-
+13Mar04
+- Added two tar files that Phil sent for adding slackware support.
+ Thanks Phil and Matt Howard.
+- Fix relabel command -- || should have been &&.
+- Correct once more the scheduling algorithm -- this time it should
+ be much better.
+12Mar04
+- Implement simple spool statistics printed by "status sd".
+- Cleanup/add spool error messages.
+- On multiple ctl-c, exit from console quickly.
+- Implement code to select a different Volume in the SD if the
+ current Volume is busy -- requested by Patrick Cole.
+- Change postgreSQL make tables to use bigint instead of integer
+ for StartBlock and EndBlock because
+ there is no unsigned integer in postgreSQL and these fields use 32 bits.
+- Implement multiple drive autochanger support -- prompted by email of
+ Patrick Cole.
+11Mar04
+- Update version and date in Win32 build.
+- Fix bug in run_conf with month range zapping wday and wom.
+- Correct editing of port for PQsetdbLogin in postgresql.c as reported
+ by J. Conroy, and do a second try.
+10Mar04
+- Integrate Phil's Makefile patch.
+- Add SpoolDirectory to SD Device resource.
+- Add MaximumJobSpoolSize to SD Device resource.
+- Add MaximumSpoolSize to SD Device resource.
+- Implement the above.
+- Make despooling lock the device so only it writes
+ to the device during despooling.
+09Mar04
+- Data spooling now passes the regression tests.
+- Added "Files Expected" to the restore job report.
+- Implemented SpoolData = yes/no in the Job resource
+- Implemented SpoolData = yes/no in the Run overrides.
+- Note, you must have a 1.33.4-09Mar05 SD or later with the Director since
+ the DIR->SD protocol has changed to support data spooling.
+- Remove the confusing "Which DBMS do you want to use (please select only one):"
+ messages from ./configure.
+- Fix broken Jmsg with missing argument in find_one.c
+08Mar04
+- Second cut data spooling code.
+- Fix missing piece of yesterday's patch.
+07Mar04
+- Integrate patches supplied by Jason Conroy that fix Slot numbers
+ getting lost.
+06Mar04
+- This version contains a *major* addition to the SD structures.
+ Many variables have been moved out of the JCR into a new DCR
+ (Device Context Record). This required quite a few changes, and
+ introduced a bit of instability. In the end, after a few more
+ rounds of changes, the DCR will replace virtually all arguments
+ to the low level SD read/write routines. This change permits
+ (with a bit more code) the SD to write to multiple devices at
+ the same time for a single Job. It also facilitates adding the
+ data spooling code.
+- Added a patch that fixed the port specification for postgresql.
+- Fixed findlib/find_one.c so that it will complain about backing
+ up a hard linked file twice, which makes a restore impossible.
+- Tighten up permissions on pid file.
+- Attempt to avoid warnings when casting 64 bit pointers to an
+ integer.
+- Added skeleton code for data spooling.
+- Make sure spooled attributes are discarded on error.
+03Mar04
+- Updated getdiskinfo in rescue/linux to handle the fact that the
+ sfdisk -s option now includes info on partitions. Also, before
+ creating the diskinfo files, clean out the old ones.
+02Mar04
+- Add debug and error output to jobq.c
+- Fix some minor errors in debug output of scheduler.c
+- Add "trace on/off" command and modify trace code in message.c
+ to work off trace flag rather than #defines.
+01Mar04
+- If console is found during install warn user (new name is bconsole).
+- On install copy console.conf to bconsole.conf if it exists.
+- Make RunAfterJob error non-fatal.
+- Rework minor details in status output.
+- Attempt to define all values for "consoles" in JCR to avoid
+ confusion.
+28Feb04
+- Use net start bacula to start Bacula on WinXP/NT/2K
+ systems after installation.
+- Change a few strcpy()s to bstrncpy()s in signal.
+- Add "gui on/off" command to console to turn on gui mode,
+ which will adapt Bacula better to batch or gui programs.
+ Currently, it prevents commas from being inserted into
+ numbers in the list command.
+- Tweak some of the Makefiles so that the install is done with
+ the right program (nothing really changed).
+- Install logrotate in the "make autostart" of Bacula on RedHat.
+- Implement SIGHUP while jobs are running. It *seems* to work but is
+ a bit fragile and still crashes if you push it. More work needed.
+- Applied Phil's GNOME 1.4 patch. Very clean. Thanks Phil.
+- Major updates to the native Win32 installer.
+27Feb04
+- Restructure free of conf resources to handle job_end_push()
+ and SIGHUP. Lots of changes little substance.
+26Feb04
+- Add mtimeonly=yes/no and keepatime=yes/no to Include list
+ 1.33 style.
+- Allow no modifier in time and size specifications.
+- Attempt to implement disk seeking. More work needed.
+- Implement more in win32 installer and test it.
+25Feb04
+- Implement kb, mb, and gb modifiers for size, which mean
+ 1,000 bytes, 1,000,000, ... bytes.
+- Add Makefile to src/win32 to allow building the full
+ release -- it calls nmake where appropriate.
+- Remove the rsi directory from win32 as suggested by Christopher
+- Add Michel Meyers' nsi file to win32 and integrate it with
+ the build.
+24Feb04
+- Add the frigging _O_BINARY to the right variable. Win32 works!
+- Add debug code to state file. There is a problem with Win32 read
+ returning one byte less than it should.
+- Implement a number of new functions for alist so that it has
+ equivalent functionality to dlist. It will be used for
+ job_end_push().
+- More work on state file.
+- Added a popup message box on Windows if Bacula Aborts.
+- Attempt to shutdown the FD server, but doesn't work on Windows. This
+ would have given a cleaner shutdown, no problem.
+23Feb04
+- Update projects
+- Attempt to fix state file on Win32 -- no luck.
+- Add HPUX tweaks to configure.in
+- Update quickstart and install.wml of doc.
+- Eliminate extern int h_errno from bnet.c by including netdb.h
+- Eliminate NumJobs from last_job structure so that FD and SD
+ correctly report what is in the state file.
+22Feb04
+- Implement daemon state file, and save/restore last_jobs list.
+- Fix time routines in win32 compat.cpp
+- Fix a bug of not checking the bpipe_open() status in do_shell_expansion().
+21Feb04
+- Add Volker's bacula script to the SuSE directory and convert it
+ to bacula.in
+- Make LockRes() and UnlockRes() use read/write locks and have
+ debug info to trace locking and unlocking.
+- Begin work of implementing SIGHUP. It is now enabled for
+ the Director and works only if there are no jobs running,
+ in addition, if the new conf file is incorrect, it will
+ terminate Bacula. Much more work to be done. I've figured
+ out a scheme to implement it with running jobs without
+ adding individual semaphores or use counts.
+- Sort "list jobs" by StartTime rather than JobId, which can
+ be random.
+16Feb04
+- Add a lot of "const" to char * arguments to keep the native
+ Sun compiler happy.
+- Added first cut of message queuing to prevent recursion in
+ low level routines. This code is not yet working.
+- Spent a *huge* amount of time looking at the tape driver
+ code in the wake of several reports of tape labels getting
+ trashed. The code looks fine.
+- Created a tape with an error (by writing a block at the
+ beginning of a tape that had valid data), and found that
+ Under Linux there is no way to read past an I/O error.
+ I tried mt; I tried my own program (btape); and I tried
+ using scsitape, which talks directly to the scsi driver.
+ This is *VERY BAD* news.
+- I spent an *enormous* amount of time enhancing the btape
+ test program as well as making the fill command work with
+ both one and two tapes and the autochanger if configured.
+ The multiple tape fill test is now quite comprehensive.
+ It checks the last block on the first tape, the first block
+ on the second tape, and the last block (11) on the second
+ tape.
+- Teaked the tape driver for several functions to use the
+ OS driver's notion of mt_file if there is an error.
+13Feb04
+- Add DB update scripts retrieved by Scott to a new updatedb
+ directory at the top level.
+- Add -p option to bcopy to allow ignoring errors on input.
+- COALESCE(xx,0) does not work on PostgreSQL because the 0 is not a
+ correct time. Replaced by adding LastWritten IS NULL to the sort
+ line as was first suggested by JML.
+12Feb04
+- Use COALESCE(LastWritten, 0) in sql_find.c to get NULLs to sort last.
+ See note above.
+- Add write/read and positioning test to btape "test" program.
+10Feb04
+- Fix bad printf of InChanger flag reported by Pascal Pederiva.
+- Fix bad indexing off stack in authenticate.c reported by
+ Pascal Pederiva.
+09Feb04
+- Add \n to hosts.access reject message.
+- Implement security message class and make hosts.access message use
+ that class.
+08Feb04
+- Fix check_memory bug in ua_query.c, which gives a bus error on Solaris.
+ This affected only the query command.
07Feb04
- Added backup to cdwriter script to examples provided by Johan Decock.
- Fixed a bug where ls really did lsmark (just invert command table
projects / bacula / bacula / blobdiff