|
Sitback is currently pre-release.. This means that
f.ex. the configuration and command-line options may still change. Also
archive-creation and the database can still change, but the code
seems to be somehow stable now. I am still running test's, but dont forget that
the data is only handled by tar so you dont risk anything by using sitback now..
Restore operation is not yet implemented, but as you know, just use tar to
extract the files. I want the backup operation to be stable and the amount
of extra information included in the archive not to change anymore before
implementing restore and list operations..
There is also some code-cleanup left, so dont blame me if you see some
messy or inefficient coding.. Im still working on that..
Please go ahead and try sitback.
Bug-reports, ideas and comments is welcome,
but please dont write to me, to tell me that you know about another application
that does the same thing as sitback, I know that already..
I know f.ex. that there is something called Taper, i dont know where to get
it but as far as i can tell, it has the same basic idea as Sitback.
Go look for it if sitback is not what you expected or looked fore. It works for me, and
that is what is important...
- Pre-release 0.2.x
-
0.2.0 ????????
-------------------------
- Fixed bug with verify failing when updating archive.
- Fixed bug that would leave report file empty after
failing verify
- Improved report content after failing verify
- Improved view of file currently being processed.
- Fixed a bug that prevented directory's with blank
spaces in the name to be included in the archive.
- Pre-release 0.1.x
-
0.1.20 03192000
-------------------------
- Fixed bug with verify failing when updating archive.
- Fixed bug that would leave report file empty after
failing verify
- Improved report content after failing verify
0.1.19 03152000
-------------------------
- Searching standard locations for tar, gzip, bzip2, compress
and mt if not found in the PATH.
- Now using best available compression, 1) bzip2, 2) gzip, 3) compress
or no compression if no tools is available.
It is possible to select one of the methods in the script, f.ex.
if you dont want to use bzip2 although it is available..
- Improved error handling when bad symlinks is found when the
archive list is being prepared.. Bad symlinks is now just
reported to the log but the operation continues.
- Various updates to the webpages (including the manual). Nothing
important, just ordinary housekeeping...
- Changes in the script parser. Invalid directives is now reported
to the log. Trailing garbage is ignored, and a line can be
'commented out' using '#'..
- If there is errors during initialization, the user must press a
key to exit, so that he/she can see the messages...
- Configure script detects and tests the Q toolkit. In the end, X
support via Qt is permanently disabled even if it is available. This
is done since the goal is to get a stable 0.1.x version, before
introducing anymore features.. The 0.2.x versions will start using X, if
available.. But still, Sitback will _never_ be an X-only application.
The primary usage of sitback is as a backup tool on console based servers.
0.1.18 03122000
-------------------------
- Fixed two bugs relating to non-rewinding tape devices.
0.1.17 03102000
-------------------------
- Hey.. it never stops... Fixed a little bug,
introduced by another bugfix, so that we again
may read the full volume name from a script, even
if there is blankspaces in it..
0.1.16 03092000
-------------------------
- Well. seems we had a little bug..
In certain situations, the smb password for
printing the backup report could be modified
due to a bug in the script parser.. This could
make the print-operation fail.... Not fatal, but
very annoying....
0.1.15 03092000
-------------------------
- Fixed a bug that would terminate operation
instead of waiting for the device to be available
if the device is a magnetic tape.
- Fixed a missing prototype..
- Better function,libs and headers check in the configure
script..
- Checking for color_set() and wcolor_set in libncurses,
if not available, not used.
- Fixed a bug where cr/lf was not removed properly
if the script is a windows-style textfile.
- Fixed a bug that would prevent the archiveid file
to be written to the archive if no scriptname
was supplied when using the '-i' switch.
- Small changes to the curses interface.
- Internal structure for detecting and using best available
interface (X,ncurses,text) ready...
- Solved several problems with backup and verify of
multiple volumes..
- Sitback tested on scsi tapes, magnetic tapes, zip disk's
floppy disk's and local filesystem.
- Found and fixed several small non-fatal bugs.
0.1.14 03032000
-------------------------
- Better error-reporting when stat() fails while checking
locations.
- Replaced sigblock wich is obsolete with POSIX signal
functions, now using sigaddset and family.
- Only using console ioctl to beep the bell on linux.
Other platforms may live without this little gimmick.
- Small fixes to increase cross-platform compatibility
- Targets that fail initial CheckLocation() is not
added to the filelist, but all other targets is, and
the operation continues without the invalid targets.
- Better user interface. Curses based interface shows
progress while doing single backup's. This interface
will later also be used for list and restore operation.
Later on there will be a sitback version for X where
Qt is used to provide the user interface.
0.1.13 03012000
-------------------------
- Bugfix: When running as daemon with retry count declared in
the script or on the command line, a debug message would prevent
the session group leader from terminating properly when starting
the daemon. This would hang the startup script, wich would have
to be ended with ctrl-c. This could hang initialization of the
whole system, since init would hang after starting sitback...
- Solved possible race condition that could remove a lock file
before allowed, thereby allowing access to a busy archive.
0.1.12 02282000
-------------------------
- Removed option '-v', verbose.
Sitback will be verbose when not running as daemon
- New option: '-y', retry count.
Retry backup or restore operation a certain number
of times if there is an error. This option is
added because some older drives might fail once in a while
for no apparent reason.
0.1.11 02242000
-------------------------
- Bugfix: Forgot to reset size counter when starting
to build the archivelist, daemons would add the previous
size to the total size. Not critical at all, just
confusing in the report
0.1.10 02232000
-------------------------
- Small cosmetic changes to error-reporting
0.1.9 02212000
-------------------------
- Small cosmetic changes to log-messages
- Better timeout control. Previous, a timeout
would make sitback terminate, now, an error
is returned to the callee. So even if a backup
daemon has timeout-error on the device
(the pretty but rather confused secretary
forgets to put a tape in the drive), the
current backup would just end with an error,
but the daemon keeps on running and start the
backup next time it is scheduled to do so..
0.1.8 02202000
-------------------------
- Internal changes to backup and verify.
Instead of using a shell-script, tar is
executed (with execvp()) and the result
is then analyzed.. This should allow
the user to interrupt tar.
- Internal changes to functions for searching
an existing archive id. Now using execvp
to ensure tar can be interrupted.
- Renamed switches '-v' and '-h-' to
'--version' and '--help'
- New switch '-v'. Allow tar to be verbose
when doing backup, restore and verify
- Bugfix: Although only update of archive
was requested ('-u' switch), 'mt' was
used to erase current archive if the
device is a magnetic tape.
- Verify of multiple volumes enabled
- Removed all 'system()' calls, so
that we dont get uninterruptible
processes.
- Locking device or archive when starting
backup operation, so no more than
one backup operation may run on a
single archive or device when having
multiple sitback instances
- Cosmetic changes to log-messages
- Moved the Logfile to /var/sitback.log
- Moved scripts to /etc/sitback/*
- Installed html documentation to /doc/sitback/html/
- Non-rewinding Magnetic and SCSI tapes is rewinded
- Backup daemons now properly setup, (become
process group leader, exit, chdir("/") etc.)
0.1.7 02152000
-------------------------
- Cosmetic changes
0.1.6 02142000
-------------------------
- Added a few extra log messages
- Modifyed the beep, when a new tape
or disk is required when writing
multiple volumes
- If the archive is one of the ftape
devices (a magnetic tape of some sort),
'mt' is called to erase, then retension
the tape
- Bugfix: Error message from backup was written to
the report _after_ the file was closed..
- Bugfix: Error message from verify was written to
the report _after_ the file was closed.
- Slight modification of Log messages
- Report is printed even if backup or verify
fails.
0.1.5 02132000
-------------------------
-Various code-clenaup.
- New switch '-u'. Update archive instead of
recreating it.
- Finally got around the documentation,
even prepared the webpages, so that
sitback can be made public available..
- No more functions should be added to sitback
at this moment, unless something is really
missing.. The code should go through some testing,
and when ready, version 0.2.x series is started.
0.2.x versions will include restore or list operation,
cant deside wich one at the moment..
0.1.4 02122000
-------------------------
- Merged -l and -m, if -l has no data, tar -M is
used, else tar -L is used.
- Better command-line help.
- If new archive-id is set on an archive with the
-i switch, no target files is included, only
archive-info.
- New switch '-s'. Skip attempt to resolve current
archive-info on a archive. Good when creating
a new archive-id f.ex. on a tape.
- Better code for getting archive-id from a archive.
tar is killed as soon as the archive-id file
is ready.
0.1.3 02102000
-------------------------
- Blocking SIGPIPE in daemon mode (why did i forgot
to do that before ???, stupid me!)
- New switch '-n'. Ignore time-info, start operation
now..
- Support for VOLUMENAME script directive
- Volume name is written into the report
- Bugfix: Check made from init.d could corrupt the
previous version of the archive, critical later
on when restore or list will be implemented.
- Added an 'archiveid' to the tapes/archives.
- New switch '-i'. Force archiveid.
- Added database of archive id's and there
content. This will be usefull later on
in the list and restore code.
- Checking that we have tar and gzip available
0.1.2 02072000
-------------------------
- Support for MULTIPLE=[YES|NO] script directive
- Support for VOLUMELENGTH script directive
- Better sanity checking
- Checking if the archive can be created or the
device opened
0.1.1 02062000
-------------------------
- Bugfix: completion time now correct reported
- Added logentry upon startup
- No longer descends into subdirs when checking
locations for the filelist
0.1.0 02042000
-------------------------
- Bugfixes, ao. to thread synchronization
- TARGET script directive changed to ARCHIVE
- FILE script directive changed to TARGET
| |