirstWritten: 0000-00-00 00:00:00
249 LastWritten: 0000-00-00 00:00:00
250 LabelDate: 2005-12-26 16:52:40
261 VolRetention: 31,536,000
274 Automatically selected Storage: LTO-changer
275 Enter autochanger drive[0]: 0
276 3301 Issuing autochanger "loaded drive 0" command.
277 3302 Autochanger "loaded drive 0", result: nothing loaded.
278 3301 Issuing autochanger "loaded drive 0" command.
279 3302 Autochanger "loaded drive 0", result: nothing loaded.
280 3902 Cannot mount Volume on Storage Device "LTO-Drive1" (/dev/nst0) because:
281 Couldn't rewind device "LTO-Drive1" (/dev/nst0): ERR=dev.c:678 Rewind error on "LTO-Drive1" (/dev/nst0). ERR=No medium found.
283 3905 Device "LTO-Drive1" (/dev/nst0) open but no Bacula volume is mounted.
284 If this is not a blank tape, try unmounting and remounting the Volume.
286 - Add VolumeState (enable, disable, archive)
287 - Add VolumeLock to prevent all but lock holder (SD) from updating
288 the Volume data (with the exception of VolumeState).
289 - The btape fill command does not seem to use the Autochanger
290 - Make Windows installer default to system disk drive.
291 - Look at using ioctl(FIOBMAP, ...) on Linux, and
292 DeviceIoControl(..., FSCTL_QUERY_ALLOCATED_RANGES, ...) on
293 Win32 for sparse files.
294 http://www.flexhex.com/docs/articles/sparse-files.phtml
295 http://www.informatik.uni-frankfurt.de/~loizides/reiserfs/fibmap.html
296 - Directive: at <event> "command"
297 - Command: pycmd "command" generates "command" event. How to
298 attach to a specific job?
299 - Integrate Christopher's St. Bernard code.
300 - run_cmd() returns int should return JobId_t
301 - get_next_jobid_from_list() returns int should return JobId_t
302 - Document export LDFLAGS=-L/usr/lib64
303 - Don't attempt to restore from "Disabled" Volumes.
304 - Network error on Win32 should set Win32 error code.
305 - What happens when you rename a Disk Volume?
306 - Job retention period in a Pool (and hence Volume). The job would
308 - Detect resource deadlock in Migrate when same job wants to read
309 and write the same device.
310 - Queue warning/error messages during restore so that they
311 are reported at the end of the report rather than being
312 hidden in the file listing ...
313 - Look at -D_FORTIFY_SOURCE=2
314 - Add Win32 FileSet definition somewhere
315 - Look at fixing restore status stats in SD.
316 - Make selection of Database used in restore correspond to
318 - Look at using ioctl(FIMAP) and FIGETBSZ for sparse files.
319 http://www.informatik.uni-frankfurt.de/~loizides/reiserfs/fibmap.html
320 - Implement a mode that says when a hard read error is
321 encountered, read many times (as it currently does), and if the
322 block cannot be read, skip to the next block, and try again. If
323 that fails, skip to the next file and try again, ...
325 create table LevelType (LevelType binary(1), LevelTypeLong tinyblob);
326 insert into LevelType (LevelType,LevelTypeLong) values
330 - Add ACL to restore only to original location.
331 - Show files/second in client status output.
332 - Add a recursive mark command (rmark) to restore.
333 - "Minimum Job Interval = nnn" sets minimum interval between Jobs
334 of the same level and does not permit multiple simultaneous
335 running of that Job (i.e. lets any previous invocation finish
336 before doing Interval testing).
337 - Look at simplifying File exclusions.
338 - New directive "Delete purged Volumes"
339 - new pool XXX with ScratchPoolId = MyScratchPool's PoolId and
340 let it fill itself, and RecyclePoolId = XXX's PoolId so I can
341 see if it become stable and I just have to supervise
343 - If I want to remove this pool, I set RecyclePoolId = MyScratchPool's
344 PoolId, and when it is empty remove it.
345 - Figure out how to recycle Scratch volumes back to the Scratch Pool.
347 - Allow Check Labels to be used with Bacula labels.
348 - "Resuming" a failed backup (lost line for example) by using the
349 failed backup as a sort of "base" job.
351 - Email to the user when the tape is about to need changing x
352 days before it needs changing.
353 - Command to show next tape that will be used for a job even
354 if the job is not scheduled.
355 - From: Arunav Mandal <amandal@trolltech.com>
356 1. When jobs are running and bacula for some reason crashes or if I do a
357 restart it remembers and jobs it was running before it crashed or restarted
358 as of now I loose all jobs if I restart it.
360 2. When spooling and in the midway if client is disconnected for instance a
361 laptop bacula completely discard the spool. It will be nice if it can write
362 that spool to tape so there will be some backups for that client if not all.
364 3. We have around 150 clients machines it will be nice to have a option to
365 upgrade all the client machines bacula version automatically.
367 4. Atleast one connection should be reserved for the bconsole so at heavy load
368 I should connect to the director via bconsole which at sometimes I can't
370 5. Another most important feature that is missing, say at 10am I manually
371 started backup of client abc and it was a full backup since client abc has
372 no backup history and at 10.30am bacula again automatically started backup of
373 client abc as that was in the schedule. So now we have 2 multiple Full
374 backups of the same client and if we again try to start a full backup of
375 client backup abc bacula won't complain. That should be fixed.
377 - Fix bpipe.c so that it does not modify results pointer.
378 ***FIXME*** calling sequence should be changed.
379 - For Windows disaster recovery see http://unattended.sf.net/
380 - regardless of the retention period, Bacula will not prune the
381 last Full, Diff, or Inc File data until a month after the
382 retention period for the last Full backup that was done.
383 - update volume=xxx --- add status=Full
384 - Remove old spool files on startup.
385 - Exclude SD spool/working directory.
386 - Refuse to prune last valid Full backup. Same goes for Catalog.
388 - Make a callback when Rerun failed levels is called.
389 - Give Python program access to Scheduled jobs.
390 - Add setting Volume State via Python.
391 - Python script to save with Python, not save, save with Bacula.
392 - Python script to do backup.
394 - Change the Priority, Client, Storage, JobStatus (error)
395 at the start of a job.
396 - Why is SpoolDirectory = /home/bacula/spool; not reported
397 as an error when writing a DVD?
398 - Make bootstrap file handle multiple MediaTypes (SD)
399 - Remove all old Device resource code in Dir and code to pass it
400 back in SD -- better, rework it to pass back device statistics.
401 - Check locking of resources -- be sure to lock devices where previously
402 resources were locked.
403 - The last part is left in the spool dir.
406 - In restore don't compare byte count on a raw device -- directory
407 entry does not contain bytes.
408 - To mark files as deleted, run essentially a Verify to disk, and
409 when a file is found missing (MarkId != JobId), then create
410 a new File record with FileIndex == -1. This could be done
411 by the FD at the same time as the backup.
415 MA = (last_MA * 3 + rate) / 4
416 rate = (bytes - last_bytes) / (runtime - last_runtime)
417 - Max Vols limit in Pool off by one?
418 - Implement Files/Bytes,... stats for restore job.
419 - Implement Total Bytes Written, ... for restore job.
420 - Despool attributes simultaneously with data in a separate
421 thread, rejoined at end of data spooling.
422 - 7. Implement new Console commands to allow offlining/reserving drives,
423 and possibly manipulating the autochanger (much asked for).
424 - Add start/end date editing in messages (%t %T, %e?) ...
425 - Add ClientDefs similar to JobDefs.
426 - Print more info when bextract -p accepts a bad block.
427 - Fix FD JobType to be set before RunBeforeJob in FD.
428 - Look at adding full Volume and Pool information to a Volume
429 label so that bscan can get *all* the info.
430 - If the user puts "Purge Oldest Volume = yes" or "Recycle Oldest Volume = yes"
431 and there is only one volume in the pool, refuse to do it -- otherwise
432 he fills the Volume, then immediately starts reusing it.
433 - Implement copies and stripes.
434 - Add history file to console.
435 - Each file on tape creates a JobMedia record. Peter has 4 million
436 files spread over 10000 tape files and four tapes. A restore takes
437 16 hours to build the restore list.
438 - Add and option to check if the file size changed during backup.
439 - Make sure SD deletes spool files on error exit.
440 - Delete old spool files when SD starts.
441 - When labeling tapes, if you enter 000026, Bacula uses
442 the tape index rather than the Volume name 000026.
443 - Add offline tape command to Bacula console.
445 Enter MediaId or Volume name: 32
446 Enter new Volume name: DLT-20Dec04
447 Automatically selected Pool: Default
448 Connecting to Storage daemon DLTDrive at 192.168.68.104:9103 ...
449 Sending relabel command from "DLT-28Jun03" to "DLT-20Dec04" ...
450 block.c:552 Write error at 0:0 on device /dev/nst0. ERR=Bad file descriptor.
451 Error writing final EOF to tape. This tape may not be readable.
452 dev.c:1207 ioctl MTWEOF error on /dev/nst0. ERR=Permission denied.
453 askdir.c:219 NULL Volume name. This shouldn't happen!!!
454 3912 Failed to label Volume: ERR=dev.c:1207 ioctl MTWEOF error on /dev/nst0. ERR=Permission denied.
455 Label command failed for Volume DLT-20Dec04.
456 Do not forget to mount the drive!!!
457 - Bug: if a job is manually scheduled to run later, it does not appear
458 in any status report and cannot be cancelled.
460 ==== Keeping track of deleted files ====
461 My "trick" for keeping track of deletions is the following.
462 Assuming the user turns on this option, after all the files
463 have been backed up, but before the job has terminated, the
464 FD will make a pass through all the files and send their
465 names to the DIR (*exactly* the same as what a Verify job
466 currently does). This will probably be done at the same
467 time the files are being sent to the SD avoiding a second
468 pass. The DIR will then compare that to what is stored in
469 the catalog. Any files in the catalog but not in what the
470 FD sent will receive a catalog File entry that indicates
471 that at that point in time the file was deleted.
473 During a restore, any file initially picked up by some
474 backup (Full, ...) then subsequently having a File entry
475 marked "delete" will be removed from the tree, so will not
476 be restored. If a file with the same name is later OK it
477 will be inserted in the tree -- this already happens. All
478 will be consistent except for possible changes during the
481 Since I'm on the subject, some of you may be wondering what
482 the utility of the in memory tree is if you are going to
483 restore everything (at least it comes up from time to time
484 on the list). Well, it is still *very* useful because it
485 allows only the last item found for a particular filename
486 (full path) to be entered into the tree, and thus if a file
487 is backed up 10 times, only the last copy will be restored.
488 I recently (last Friday) restored a complete directory, and
489 the Full and all the Differential and Incremental backups
490 spanned 3 Volumes. The first Volume was not even mounted
491 because all the files had been updated and hence backed up
492 since the Full backup was made. In this case, the tree
493 saved me a *lot* of time.
495 Make sure this information is stored on the tape too so
496 that it can be restored directly from the tape.
498 Comments from Martin Simmons (I think they are all covered):
499 Ok, that should cover the basics. There are few issues though:
501 - Restore will depend on the catalog. I think it is better to include the
502 extra data in the backup as well, so it can be seen by bscan and bextract.
504 - I'm not sure if it will preserve multiple hard links to the same inode. Or
505 maybe adding or removing links will cause the data to be dumped again?
507 - I'm not sure if it will handle renamed directories. Possibly it will work
508 by dumping the whole tree under a renamed directory?
510 - It remains to be seen how the backup performance of the DIR's will be
511 affected when comparing the catalog for a large filesystem.
515 How about introducing a Type = MgmtPolicy job type? That job type would
516 be responsible for scanning the Bacula environment looking for specific
517 conditions, and submitting the appropriate jobs for implementing said
521 Name = "Migration-Policy"
523 Policy Selection Job Type = Migrate
524 Scope = "<keyword> <operator> <regexp>"
525 Threshold = "<keyword> <operator> <regexp>"
526 Job Template = <template-name>
529 Where <keyword> is any legal job keyword, <operator> is a comparison
530 operator (=,<,>,!=, logical operators AND/OR/NOT) and <regexp> is a
531 appropriate regexp. I could see an argument for Scope and Threshold
532 being SQL queries if we want to support full flexibility. The
533 Migration-Policy job would then get scheduled as frequently as a site
534 felt necessary (suggested default: every 15 minutes).
539 Name = "Migration-Policy"
541 Policy Selection Job Type = Migration
543 Threshold = "Migration Selection Type = LowestUtil"
544 Job Template = "MigrationTemplate"
547 would select all pools for examination and generate a job based on
548 MigrationTemplate to automatically select the volume with the lowest
549 usage and migrate it's contents to the nextpool defined for that pool.
551 This policy abstraction would be really handy for adjusting the behavior
552 of Bacula according to site-selectable criteria (one thing that pops
553 into mind is Amanda's ability to automatically adjust backup levels
554 depending on various criteria).
560 - Add Pool/Storage override regression test.
561 - Add delete JobId to regression.
562 - Add a regression test for dbcheck.
563 - New test to add bscan to four-concurrent-jobs regression,
564 i.e. after the four-concurrent jobs zap the
565 database as is done in the bscan-test, then use bscan to
566 restore the database, do a restore and compare with the
568 - Add restore of specific JobId to regression (item 3
569 on the restore prompt)
570 - Add IPv6 to regression
571 - Add database test to regression. Test each function like delete,
574 - AntiVir can slow down backups on Win32 systems.
575 - Win32 systems with FAT32 can be much slower than NTFS for
576 more than 1000 files per directory.
580 - A HOLD command to stop all jobs from starting.
581 - A PAUSE command to pause all running jobs ==> release the
583 - Media Type = LTO,LTO-2,LTO-3
584 Media Type Read = LTO,LTO2,LTO3
585 Media Type Write = LTO2, LTO3
587 === From Carsten Menke <bootsy52@gmx.net>
589 Following is a list of what I think in the situations where I'm faced with,
590 could be a usefull enhancement to bacula, which I'm certain other users will
591 benefit from as well.
593 1. NextJob/NextJobs Directive within a Job Resource in the form of
594 NextJobs = job1,job2.
597 I currently solved the problem with running multiple jobs each after each
598 by setting the Max Wait Time for a job to 8 hours, and give
599 the jobs different Priorities. However, there scenarios where
600 1 Job is directly depending on another job, so if the former job fails,
601 the job after it needn't to be run
602 while maybe other jobs should run despite of that
605 A Backup Job and a Verify job, if the backup job fails there is no need to run
606 the verify job, as the backup job already failed. However, one may like
607 to backup the Catalog to disk despite of that the main backup job failed.
610 I see that this is related to the Event Handlers which are on the ToDo
611 list, also it is maybe a good idea to check for the return value and
612 execute different actions based on the return value
615 3. offline capability to bconsole
618 Currently I use a script which I execute within the last Job via the
619 RunAfterJob Directive, to release and eject the tape.
620 So I have to call bconsole "release=Storage-Name" and afterwards
621 mt -f /dev/nst0 eject to get the tape out.
623 If I have multiple Storage Devices, than these may not be /dev/nst0 and
624 I have to modify the script or call it with parameters etc.
625 This would actually not be needed, as everything is already defined
626 in bacula-sd.conf and if I can invoke bconsole with the
627 storage name via $1 in the script than I'm done and information is
630 4. %s for Storage Name added to the chars being substituted in "RunAfterJob"
634 For the reason mentioned in 3. to have the ability to call a
635 script with /scripts/foobar %s and in the script use $1
636 to pass the Storage Name to bconsole
638 5. Setting Volume State within a Job Resource
641 Instead of using "Maximum Volume Jobs" in the Pool Resource,
642 I would have the possibilty to define
643 in a Job Resource that after this certain job is run, the Volume State
644 should be set to "Volume State = Used", this give more flexibility (IMHO).
646 6. Localization of Bacula Messages
649 Unfortunatley many,many people I work with don't speak english very well.
650 So if at least the Reporting messages would be localized then they
651 would understand that they have to change the tape,etc. etc.
653 I volunteer to do the german translations, and if I can convince my wife also
654 french and Morre (western african language).
656 7. OK, this is evil, probably bound to security risks and maybe not possible
657 due to the design of bacula.
659 Implementation of Backtics ( `command` ) for shell comand execution to
660 the "Label Format" Directive.
664 Currently I have defined BACULA_DAY_OF_WEEK="day1|day2..." resulting in
665 Label Format = "HolyBackup-${BACULA_DAY_OF_WEEK[${WeekDay}]}". If I could
666 use backticks than I could use "Label Format = HolyBackup-`date +%A` to have
667 the localized name for the day of the week appended to the
668 format string. Then I have the tape labeled automatically with weekday
669 name in the correct language.
671 - Yes, that is surely the case. I probably should turn those into Warning
672 errors. In addition, you just made me think that it might not be bad to
673 add an option to check the file size after backing up the file and
674 report if it changes. This would be done as an option because it would
677 Kern, good idea. If you do do that, mention in the output: file
678 shrunk, or file expanded, just to make it obvious to the user
679 (without having to the refer to file size), just how the file size
682 Would this option be for all file, or just one file? Or a fileset?
683 - Make output from status use html table tags for nicely
684 presenting in a browser.
685 - Can one write tapes faster with 8192 byte block sizes?
686 - Document security problems with the same password for everyone in
687 rpm and Win32 releases.
688 - Browse generations of files.
689 - I've seen an error when my catalog's File table fills up. I
690 then have to recreate the File table with a larger maximum row
691 size. Relevant information is at
692 http://dev.mysql.com/doc/mysql/en/Full_table.html ; I think the
693 "Installing and Configuring MySQL" chapter should talk a bit
694 about this potential problem, and recommend a solution.
695 - For Solaris must use POSIX awk.
696 - Want speed of writing to tape while despooling.
697 - Supported autochanger:
705 Wangtek 6525ES (SCSI-1 QIC drive, 525MB), under Linux 2.4.something,
706 bacula 1.36.0/1 works with blocksize 16k INSIDE bacula-sd.conf.
707 - Add regex from http://www.pcre.org to Bacula for Win32.
708 - Use only shell tools no make in CDROM package.
709 - Include within include does it work?
710 - Implement a Pool of type Cleaning?
711 - Implement VolReadTime and VolWriteTime in SD
712 - Modify Backing up Your Database to include a bootstrap file.
713 - Think about making certain database errors fatal.
714 - Look at correcting the time jump in the scheduler for daylight
715 savings time changes.
716 - Add a "real" timer to network connections.
717 - Promote to Full = Time period
718 - Check dates entered by user for correctness (month/day/... ranges)
719 - Compress restore Volume listing by date and first file.
720 - Look at patches/bacula_db.b2z postgresql that loops during restore.
722 - Perhaps add read/write programs and/or plugins to FileSets.
723 - How to handle backing up portables ...
724 - Add some sort of guaranteed Interval for upgrading jobs.
725 - Can we write the state file after every job terminates? On Win32
726 the system crashes and the state file is not updated.
729 Documentation to do: (any release a little bit at a time)
730 - Doc to do unmount before removing magazine.
731 - Alternative to static linking "ldd prog" save all binaries listed,
732 restore them and point LD_LIBRARY_PATH to them.
733 - Document add "</dev/null >/dev/null 2>&1" to the bacula-fd command line
734 - Document query file format.
735 - Add more documentation for bsr files.
736 - Document problems with Verify and pruning.
737 - Document how to use multiple databases.
738 - VXA drives have a "cleaning required"
739 indicator, but Exabyte recommends preventive cleaning after every 75
742 In this context, it should be noted that Exabyte has a command-line
743 vxatool utility available for free download. (The current version is
744 vxatool-3.72.) It can get diagnostic info, read, write and erase tapes,
745 test the drive, unload tapes, change drive settings, flash new firmware,
747 Of particular interest in this context is that vxatool <device> -i will
748 report, among other details, the time since last cleaning in tape motion
749 minutes. This information can be retrieved (and settings changed, for
750 that matter) through the generic-SCSI device even when Bacula has the
751 regular tape device locked. (Needless to say, I don't recommend
752 changing tape settings while a job is running.)
753 - Lookup HP cleaning recommendations.
754 - Lookup HP tape replacement recommendations (see trouble shooting autochanger)
755 - Document doing table repair
758 ===================================
759 - Add macro expansions in JobDefs.
760 Run Before Job = "SomeFile %{Level} %{Client}"
761 Write Bootstrap="/some/dir/%{JobName}_%{Client}.bsr"
762 - Use non-blocking network I/O but if no data is available, use
764 - Use gather write() for network I/O.
765 - Autorestart on crash.
766 - Add bandwidth limiting.
767 - Add acks every once and a while from the SD to keep
768 the line from timing out.
769 - When an error in input occurs and conio beeps, you can back
770 up through the prompt.
771 - Detect fixed tape block mode during positioning by looking at
772 block numbers in btape "test". Possibly adjust in Bacula.
773 - Fix list volumes to output volume retention in some other
774 units, perhaps via a directive.
775 - If opening a tape in read/write mode fails attempt to open
776 it in read-only mode, and mark the tape for read only.
777 - Allow Simultaneous Priorities = yes => run up to Max concurrent jobs even
778 with multiple priorities.
779 - If you use restore replace=never, the directory attributes for
780 non-existent directories will not be restored properly.
782 - see lzma401.zip in others directory for new compression
784 - Minimal autochanger handling in Bacula and in btape.
785 - Look into how tar does not save sockets and the possiblity of
786 not saving them in Bacula (Martin Simmons reported this).
787 - Fix restore jobs so that multiple jobs can run if they
788 are not using the same tape(s).
789 - Allow the user to select JobType for manual pruning/purging.
790 - bscan does not put first of two volumes back with all info in
792 - Implement the FreeBSD nodump flag in chflags.
793 - Figure out how to make named console messages go only to that
794 console and to the non-restricted console (new console class?).
795 - Make restricted console prompt for password if *ask* is set or
796 perhaps if password is undefined.
797 - Implement "from ISO-date/time every x hours/days/weeks/months" in
800 ==== from Marc Schoechlin
801 - the help-command should be more verbose
802 (it should explain the paramters of the different
804 -> its time-comsuming to consult the manual anytime
805 you need a special parameter
806 -> maybe its more easy to maintain this, if the
807 descriptions of that commands are outsourced to
809 - the cd-command should allow complete paths
810 i.e. cd /foo/bar/foo/bar
811 -> if a customer mails me the path to a certain file,
812 its faster to enter the specified directory
813 - if the password is not configured in bconsole.conf
814 you should be asked for it.
815 -> sometimes you like to do restore on a customer-machine
816 which shouldnt know the password for bacula.
817 -> adding the password to the file favours admins
818 to forget to remove the password after usage
820 the protection of that file is less important
821 - long-listed-output of commands should be scrollable
822 like the unix more/less-command does
823 -> if someone runs 200 and more machines, the lists could
824 be a little long and complex
825 - command-output should be shown column by column
826 to reduce scrolling and to increase clarity
828 - lsmark should list the selected files with full
830 - wildcards for selecting and file and directories would be nice
831 - any actions should be interuptable with STRG+C
832 - command-expansion would be pretty cool
834 - When the replace Never option is set, new directory permissions
835 are not restored. See bug 213. To fix this requires creating a
836 list of newly restored directories so that those directory
837 permissions *can* be restored.
838 - Compaction of Disk space by "migrating" Volumes that have pruned
839 Jobs (what criteria? size, #jobs, time).
840 - Add prune all command
841 - Document fact that purge can destroy a part of a restore by purging
842 one volume while others remain valid -- perhaps mark Jobs.
843 - Add multiple-media-types.txt
844 - look at mxt-changer.html
845 - Make ? do a help command (no return needed).
846 - Implement restore directory.
847 - Document streams and how to implement them.
848 - Try not to re-backup a file if a new hard link is added.
849 - Add feature to backup hard links only, but not the data.
850 - Fix stream handling to be simpler.
851 - Add Priority and Bootstrap to Run a Job.
852 - Eliminate Restore "Run Restore Job" prompt by allowing new "run command
854 - Remove View FileSet button from Run a Job dialog.
855 - Handle prompt for restore job at end of Restore command.
856 - Add display of total selected files to Restore window.
857 - Add tree pane to left of window.
858 - Add progress meter.
859 - Max wait time or max run time causes seg fault -- see runtime-bug.txt
860 - Document writing to a CD/DVD with Bacula.
861 - Add a "base" package to the window installer for pthreadsVCE.dll
862 which is needed by all packages.
863 - Add message to user to check for fixed block size when the forward
864 space test fails in btape.
865 - When unmarking a directory check if all files below are unmarked and
866 then remove the + flag -- in the restore tree.
867 - Possibly implement: Action = Unmount Device="TapeDrive1" in Admin jobs.
868 - Setup lrrd graphs: (http://www.linpro.no/projects/lrrd/) Mike Acar.
869 - Revisit the question of multiple Volumes (disk) on a single device.
870 - Add a block copy option to bcopy.
871 - Investigate adding Mac Resource Forks.
872 - Finish work on Gnome restore GUI.
873 - Fix "llist jobid=xx" where no fileset or client exists.
874 - For each job type (Admin, Restore, ...) require only the really necessary
875 fields.- Pass Director resource name as an option to the Console.
876 - Add a "batch" mode to the Console (no unsolicited queries, ...).
877 - Add a .list all files in the restore tree (probably also a list all files)
878 Do both a long and short form.
879 - Allow browsing the catalog to see all versions of a file (with
880 stat data on each file).
881 - Restore attributes of directory if replace=never set but directory
883 - Use SHA1 on authentication if possible.
884 - See comtest-xxx.zip for Windows code to talk to USB.
885 - Add John's appended files:
886 Appended = { /files/server/logs/http/*log }
887 and such files would be treated as follows.On a FULL backup, they would
888 be backed up like any other file.On an INCREMENTAL backup, where a
889 previous INCREMENTAL or FULL was already in thecatalogue and the length
890 of the file wasgreater than the length of the last backup, only thedata
891 added since the last backup will be dumped.On an INCREMENTAL backup, if
892 the length of the file is less than thelength of the file with the same
893 name last backed up, the completefile is dumped.On Windows systems, with
894 creation date of files, we can be evensmarter about this and not count
895 entirely upon the length.On a restore, the full and all incrementals
896 since it will beapplied in sequence to restore the file.
897 - Check new HAVE_WIN32 open bits.
898 - Check if the tape has moved before writing.
899 - Handling removable disks -- see below:
900 - Keep track of tape use time, and report when cleaning is necessary.
901 - Add FromClient and ToClient keywords on restore command (or
902 BackupClient RestoreClient).
903 - Implement a JobSet, which groups any number of jobs. If the
904 JobSet is started, all the jobs are started together.
905 Allow Pool, Level, and Schedule overrides.
906 - Enhance cancel to timeout BSOCK packets after a specific delay.
907 - Do scheduling by UTC using gmtime_r() in run_conf, scheduler, and
908 ua_status.!!! Thanks to Alan Brown for this tip.
909 - Look at updating Volume Jobs so that Max Volume Jobs = 1 will work
910 correctly for multiple simultaneous jobs.
911 - Correct code so that FileSet MD5 is calculated for < and | filename
913 - Implement the Media record flag that indicates that the Volume does disk
915 - Implement VolAddr, which is used when Volume is addressed like a disk,
916 and form it from VolFile and VolBlock.
917 - Make multiple restore jobs for multiple media types specifying
918 the proper storage type.
919 - Fix fast block rejection (stored/read_record.c:118). It passes a null
920 pointer (rec) to try_repositioning().
921 - Look at extracting Win data from BackupRead.
922 - Implement RestoreJobRetention? Maybe better "JobRetention" in a Job,
923 which would take precidence over the Catalog "JobRetention".
924 - Implement Label Format in Add and Label console commands.
925 - Possibly up network buffers to 65K. Put on variable.
926 - Put email tape request delays on one or more variables. User wants
927 to cancel the job after a certain time interval. Maximum Mount Wait?
928 - Job, Client, Device, Pool, or Volume?
929 Is it possible to make this a directive which is *optional* in multiple
930 resources, like Level? If so, I think I'd make it an optional directive
931 in Job, Client, and Pool, with precedence such that Job overrides Client
932 which in turn overrides Pool.
934 - New Storage specifications:
935 - Want to write to multiple storage devices simultaneously
936 - Want to write to multiple storage devices sequentially (in one job)
937 - Want to read/write simultaneously
938 - Key is MediaType -- it must match
940 Passed to SD as a sort of BSR record called Storage Specification
944 MediaType -> Next MediaType
946 Device -> Next Device
948 Allow multiple Storage specifications
956 Allow Multiple Pool specifications (note, Pool currently
958 Allow Multiple MediaType specifications in Dir conf
959 Allow Multiple Device specifications in Dir conf
960 Perhaps keep this in a single SSR
961 Tie a Volume to a specific device by using a MediaType that
962 is contained in only one device.
963 In SD allow Device to have Multiple MediaTypes
965 - Ideas from Jerry Scharf:
966 First let's point out some big pluses that bacula has for this
968 more importantly it's active. Thank you so much for that
969 even more important, it's not flaky
970 it has an open access catalog, opening many possibilities
971 it's pushing toward heterogeneous systems capability
973 Macintosh file client
974 macs are an interesting niche, but I fear a server is a rathole
975 working bare iron recovery for windows
976 the option for inc/diff backups not reset on fileset revision
977 a) use both change and inode update time against base time
978 b) do the full catalog check (expensive but accurate)
979 sizing guide (how much system is needed to back up N systems/files)
980 consultants on using bacula in building a disaster recovery system
982 or how to get at fancy things that one could do with bacula
983 logwatch code for bacula logs (or similar)
984 linux distro inclusion of bacula (brings good and bad, but necessary)
985 win2k/XP server capability (icky but you asked)
986 support for Oracle database ??
988 - Look at adding SQL server and Exchange support for Windows.
989 - Make dev->file and dev->block_num signed integers so that -1 can
990 be an invalid value which happens with BSR.
991 - Create VolAddr for disk files in place of VolFile and VolBlock. This
992 is needed to properly specify ranges.
993 - Add progress of files/bytes to SD and FD.
994 - Print warning message if FileId > 4 billion
995 - do a "messages" before the first prompt in Console
996 - Client does not show busy during Estimate command.
997 - Implement Console mtx commands.
998 - Implement a Mount Command and an Unmount Command where
999 the users could specify a system command to be performed
1000 to do the mount, after which Bacula could attempt to
1001 read the device. This is for Removeable media such as a CDROM.
1002 - Most likely, this mount command would be invoked explicitly
1003 by the user using the current Console "mount" and "unmount"
1004 commands -- the Storage Daemon would do the right thing
1005 depending on the exact nature of the device.
1006 - As with tape drives, when Bacula wanted a new removable
1007 disk mounted, it would unmount the old one, and send a message
1008 to the user, who would then use "mount" as described above
1009 once he had actually inserted the disk.
1010 - Implement dump/print label to UA
1011 - Spool to disk only when the tape is full, then when a tape is hung move
1013 - bextract is sending everything to the log file ****FIXME****
1014 - Allow multiple Storage specifications (or multiple names on
1015 a single Storage specification) in the Job record. Thus a job
1016 can be backed up to a number of storage devices.
1017 - Implement some way for the File daemon to contact the Director
1018 to start a job or pass its DHCP obtained IP number.
1019 - Implement a query tape prompt/replace feature for a console
1020 - Copy console @ code to gnome2-console
1021 - Make AES the only encryption algorithm see
1022 http://csrc.nist.gov/CryptoToolkit/aes/). It's
1023 an officially adopted standard, has survived peer
1024 review, and provides keys up to 256 bits.
1025 - Take a careful look at SetACL http://setacl.sourceforge.net
1026 - Make tree walk routines like cd, ls, ... more user friendly
1027 by handling spaces better.
1028 - Make sure that Bacula rechecks the tape after the 20 min wait.
1029 - Set IO_NOWAIT on Bacula TCP/IP packets.
1030 - Try doing a raw partition backup and restore by mounting a
1032 - From Lars Kellers:
1033 Yes, it would allow to highly automatic the request for new tapes. If a
1034 tape is empty, bacula reads the barcodes (native or simulated), and if
1035 an unused tape is found, it runs the label command with all the
1036 necessary parameters.
1038 By the way can bacula automatically "move" an empty/purged volume say
1039 in the "short" pool to the "long" pool if this pool runs out of volume
1041 - What to do about "list files job=xxx".
1042 - Get and test MySQL 4.0
1043 - Look at how fuser works and /proc/PID/fd that is how Nic found the
1044 file descriptor leak in Bacula.
1045 - Implement WrapCounters in Counters.
1046 - Add heartbeat from FD to SD if hb interval expires.
1047 - Can we dynamically change FileSets?
1048 - If pool specified to label command and Label Format is specified,
1049 automatically generate the Volume name.
1050 - Why can't SQL do the filename sort for restore?
1051 - Add ExhautiveRestoreSearch
1052 - Look at the possibility of loading only the necessary
1053 data into the restore tree (i.e. do it one directory at a
1054 time as the user walks through the tree).
1055 - Possibly use the hash code if the user selects all for a restore command.
1056 - Fix "restore all" to bypass building the tree.
1057 - Prohibit backing up archive device (findlib/find_one.c:128)
1058 - Implement Release Device in the Job resource to unmount a drive.
1059 - Implement Acquire Device in the Job resource to mount a drive,
1060 be sure this works with admin jobs so that the user can get
1061 prompted to insert the correct tape. Possibly some way to say to
1062 run the job but don't save the files.
1063 - Make things like list where a file is saved case independent for
1066 - Use autochanger to handle multiple devices.
1067 - On Windows with very long path names, it may be impossible to create
1068 a file (and thus restore it) because the total length is too long.
1069 We must cd into the directory then create the file without the
1071 - Implement a Recycle command
1072 - Test a second language e.g. french.
1073 - Start working on Base jobs.
1074 - Implement UnsavedFiles DB record.
1075 - From Phil Stracchino:
1076 It would probably be a per-client option, and would be called
1077 something like, say, "Automatically purge obsoleted jobs". What it
1078 would do is, when you successfully complete a Differential backup of a
1079 client, it would automatically purge all Incremental backups for that
1080 client that are rendered redundant by that Differential. Likewise,
1081 when a Full backup on a client completed, it would automatically purge
1082 all Differential and Incremental jobs obsoleted by that Full backup.
1083 This would let people minimize the number of tapes they're keeping on
1084 hand without having to master the art of retention times.
1085 - When doing a Backup send all attributes back to the Director, who
1086 would then figure out what files have been deleted.
1087 - Currently in mount.c:236 the SD simply creates a Volume. It should have
1088 explicit permission to do so. It should also mark the tape in error
1089 if there is an error.
1090 - Cancel waiting for Client connect in SD if FD goes away.
1092 - Implement timeout in response() when it should come quickly.
1093 - Implement a Slot priority (loaded/not loaded).
1094 - Implement "vacation" Incremental only saves.
1095 - Implement create "FileSet"?
1096 - Add prefixlinks to where or not where absolute links to FD.
1097 - Issue message to mount a new tape before the rewind.
1098 - Simplified client job initiation for portables.
1099 - If SD cannot open a drive, make it periodically retry.
1100 - Add more of the config info to the tape label.
1102 - If tape is marked read-only, then try opening it read-only rather than
1103 failing, and remember that it cannot be written.
1104 - Refine SD waiting output:
1105 Device is being positioned
1106 > Device is being positioned for append
1107 > Device is being positioned to file x
1109 - Figure out some way to estimate output size and to avoid splitting
1110 a backup across two Volumes -- this could be useful for writing CDROMs
1111 where you really prefer not to have it split -- not serious.
1112 - Have SD compute MD5 or SHA1 and compare to what FD computes.
1113 - Make VolumeToCatalog calculate an MD5 or SHA1 from the
1114 actual data on the Volume and compare it.
1115 - Implement Bacula plugins -- design API
1116 - Make bcopy read through bad tape records.
1117 - Program files (i.e. execute a program to read/write files).
1118 Pass read date of last backup, size of file last time.
1119 - Add Signature type to File DB record.
1120 - CD into subdirectory when open()ing files for backup to
1121 speed up things. Test with testfind().
1122 - Priority job to go to top of list.
1123 - Why are save/restore of device different sizes (sparse?) Yup! Fix it.
1124 - Implement some way for the Console to dynamically create a job.
1125 - Solaris -I on tar for include list
1126 - Need a verbose mode in restore, perhaps to bsr.
1127 - bscan without -v is too quiet -- perhaps show jobs.
1128 - Add code to reject whole blocks if not wanted on restore.
1129 - Check if we can increase Bacula FD priorty in Win2000
1130 - Make sure the MaxVolFiles is fully implemented in SD
1131 - Check if both CatalogFiles and UseCatalog are set to SD.
1132 - Possibly add email to Watchdog if drive is unmounted too
1133 long and a job is waiting on the drive.
1134 - After unmount, if restore job started, ask to mount.
1135 - Add UA rc and history files.
1136 - put termcap (used by console) in ./configure and
1137 allow -with-termcap-dir.
1138 - Fix Autoprune for Volumes to respect need for full save.
1139 - Compare tape to Client files (attributes, or attributes and data)
1140 - Make all database Ids 64 bit.
1141 - Allow console commands to detach or run in background.
1142 - Fix status delay on storage daemon during rewind.
1143 - Add SD message variables to control operator wait time
1144 - Maximum Operator Wait
1145 - Minimum Message Interval
1146 - Maximum Message Interval
1147 - Send Operator message when cannot read tape label.
1148 - Verify level=Volume (scan only), level=Data (compare of data to file).
1149 Verify level=Catalog, level=InitCatalog
1151 - Add keyword search to show command in Console.
1152 - Events : tape has more than xxx bytes.
1153 - Complete code in Bacula Resources -- this will permit
1154 reading a new config file at any time.
1155 - Handle ctl-c in Console
1156 - Implement script driven addition of File daemon to config files.
1157 - Think about how to make Bacula work better with File (non-tape) archives.
1158 - Write Unix emulator for Windows.
1159 - Put memory utilization in Status output of each daemon
1160 if full status requested or if some level of debug on.
1161 - Make database type selectable by .conf files i.e. at runtime
1162 - Set flag for uname -a. Add to Volume label.
1163 - Restore files modified after date
1164 - SET LD_RUN_PATH=$HOME/mysql/lib/mysql
1165 - Remove duplicate fields from jcr (e.g. jcr.level and jcr.jr.Level, ...).
1166 - Timout a job or terminate if link goes down, or reopen link and query.
1167 - Concept of precious tapes (cannot be reused).
1168 - Make bcopy copy with a single tape drive.
1169 - Permit changing ownership during restore.
1172 > My suggestion: Add a feature on the systray menu-icon menu to request
1173 > an immediate backup now. This would be useful for laptop users who may
1174 > not be on the network when the regular scheduled backup is run.
1176 > My wife's suggestion: Add a setting to the win32 client to allow it to
1177 > shut down the machine after backup is complete (after, of course,
1178 > displaying a "System will shut down in one minute, click here to cancel"
1179 > warning dialog). This would be useful for sites that want user
1180 > woorkstations to be shut down overnight to save power.
1183 - Autolabel should be specified by DIR instead of SD.
1185 - Add media capacity
1186 - AutoScan (check checksum of tape)
1187 - Format command = "format /dev/nst0"
1191 - Seek resolution (usually corresponds to buffer size)
1192 - EODErrorCode=ENOSPC or code
1193 - Partial Read error code
1194 - Partial write error code
1195 - Nonformatted read error
1196 - Nonformatted write error
1197 - WriteProtected error
1201 - IgnoreCloseErrors=yes
1211 - FD sends unsaved file list to Director at end of job (see
1213 - File daemon should build list of files skipped, and then
1214 at end of save retry and report any errors.
1215 - Write a Storage daemon that uses pipes and
1216 standard Unix programs to write to the tape.
1218 - Need something that monitors the JCR queue and
1219 times out jobs by asking the deamons where they are.
1220 - Enhance Jmsg code to permit buffering and saving to disk.
1221 - device driver = "xxxx" for drives.
1222 - Verify from Volume
1223 - Ensure that /dev/null works
1224 - Need report class for messages. Perhaps
1225 report resource where report=group of messages
1226 - enhance scan_attrib and rename scan_jobtype, and
1227 fill in code for "since" option
1228 - Director needs a time after which the report status is sent
1229 anyway -- or better yet, a retry time for the job.
1230 - Don't reschedule a job if previous incarnation is still running.
1231 - Some way to automatically backup everything is needed????
1232 - Need a structure for pending actions:
1234 - termination status (part of buffered msgs?)
1236 Read, Write, Clean, Delete
1237 - Login to Bacula; Bacula users with different permissions:
1238 owner, group, user, quotas
1239 - Store info on each file system type (probably in the job header on tape.
1240 This could be the output of df; or perhaps some sort of /etc/mtab record.
1242 ========= ideas ===============
1243 From: "Jerry K. Schieffer" <jerry@skylinetechnology.com>
1244 To: <kern@sibbald.com>
1245 Subject: RE: [Bacula-users] future large programming jobs
1246 Date: Thu, 26 Feb 2004 11:34:54 -0600
1248 I noticed the subject thread and thought I would offer the following
1249 merely as sources of ideas, i.e. something to think about, not even as
1250 strong as a request. In my former life (before retiring) I often
1251 dealt with backups and storage management issues/products as a
1252 developer and as a consultant. I am currently migrating my personal
1253 network from amanda to bacula specifically because of the ability to
1254 cross media boundaries during storing backups.
1255 Are you familiar with the commercial product called ADSM (I think IBM
1256 now sells it under the Tivoli label)? It has a couple of interesting
1257 ideas that may apply to the following topics.
1259 1. Migration: Consider that when you need to restore a system, there
1260 may be pressure to hurry. If all the information for a single client
1261 can eventually end up on the same media (and in chronological order),
1262 the restore is facillitated by not having to search past information
1263 from other clients. ADSM has the concept of "client affinity" that
1264 may be associated with it's storage pools. It seems to me that this
1265 concept (as an optional feature) might fit in your architecture for
1268 ADSM also has the concept of defining one or more storage pools as
1269 "copy pools" (almost mirrors, but only in the sense of contents).
1270 These pools provide the ability to have duplicte data stored both
1271 onsite and offsite. The copy process can be scheduled to be handled
1272 by their storage manager during periods when there is no backup
1273 activity. Again, the migration process might be a place to consider
1274 implementing something like this.
1277 > It strikes me that it would be very nice to be able to do things
1279 > have the Job(s) backing up the machines run, and once they have all
1280 > completed, start a migration job to copy the data from disks Volumes
1282 > a tape library and then to offsite storage. Maybe this can already
1284 > done with some careful scheduling and Job prioritzation; the events
1285 > mechanism described below would probably make it very easy.
1287 This is the goal. In the first step (before events), you simply
1289 the Migration to tape later.
1291 2. Base jobs: In ADSM, each copy of each stored file is tracked in
1292 the database. Once a file (unique by path and metadata such as dates,
1293 size, ownership, etc.) is in a copy pool, no more copies are made. In
1294 other words, when you start ADSM, it begins like your concept of a
1295 base job. After that it is in the "incremental" mode. You can
1296 configure the number of "generations" of files to be retained, plus a
1297 retention date after which even old generations are purged. The
1298 database tracks the contents of media and projects the percentage of
1299 each volume that is valid. When the valid content of a volume drops
1300 below a configured percentage, the valid data are migrated to another
1301 volume and the old volume is marked as empty. Note, this requires
1302 ADSM to have an idea of the contents of a client, i.e. marking the
1303 database when an existing file was deleted, but this would solve your
1304 issue of restoring a client without restoring deleted files.
1306 This is pretty far from what bacula now does, but if you are going to
1307 rip things up for Base jobs,.....
1308 Also, the benefits of this are huge for very large shops, especially
1309 with media robots, but are a pain for shops with manual media
1313 > Base jobs sound pretty useful, but I'm not dying for them.
1315 Nobody is dying for them, but when you see what it does, you will die
1318 3. Restoring deleted files: Since I think my comments in (2) above
1319 have low probability of implementation, I'll also suggest that you
1320 could approach the issue of deleted files by a mechanism of having the
1321 fd report to the dir, a list of all files on the client for every
1322 backup job. The dir could note in the database entry for each file
1323 the date that the file was seen. Then if a restore as of date X takes
1324 place, only files that exist from before X until after X would be
1325 restored. Probably the major cost here is the extra date container in
1326 each row of the files table.
1328 Thanks for "listening". I hope some of this helps. If you want to
1329 contact me, please send me an email - I read some but not all of the
1330 mailing list traffic and might miss a reply there.
1332 Please accept my compliments for bacula. It is doing a great job for
1333 me!! I sympathize with you in the need to wrestle with excelence in
1334 execution vs. excelence in feature inclusion.
1339 ==============================
1342 - Design at hierarchial storage for Bacula. Migration and Clone.
1343 - Implement FSM (File System Modules).
1344 - Audit M_ error codes to ensure they are correct and consistent.
1345 - Add variable break characters to lex analyzer.
1346 Either a bit mask or a string of chars so that
1347 the caller can change the break characters.
1348 - Make a single T_BREAK to replace T_COMMA, etc.
1349 - Ensure that File daemon and Storage daemon can
1350 continue a save if the Director goes down (this
1351 is NOT currently the case). Must detect socket error,
1352 buffer messages for later.
1353 - Enhance time/duration input to allow multiple qualifiers e.g. 3d2h
1354 - Add ability to backup to two Storage devices (two SD sessions) at
1355 the same time -- e.g. onsite, offsite.
1356 - Add the ability to consolidate old backup sets (basically do a restore
1357 to tape and appropriately update the catalog). Compress Volume sets.
1358 Might need to spool via file is only one drive is available.
1359 - Compress or consolidate Volumes of old possibly deleted files. Perhaps
1360 someway to do so with every volume that has less than x% valid
1364 Migration: Move a backup from one Volume to another
1365 Clone: Copy a backup -- two Volumes
1367 Bacula Migration is based on Jobs (apparently Networker is file by file).
1369 Migration triggered by:
1373 Highwater mark (keep total size)
1378 ======================================================
1380 It is somewhat like a Full save becomes an incremental since
1381 the Base job (or jobs) plus other non-base files.
1383 - A Base backup is same as Full backup, just different type.
1384 - New BaseFiles table that contains:
1386 BaseJobId - Base JobId referenced for this FileId (needed ???)
1387 JobId - JobId currently running
1388 FileId - File not backed up, exists in Base Job
1389 FileIndex - FileIndex from Base Job.
1390 i.e. for each base file that exists but is not saved because
1391 it has not changed, the File daemon sends the JobId, BaseId,
1392 FileId, FileIndex back to the Director who creates the DB entry.
1393 - To initiate a Base save, the Director sends the FD
1394 the FileId, and full filename for each file in the Base.
1395 - When the FD finds a Base file, he requests the Director to
1396 send him the full File entry (stat packet plus MD5), or
1397 conversely, the FD sends it to the Director and the Director
1398 says yes or no. This can be quite rapid if the FileId is kept
1399 by the FD for each Base Filename.
1400 - It is probably better to have the comparison done by the FD
1401 despite the fact that the File entry must be sent across the
1403 - An alternative would be to send the FD the whole File entry
1404 from the start. The disadvantage is that it requires a lot of
1405 space. The advantage is that it requires less communications
1407 - The Job record must be updated to indicate that one or more
1409 - At end of Job, FD returns:
1410 1. Count of base files/bytes not written to tape (i.e. matches)
1411 2. Count of base file that were saved i.e. had changed.
1412 - No tape record would be written for a Base file that matches, in the
1413 same way that no tape record is written for Incremental jobs where
1414 the file is not saved because it is unchanged.
1415 - On a restore, all the Base file records must explicitly be
1416 found from the BaseFile tape. I.e. for each Full save that is marked
1417 to have one or more Base Jobs, search the BaseFile for all occurrences
1419 - An optimization might be to make the BaseFile have:
1425 This would avoid the need to explicitly fetch each File record for
1426 the Base job. The Base Job record will be fetched to get the
1427 VolSessionId and VolSessionTime.
1428 =========================================================
1431 ==========================================================
1433 For each Incremental job that is run, there may be files that
1434 were found but not saved because they were locked (this applies
1435 only to Windows). Such a system could send back to the Director
1436 a list of Unsaved files.
1438 - New UnSavedFiles table that contains:
1442 - Then in the next Incremental job, the list of Unsaved Files will be
1443 feed to the FD, who will ensure that they are explicitly chosen even
1444 if standard date/time check would not have selected them.
1445 =============================================================
1449 Multiple drive autochanger data: see Alan Brown
1450 mtx -f xxx unloadStorage Element 1 is Already Full(drive 0 was empty)
1451 Unloading Data Transfer Element into Storage Element 1...source Element
1452 Address 480 is Empty
1454 (drive 0 was empty and so was slot 1)
1455 > mtx -f xxx load 15 0
1456 no response, just returns to the command prompt when complete.
1457 > mtx -f xxx status Storage Changer /dev/changer:2 Drives, 60 Slots ( 2 Import/Export )
1458 Data Transfer Element 0:Full (Storage Element 15 Loaded):VolumeTag = HX001
1459 Data Transfer Element 1:Empty
1460 Storage Element 1:Empty
1461 Storage Element 2:Full :VolumeTag=HX002
1462 Storage Element 3:Full :VolumeTag=HX003
1463 Storage Element 4:Full :VolumeTag=HX004
1464 Storage Element 5:Full :VolumeTag=HX005
1465 Storage Element 6:Full :VolumeTag=HX006
1466 Storage Element 7:Full :VolumeTag=HX007
1467 Storage Element 8:Full :VolumeTag=HX008
1468 Storage Element 9:Full :VolumeTag=HX009
1469 Storage Element 10:Full :VolumeTag=HX010
1470 Storage Element 11:Empty
1471 Storage Element 12:Empty
1472 Storage Element 13:Empty
1473 Storage Element 14:Empty
1474 Storage Element 15:Empty
1475 Storage Element 16:Empty....
1476 Storage Element 28:Empty
1477 Storage Element 29:Full :VolumeTag=CLNU01L1
1478 Storage Element 30:Empty....
1479 Storage Element 57:Empty
1480 Storage Element 58:Full :VolumeTag=NEX261L2
1481 Storage Element 59 IMPORT/EXPORT:Empty
1482 Storage Element 60 IMPORT/EXPORT:Empty
1484 Unloading Data Transfer Element into Storage Element 15...done
1486 (just to verify it remembers where it came from, however it can be
1487 overrriden with mtx unload {slotnumber} to go to any storage slot.)
1489 There needs to be a table of drive # to devices somewhere - If there are
1490 multiple changers or drives there may not be a 1:1 correspondance between
1491 changer drive number and system device name - and depending on the way the
1492 drives are hooked up to scsi busses, they may not be linearly numbered
1493 from an offset point either.something like
1495 Autochanger drives = 2
1496 Autochanger drive 0 = /dev/nst1
1497 Autochanger drive 1 = /dev/nst2
1498 IMHO, it would be _safest_ to use explicit mtx unload commands at all
1499 times, not just for multidrive changers. For a 1 drive changer, that's
1505 MTX's manpage (1.2.15):
1506 unload [<slotnum>] [ <drivenum> ]
1507 Unloads media from drive <drivenum> into slot
1508 <slotnum>. If <drivenum> is omitted, defaults to
1509 drive 0 (as do all commands). If <slotnum> is
1510 omitted, defaults to the slot that the drive was
1511 loaded from. Note that there's currently no way
1512 to say 'unload drive 1's media to the slot it
1513 came from', other than to explicitly use that
1514 slot number as the destination.AB
1520 undef# camcontrol devlist
1521 <WANGTEK 51000 SCSI M74H 12B3> at scbus0 target 2 lun 0 (pass0,sa0)
1522 <ARCHIVE 4586XX 28887-XXX 4BGD> at scbus0 target 4 lun 0 (pass1,sa1)
1523 <ARCHIVE 4586XX 28887-XXX 4BGD> at scbus0 target 4 lun 1 (pass2)
1525 tapeinfo -f /dev/sg0 with a bad tape in drive 1:
1526 [kern@rufus mtx-1.2.17kes]$ ./tapeinfo -f /dev/sg0
1527 Product Type: Tape Drive
1529 Product ID: 'C5713A '
1531 Attached Changer: No
1532 TapeAlert[3]: Hard Error: Uncorrectable read/write error.
1533 TapeAlert[20]: Clean Now: The tape drive neads cleaning NOW.
1540 Medium Type: Not Loaded
1543 DataCompEnabled: yes
1544 DataCompCapable: yes
1545 DataDeCompEnabled: yes
1552 Handling removable disks
1554 From: Karl Cunningham <karlc@keckec.com>
1556 My backups are only to hard disk these days, in removable bays. This is my
1557 idea of how a backup to hard disk would work more smoothly. Some of these
1558 things Bacula does already, but I mention them for completeness. If others
1559 have better ways to do this, I'd like to hear about it.
1561 1. Accommodate several disks, rotated similar to how tapes are. Identified
1562 by partition volume ID or perhaps by the name of a subdirectory.
1563 2. Abort & notify the admin if the wrong disk is in the bay.
1564 3. Write backups to different subdirectories for each machine to be backed
1566 4. Volumes (files) get created as needed in the proper subdirectory, one
1568 5. When a disk is recycled, remove or zero all old backup files. This is
1569 important as the disk being recycled may be close to full. This may be
1570 better done manually since the backup files for many machines may be
1571 scattered in many subdirectories.
1576 - Make sure that all do_prompt() calls in Dir check for
1577 -1 (error) and -2 (cancel) returns.
1578 - Fix foreach_jcr() to have free_jcr() inside next().
1579 jcr=jcr_walk_start();
1580 for ( ; jcr; (jcr=jcr_walk_next(jcr)) )
1583 - A Volume taken from Scratch should take on the retention period
1585 - Correct doc for Maximum Changer Wait (and others) accepting only
1587 - Implement status that shows why a job is being held in reserve, or
1588 rather why none of the drives are suitable.
1589 - Implement a way to disable a drive (so you can use the second
1590 drive of an autochanger, and the first one will not be used or
1592 - Make sure Maximum Volumes is respected in Pools when adding
1593 Volumes (e.g. when pulling a Scratch volume).
1594 - Keep same dcr when switching device ...
1595 - Implement code that makes the Dir aware that a drive is an
1596 autochanger (so the user doesn't need to use the Autochanger = yes
1598 - Make catalog respect ACL.
1599 - Add recycle count to Media record.
1600 - Add initial write date to Media record.
1601 - Fix store_yesno to be store_bitmask.
1602 --- create_file.c.orig Fri Jul 8 12:13:05 2005
1603 +++ create_file.c Fri Jul 8 12:13:07 2005
1605 attr->ofname, be.strerror());
1608 + } else if(S_ISSOCK(attr->statp.st_mode)) {
1609 + Dmsg1(200, "Skipping socket: %s\n", attr->ofname);
1611 Dmsg1(200, "Restore node: %s\n", attr->ofname);
1612 if (mknod(attr->ofname, attr->statp.st_mode, attr->statp.st_rdev) != 0 && errno != EEXIST) {
1613 - Add true/false to conf same as yes/no
1614 - Reserve blocks other restore jobs when first cannot connect to SD.
1615 - Fix Maximum Changer Wait, Maximum Open Wait, Maximum Rewind Wait to
1616 accept time qualifiers.
1617 - Does ClientRunAfterJob fail the job on a bad return code?
1618 - Make hardlink code at line 240 of find_one.c use binary search.
1619 - Add ACL error messages in src/filed/acl.c.
1620 - Make authentication failures single threaded.
1621 - Make Dir and SD authentication errors single threaded.
1623 - Fix catreq.c digestbuf at line 411 in src/dird/catreq.c
1624 - Make base64.c (bin_to_base64) take a buffer length
1625 argument to avoid overruns.
1626 and verify that other buffers cannot overrun.
1627 - Implement VolumeState as discussed with Arno.
1628 - Add LocationId to update volume
1636 - Add Comment to Media record
1637 - Fix auth compatibility with 1.38
1638 - Update dbcheck to include Log table
1639 - Update llist to include new fields.