Kern Sibbald [Sat, 23 Dec 2006 16:33:53 +0000 (16:33 +0000)]
kes Add code in catreq.c to reject volumes not marked Enabled.
kes Add a few more ACL access checks.
kes Add \n in gnome2-console restore command so that output prints
nicer -- i.e. lines do not run together.
kes Fix autolabeling so that it will not mark a volume in error
if the volume was not actually opened. This should fix bugs
#737 and 738.
kes Require tape drive to be open before autolabeling.
kes Require explicit mount command on non-tapes before autolabeling.
kes Implement an open_client_db() that searches the keywords for
either a catolg or client, and opens the appropriate catalog.
This makes Bacula adjust better to multiple catalogs.
kes Use more name() methods for resources rather than hdr.name.
Kern Sibbald [Fri, 22 Dec 2006 15:01:05 +0000 (15:01 +0000)]
kes Apply Eric's patch for recycling the Scratch pool.
kes Note, for the tray-monitor (or almost any restricted console_
to work, in version 1.39.x you must add CatalogACL = catalog-name
kes Add appropriate new #undefs to autoconf/acconfig.h so that
configure works correctly.
kes Implement configure code to allow the user to disable IPv6 with
--disable-ipv6
kes Apply Richard Mortimer's patch to prevent seg fault if mount point
not specified -- SD refuses to start.
kes Add magic.bacula from bug #715 to the scripts directory along with
Arno's explanation of how to install it.
kes Convert a bunch of hdr.name into name() method calls.
kes Fix the code to remove .bsr files. Fixes bug #736 -- Martin.
Jeremy C. Reed [Wed, 20 Dec 2006 22:32:48 +0000 (22:32 +0000)]
Another mega patch. Sorry it includes so many unrelated changes.
They are summarized below. Kern okayed this via bacula-devel list.
If you notice any issues, please let me know.
- Use latex2html -split 3 instead of -split 4 for the "make web" target.
NOTE: be sure to update your Makefile manually or from the Makefile.in.
- Switches top level sections to \chapter.
- Gets rid of asterisks for sections so they are all numbered now.
- Removes the now un-necessary and now redundant \addcontentsline for
table of contents (not needed because numbered sections are
automatically added so was doubling them).
- Shortens length of several section (and now chapter) titles (so this can
look better in table of contents and fit better in printed book.
Also a few of these are rewritten. (Note that I didn't modify the long
FAQ section names.)
- Removed some tex formatting that were embedded in some \section* because
now they cause tex errors.
- Adds many of my TODO: items as tex comments directly near problems.
Some of these are just questions.
- Uncomment label_fd label (as in my custom book tex file references the
GNU Free Documentation License).
- Change label name of PostgreSQL_phase2 to PostgreSQL_configure and
renamed that section.
- Removes an old comment about old FreeBSD port before it was updated.
- Removes some index items that were bogus or redundant; or rewords some
of index items.
Hopefully nothing else. (After this last batch of changes, I won't
get so far behind and will be able to commit related items at one
time.)
TODO: grep the tex files for TODO: and share your feedback or fix :)
Jeremy C. Reed [Wed, 20 Dec 2006 18:17:01 +0000 (18:17 +0000)]
Many, many changes but no real changes to content nor formatting.
- many spelling fixes
- various typo and punctuation fixes
- spell out digits as appropriate (help doc to be consistent)
- or use digits instead of words as appropriate
- fix some minor grammar issues
- mention "Linux" if referring to a Linux "kernel"
- proper use of a couple vendor names (to be consistent)
Okayed by Kern. (Note this does NOT include the TODO: nor \chapter changes.)
Kern Sibbald [Wed, 20 Dec 2006 11:35:40 +0000 (11:35 +0000)]
kes Add Volume name to Forward spacing message.
kes Update migration job start time (so it is not zero) to prevent
immediate recycling of the job. This *should* fix bug #733.
Kern Sibbald [Sun, 17 Dec 2006 12:42:56 +0000 (12:42 +0000)]
kes Add additional messages when Job canceled automatically for
max run time or max start wait exceeded. This fixes bug #621.
kes Update maxruntime-test in regression script to properly test
max run time -- from info given in bug #621.
kes Simplify automatic cancel code.
kes Add check for job_canceled() in FD when contacting SD so that if
job is canceled from max runtime, it terminates faster.
Kern Sibbald [Sat, 16 Dec 2006 11:10:17 +0000 (11:10 +0000)]
kes Update projects file for next vote.
kes Implement a fix (more could be done) for bug #728 where a double
EOF is writing on the Volume when spooling is enabled and the
connection with the FD fails (FD crash, comm error).
kes Disable attempting to autolabel when polling and device is not
a tape. This fixes bug #730 where polling with a removable disk
not mounted caused lots of new Volumes to be created then marked
in error.
Kern Sibbald [Thu, 14 Dec 2006 11:41:02 +0000 (11:41 +0000)]
kes Complete hopefully the last of the copyright transfer changes.
kes Separate NumVolumes in SD to NumReadVolumes and NumWrite volumes so
that migration writing to multiple volumes doesn't mess up reading
Volumes. This should fix bug #720 -- multivol migration fails.
kes Make migration SQL look only for Type='B' jobs to migrate. This should
fix bug #724 jobs getting migrated twice.
kes Convert remaining dev_cap(dev, ...) to use dev->has_cap().
kes Implement dev->clear_cap() and dev->set_cap() methods.
kes Turn off disk seeking on restore for DVDs since I am not 100%
sure it works.
Kern Sibbald [Wed, 13 Dec 2006 19:42:12 +0000 (19:42 +0000)]
kes Implement code that should properly set that a job was migrated,
if the migration was successful even if there were no file to
migrate.
kes Apply Richard's patch to fix building or filed/restore.c without
OpenSSL support.
kes Ensure that at least one block is written by the SD even if the
FD dies or fails -- should fix bug #728.
kes Convert stored/record.c ASSERT into a soft fail if the length of
a record is too insanely long. This should fix bug #???
Landon Fuller [Sun, 10 Dec 2006 01:37:49 +0000 (01:37 +0000)]
Fix job byte reporting -- the digest code runs in verify.c and increases this value.
The better solution is to digest the stream as we receive it, thus removing the need to re-read from disk at all, but that requires more extensive and potentially destabilizing changes.
Landon Fuller [Sat, 9 Dec 2006 23:50:00 +0000 (23:50 +0000)]
This commit fixes backup and restore of HFS+ resource forks.
Unrelated to the above, I added a long-missing 'break' statement in HFS+ attribute handling that resulted in restore failures.
Kern Sibbald [Fri, 8 Dec 2006 14:27:10 +0000 (14:27 +0000)]
kes Mark build-depkgs-mingw3 executable in configure process.
kes Update autoconf/config.h.in to include LOCALEDIR
kes Undefine NLS settings for wxWidgets as there is a conflict with
their way of doing things.
kes Implement a rpool and rpool_source in jcr.h for Migration jobs.
kes Change a bunch of ->hdr.name to ->name()
kes Straighten out handling of read and write pools for Migration, including
getting the code in the proper order.
kes Improve migration error messages and debug output.
kes Shorten code path in migration if nothing found to do.
kes Implement read/write pool in migration job report.
kes Move ua_run.c select_job_level code into a subroutine (readability).
kes Move ua_run.c display_job_parameters into a subroutine.
kes Do not bash dcr->VolumeName in get_volume_info() because it makes
restore dependent on a valid catalog containing the Volume name.
Kern Sibbald [Wed, 6 Dec 2006 18:30:08 +0000 (18:30 +0000)]
kes Put make_catalog_backup SQL_BINDIR in an environment variable to
make it easier for the user to change.
kes Fail any backup job that tries to use encryption and sparse options.
Hopefully this is a temporary change.
kes Correct the EndJob return string for non-backup jobs.
kes Review the encryption restore code.
kes Implement a new encryption regression test with compression.
Kern Sibbald [Wed, 6 Dec 2006 08:39:18 +0000 (08:39 +0000)]
kes Modify migrate not to fail if nothing found to migrate.
kes Add #undef HAVE_OPENSSL, HAVE_TLS, HAVE_CYRPTO to autoconf/config.h.in
because they were not always set correctly.