+ Release Notes for Bacula 5.2.0
+
+ Bacula code: Total files = 1,088 Total lines = 219,172 (Using SLOCCount)
+
+General:
+--------
+
+This 5.2.0 version is a major release since the last version 5.0.3.
+It includes well over 50,000 lines of changes and a number of important
+new features:
+
+Compatibility:
+--------------
+ As always, both the Director and Storage daemon must be upgraded at
+ the same time.
+
+ Older 5.0.x and 3.0.x File Daemons are compatible with the 5.2.0
+ Director and Storage daemons. There should be no need to upgrade older File
+ Daemons.
+
+Changes and New Features:
+- Use of PostgreSQL cursor when doing large queries, should reduce the
+ memory usage during restore and accurate backups.
+- Includes better handling for Windows repares points, mount points, and
+ junction points.
+- A few path name length limits in Windows have been removed.
+- New SQL backend and ./configure that permits building multiple catalog
+ backends in one build, and has the main Bacula core code totally independent
+ of the SQL backend, which should simplify packaging (only the libbacsql.so/dll
+ needs to be changed.
+- New database format
+- Director selection option in bconsole
+- restorejob command in bconsole
+- Checkpoint of jobs so that in many cases they can be restarted
+- Maximum Bandwidth Per Job in several places
+- Disable batch insert config item
+- setbandwith command
+- limit command
+- Bacula Virtual File System for fast catalog access
+- New Pruning algoritm -- safer and faster
+- Additional security against injection of illegal characters
+- New block delta backup/restore
+- Indent lsmark during restore
+- Additional script variable editing
+- Improvements in ACL and xattr handling on additional platforms:
+ AIX, Darwind, FreeBSD, HPUX, IRIX, Solaris, Tru64
+- Many new plugin API features
+- Faster crc32 algorithm
+- Improved memory handling for restores
+- Improved hash table for hard links
+- Significant stability enhancements to bat
+- Many additional features for bat
+- Tray monitor for Windows
+- New FTP device handler for Storage daemon
+
+Bug fixes
+
+
+===========================================================================
+
+This 5.0.3 version is significant bug fix update to version 5.0.2.
+It includes new code and some new features.
+There is no database change since prior versions of 5.0.
+
+Note, if you are running 5.0.0, read the 5.0.1 and 5.0.2 release notes below in
+ the 5.0.1 release section.
+ Important!!!
+
+Compatibility:
+--------------
+ As always, both the Director and Storage daemon must be upgraded at
+ the same time.
+
+ Older 5.0.x and 3.0.x File Daemons are compatible with the 5.0.3
+ Director and Storage daemons. There should be no need to upgrade older File
+ Daemons.
+
+Changes and New Features:
+- If a VSS (Windows) snapshot fails, the whole Job is failed, previously
+ it printed a warning message and continued.
+- New editing codes %b = Job Bytes and %f = Job Files for runscripts
+- Any non-Bacula project plugins (i.e. third party plugins)
+ must be upgraded to specify AGPLv3 in place of the GPLv2 used previously
+ for the PLUGIN_LICENSE.
+- The bat Media page text columns can be sorted by clicking on the column.
+- If possible, the SD will automatically update the catalog Volume size if
+ it does not match the Volume size. The mismatch between the Volume sizes
+ typically occurs when the FD has a comm line drop or the FD crashes.
+
+
+Bug fixes
+1538 1554 1564 1567 1568 1569 1571 1577 1581 1582 1587 1594 1595 1606 1610
+
+- Change license from GPLv2 to AGPLv3
+- Make Win32 no filesystem change more explicit
+- Backport Branch-4.0 release to Branch-5.0
+- Massive backport from Branch-5.1 to Branch-5.0
+- Fix postgresql catalog creation when version is not on the first line
+- Fix bug #1610 handle empty xattr values on Linux and xBSD.
+- Garbage collect memory pool at end of job and before sm_dump
+- Add xattr seg fault protection suggested by Marco for bug #1610
+- Add archlinux to os.m4
+- More fixes for 32/64 bit problems in smartall
+- Fix 32/64 bit problem in smartalloc dump routine
+- Fix bad copy/paste in commit c88dccb88 prably a seg fault
+- Add .dump and .exit commands for daemons
+- Make SD automatically fix the Volume size in the Catalog when out of sync
+- Use Pmsg in smartall.c rather than printf for tracing Windows
+- Up maximum block size to 20M
+- Fix for TLS bugs #1568 and #1599
+- Improve comm line error handling for TLS, may fix bug #1568 and #1599
+- Fix problem when sending Cancel event to plugin
+- Convert all Jmsg on the watchdog thread to Qmsg
+- Fix SD crash due to mismatched lock/unlock in error condition
+- Fix bug #1587 if you have clients with different catalogs configured, both
+ catalogs contains all clients after a reload or restart of the dir
+- Fix bug #1577 During migration: ERROR in block.c:950 Failed ASSERT: dev->is_open()
+ by moving and commenting out the ASSERT()
+- Skip HB kill in FD if HB thread terminated
+- Fix crash from unequal volume_lock/unlock calls
+- New prunning algorithm -- should be more accurate
+- Save any dumps during regression to dumps directory
+- Fix crash from rw_lock/unlock miss match
+- Attempt to fix duplicate job kill seg fault
+- Replace pthread_kill by my_thread_send_signal() in jcr.c
+- fix #1595 about batch mode detection problem for postgresql with non standard
+ install dir
+- Fix #1594 about prune copy jobs
+- Apply literal string fix in printf submitted by Luca Berra
+- Fix bug #1582 Restore from multiple storage daemons breaks subsequent backups
+- Fix segfault on "cancel jobid=" command
+- Fix bug #1554 Windows installer doesn't honor /S flag
+- Remove pool zap code that breaks 2drive-concurrent-test
+- Fix #1571 and install libraries with 755 perms
+- Fix bug #1581 ASSERTs in src/dird/jobq.c can never trigger
+- Fix problem with BaseJob and Accurate FileSet options on client side
+- Fix for bug #1569 deadlock/crash in Dir
+- Fix segfault when loading Plugins
+- Fix #1567 about display of long volume names truncated during restore
+- Added new bacula-libs package.
+- Fix bug #1538 Start Scripts return incorrect value
+- Fix bug #1564 init scripts not LSB compliant
+
+See ChangeLog for more details.
+
+=============================================================================
+
+
+============= Release 5.0.2 ========================
+
+Changes since 5.0.1
+-------------------
+
+Bug fixes
+1502 1511 1517 1524 1527 1532 1536 1541 1549 1551 1553 1559 1560
+
+- Probable fix for SD crash bug #1553
+- Fix #1559 problem when restoring pruned jobs with a regexp
+- Fix for bug #1560 bcopy cannot find Volume
+- Fix cancel crash bug #1551
+- Check if sql backend is thread-safe
+- Correct Pool display in SD status. Fixes bug #1541
+- Fix cancel crash reported by Stephen Thompson
+- Rewind on close to fix #1549
+- Remove closelog() in bpipe fixes bug #1536
+- Fix #1517 about missing Base level in .level command
+- Replace ASSERT in block.c with fail Job
+- Fix database locking calling db_lock and returning from function without
+ calling db_unlock.
+- Add missing db_unlock to bvfs_update_cache.
+- Fix #1532 about permission on binaries
+- Fix #1527 about deadlock during migration
+- Another fix for OpenSSLv1
+- Add -lrt to Solaris links
+- Fix tls.c for OpenSSLv1
+- Fix #1511 when trying to insert more than 50.000 directories in bvfs
+- Fix plugin load not to stop if one plugin bad -- pointed out by James
+- Remove --without-qwt from configure statement.
+- Second correct fix to bug #1524 verify fails after adding or removing files
+- Fix bug #1524 verify fails after adding or removing files
+- Apply fix suggested by Andreas in bug #1502 for mediaview column sort problem
+- Fix OpenSSL 1.x problem in crypto.c on Fedora 12
+- Display AllowCompress warning message only if compression used in FileSet
+
+=============================================================================
+
+
+============= Release 5.0.1 ========================
+
+!!!!!!!!!!!!!!! NOTE FOR 5.0.0 USERS !!!!!!!!!!!!!!!!!!!!
+The Allow Duplicate Jobs directive has been significantly
+reworked, and the default value has changed. See below.
+
+Truncate On Purge has been totally rewritten. See the new
+features section of the manual.
+
+When Volume Poll Interval is set in the SD DEVICE configuration,
+(default 5 mins), after a certain number of polling tries (approx
+10) polling will stop and the operator will be asked to
+resolve the problem. Previously there was no limit, and an
+error message could be produced at each poll attempt.
+!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
+
+Changes since 5.0.0
+-------------------
+- We believe that we have resolved most of the problems
+ concerning canceled or failed jobs being "stuck" in the
+ Director. There is one outstanding problem in the SD when
+ canceling jobs that we will fix in the next major release.
+ If you see jobs that seem to be stuck, in general issuing
+ a cancel command in bconsole should now make them go away.
+
+Directives:
+- The default for "Allow Duplicate Jobs" has been changed from
+ no to yes. If you use this directive, please check your
+ conf file, and note the next two items !!!!!!!!!!!!!!!!!!!
+- AllowHigherDuplicates disabled. It did not work as documented
+ and was confusing.
+- New directive "CancelLowerLevelDuplicates" See New Features
+ section in the manual.
+- Truncate on Purge rewritten. See New Features section in the manual.