1 Technical notes on version 1.32 28Sep03
7 Changes submitted this submission:
9 - Enhance manual faq, regression ...
10 - Make FreeBSD read sizes always be a multiple of 512. Needed
12 - Make FreeBSD accept block AND character devices for raw
13 disk reads. On FreeBSD there are no block devices.
15 - Fix printing of EndTime on job report after rescheduling.
16 - Fix jobq.c error returns to clean up before returning.
17 - Make Cleaning tape aware of "unlabeled" volumes.
18 - If a job is rescheduled, ensure that old job is
19 removed from SD's jcr queue -- i.e. it is still waiting
21 - Test rescheduling code.
22 - Change ./configure to detect Cygwin enviornments.
24 - Release beta version.
26 - Add regression and GUI-interface chapters to the manual.
27 - Fix "label" of a volume that is already in the catalog,
29 - Correct the test for spooling attributes which was backward.
30 This caused the creation of a spool file in the working
31 directory for every job run.
32 - Print to the job output stream when block checksum errors
33 occur, but continue processing. If more than one occurs,
34 print number at end of job.
36 - Remove a few unneeded tests from configure.in
37 - Modify configure to use -pthread on FreeBSD 4.8,
38 but nothing on FreeBSD 5.1
39 - Clarify that bmicrosleep() takes sec and microsecs
40 - Apply Franc Carter's code to specify a DB host in
42 - Fix sanity login length test in stored/dircmd.c
43 - remove old code fro stored/job.c
44 - Lots of work on regress so that it works on FreeBSD
47 - Make the default gnome-console wider (console.c)
48 - Add \r to the items that terminate a token.
49 - Replace a number strcpy() calls with bstrncpy().
50 - Added code to ensure that the names for each resource
51 type are unique. Two resources of different types can
54 - Added new full length modifiers for time intervals.
55 Compatible with previous method EXCEPT a modifier is now
58 - Completely restructured recycling. It should work now, but
59 in any case, the logic is much cleaner. A lot of new
61 - Added a few pthread_xx_destroy() for items inited.
62 - Nic Bellamy pointed out that it wasn't necessary to do
63 destroy() of static initialized pthread variables -- fixed
65 - update VolStatus asked for the media and volume twice.
66 - Changed a few more strcpy() to bstrncpy().
68 - Made VolBytes=1 as indicator that the Volume is labeled.
69 - Modified creation of Media record to include VolBytes,
70 same for sql_update - also set LabelDate if VolBytes = 1.
71 - Copy any statically linked programs to install directory.
72 - Change relabel flag to label in Update_media protocol Dir<->SD.
73 - Change a few strcpy to bstrncpy ...
74 - Update Media record after an automatic tape label.
75 - Added more debug code to bnet_server and bnet.c to attempt to
76 track down Alex's SD segfault (BSOCK==0)
77 - Add additional debug code for authentication errors in cram-md5.
79 - Implement "list nextvol job=xx" and add Volume to Dir status.
80 - Retry waitpid if interrupted -- needed for FreeBSD!
81 - Move lex.c debug level to 900
83 - Add new aliases SDAddress, ...
84 - Final changes for variable expansion
85 - Possible fix to Alex's SD crashes
87 - Correct incrementing counters not in catalog -- reported by
89 - Correct memory leak in core var.c code -- reported by
91 - Make documentation of variable substitution a bit clearer.
92 - Apply Nic Bellamy's patch to dbcheck.c to fix SQL to work
93 on both MySQL and SQLite.
95 - Finally fixed the .messages that was sent during prompts
96 in the console with no readline.
97 - Implement new Include/Exclude semantics.
98 - Fix a good number of segfaults reported by Chris Allen in
99 the variable expansion code.
100 - Print warning message if a job is blocked by user unmount.
101 - Fix additional variable expansion problem reported by Chris Allen.
102 - Fix pathconf() error status bug that caused a segfault on
103 a FreeBSD system -- reported by Gernot Hueber.
105 - Make sure the console sends ".messages" only when at the
107 - Apply a patch from Nic Bellamy that corrects a file descriptor
108 leak in the Storage deamon when a job is canceled. (Thanks)
110 - Implement < and | in File daemon.
111 - Implement ClientRunBeforeJob and ClientRunAfterJob.
112 - Corrected BSRatEOF to be BSFatEOF in btape --
113 reported by Lars Koller.
114 - Documented BSFatEof
115 - Eliminated save_level in FD replaced by JobLevel.
116 - Increased MAX_RES_ITEMS from 35 to 50 to handle
117 new ClientRunBeforeJob ...
119 - Add line number in error message for restore from file.
120 - Correct editing of jobids (misplaced comma).
122 - Implement restore files.
124 - Quickie patch to allow Purged Volumes to be mounted.
125 Must review algorithm.
127 - Eliminate a duplicated query from query.sql
128 - Restructure ua_restore.c so that I can add restore files.
129 - Correct positioning problem at beginning of a second volume
130 introduced with the new forward spacing code.
132 - Eliminated a sprintf() in the tree routines.
134 - Added HOST_OS, DISTNAME, and DISTVER to the status output.
135 - Eliminated a few stray Dmsgn(000, messages making them either
136 debug_level 100, or Pmsg().
137 - Made the default "Maximum File Size" 1Gbyte in SD.
138 - Cache path in tree.c to reduce calls to make_tree_path
140 - Documentation updates
141 - Implement forward space file and block when reading a bsr.
142 - Fixed a bug in db_find_next_volume() where the VolStatus was
144 - Rework some of the query.sql records that were incomplete.
145 - Fixed an ambigous SQL statement in restore.
146 - Fix proper sorting order in restore display last 20 jobs.
147 - Remove duplicate JobId's in feeding the directory tree.
148 - Fix an orphaned ua buffer due to a return that should have been
149 break so that cleanup code executed.
150 - Changed some strcats and sprintfs int bstrncat() ... to prevent
151 possible buffer overflows.
152 - Fix max file size code on tape so that after writing EOF,
153 an appropriate jobmedia record is created.
154 - Fix error messages in dev.c, which were copied into lots of
155 different subroutines without updating the text.
156 - Add reposition_dev(), and make more of the return statuses of
157 dev routine standard.
159 - Eliminate old semaphore and workq code. Keep only new jobq code.
160 - Try to get a better default size for the gnome-console
161 - Add code to avoid race conditions in starting/stoping the
164 - Correct potential race condition in heartbeat_stop().
166 - Correct segmentation fault in mysql.c if no password given.
168 - make pm_strcat and pm_strcpy return the string length
170 - Use Phil's code to get the unadorned job name.
172 - Move the MTIOCERRSTAT from just after a write() failure to
173 after writing the EOF marks on FreeBSD systems.
174 - Enhance packet too big error in bnet.c and add some
175 ASSERTs in the send code.
176 - Set heartbeat interval to zero by default.
177 - Add Recycle to list of Update Volume parameters.
178 - Use bget_dirmsg() wherever possible in Director.
179 - Split next_volume code from catreq.c to next_vol.c
180 - Consolidate editing job codes into a single routine.
181 - Add Job resource name
182 - Remove check for Win32 attributes in bls. It doesn't
183 need to read them, only print ls -l.
185 - Add SDConnectTimeout in FD.
187 - Add Scott's perlgui directory
189 - Upgrade from RH7.3 to RH9. New autoconf.
190 - Eliminate gnome2-console directory.
192 - Correct English in status command.
193 - Eliminate old shell expansion code and use BPIPE to call shell
196 - Documentation as usual.
197 - Add a new bacula.spec for Mandrake
198 - Add ownership to alist items. Default the list owns the items.
199 - Make record_cb return a status (preparation for internal use).
200 - Remove all clearing of remainder in read_record.c -- not necessary
201 - Write a tapetest program for FreeBSD end of tape testing.
202 - Modify read.c in Bacula to use read_record.c
203 - Implement multiple records in read_record.c. One for each session that
204 is open. Free the record with the EOS_LABEL is found (or at the