On Mon, 8 Dec 2008 06:48:52 -0800 Kern Sibbald wrote
> This is a support issue and is either a:
> 1. Configuration error (most likely)
> or
> 2. An OS (driver) bug (less likely)
> > After noticing a curious warning message from a catalog backup, I have
> > been doing a little investigation and I
Kern Sibbald wrote:
> 3.0.3 in the dashboard regression testing -- particularly the Solaris
> problems to ensure that they are not Bacula bugs.
>From what I can see, it looks as if they are not Bacula problems. The warnings:
ld: warning: file /usr/sfw/lib/libstdc++.so: attempted multiple inclusi
Dan Langille wrote:
> There was a problem, a few years back, but since solved, with FreeBSD's
> pthreads. It would cause data to be "written" to the end of the tape,
> but it would not actually reach the tape.
> http://mlblog.osdir.com/os.freebsd.devel.scsi/2003-06/msg00018.shtml
I am actually
Hi, all,
After noticing a curious warning message from a catalog backup, I have
been doing a little investigation and I now think that, with certain
types of drive, a backup which spans more than one tape can corrupt the
data being written at the point where the tape change takes place.
When a ta
JCR's job type
2) Only increments num_jobs_run if jcr->JobId > 0
Allan
Original Message
Subject: Status Command Changes The Job Type
Date: Wed, 13 Feb 2008 20:23:21 +0000
From: Allan Black <[EMAIL PROTECTED]>
To: bacula-devel@lists.sourceforge.net
Just had a l
I am trying to run the regression tests:
ctest -D Experimental -R all-non-root:
but I get the message:
No tests were found!!!
I have searched my copy of the regress directory, but cannot
find any reference to all-non-root. Perhaps I am trying to
run a test set which has recently been removed ..
Hi,
Just run the regression tests on Solaris, and scripts/create_sed
generated a few error messages. The script has:
LIBDBI=${LIBDBI:- }
which causes an error with the Bourne shell, because it parses
the space before the brackets, causing it to interpret the } as
a command, resulting in:
scrip
Kern Sibbald wrote:
> Nice. I *did* however change a couple of lines of code that were probably
> causing a failure on Solaris when executing the unload plugin code when none
> existed.
Yes, I had noticed that bacula-dir and bacula-fd had stopped dumping core
during regression testing.
> Well,
what more can I say?
I did not have to change a line of the source code, it just
worked (once I got the configuration sorted out and realised
that the PluginDirectory was supposed to contain bpipe-fd.so
and not /usr/mysql/bin/mysqldump - doh!)
FileSet {
Name = "MySQL:regress"
Include { O
Kern Sibbald wrote:
On Friday 29 February 2008 00.33:23 Allan Black wrote:
Eric Bollengier wrote:
Good idea, but it would be much greater if this option was available in
user interface (menu). Direct command line is always
optional/obscure/undocumented.
OK, I have done that, and the patch is
Kern Sibbald wrote:
> I would appreciate it if everyone could try the latest SVN code. In general
> after pulling it down, you need to do a "make distclean" to get it to
> synchronize correctly -- i.e. to start from scratch ...
Much better ctest did everything that time, whereas before, I
Kern Sibbald wrote:
>> If that is OK, I am more than happy to go for it.
>
> Yes that is OK -- it is how I would have resolved the problem.
I will work on that, then (and also take Eric's comments into account).
When I resubmit the patch, I will include the changes I already sent, so
you can thro
Eric Bollengier wrote:
> Good idea, but it would be much greater if this option was available in
> user interface (menu). Direct command line is always
> optional/obscure/undocumented.
>
> We can imagine something like
>
> update -> Volume -> RecyclePool
> 1. Default
> 2. Scratch
> 3. ...
> 4. *N
The second patch (fix-update-ctest.diff) looks as if it has already
been applied to SVN - is that correct?
Also, scripts/update-ctest.in (and the patch) contain this code:
if [ -d build ] ; then
echo "Build directory not present, will run make setup"
[...]
which will run make setup if th
If I apply these patches, will there be any way to run the
regression tests without submitting the results to the dashboard?
If I am testing a patch, for example, we would not really want any
failures to be logged on the dashboard, if the problems are only
transient ones while hacking at a bit of
Hi,
The RecyclePoolId field in the Media table has a default value
of 0, which means the pool should be left unchanged when the volume
is recycled.
Currently, if RecyclePoolId is changed, there is no way to change it
back to zero again using the director (it can be done by changing the
database
Hi,
Just run the regression tests on Solaris, and tests/accurate-test
generated a few error messages. The script has some ksh/bash-isms,
which do not work with the Bourne shell, e.g.
$(command)
instead of
`command`
and
if ! command; then
instead of
command
if [ $? != 0 ]; then
The attache
Hi, Dirk,
I hope you do not mind, but I would like to suggest a modification
to the behaviour of the mediaedit dialog in BAT, WRT the Recycle
Pool.
Currently, the code will work exactly as expected if the Volume's
recycle pool is explicitly set to a pool (Media.RecyclePoolId != 0).
However, if R
Kern Sibbald wrote:
> Mark today in your calendar. Bacula just did its first backup and restore of
> a MySQL database using a plugin.
> I did it with using a simplistic "pipe" plugin.
Excellent!
I notice that the trunk code now dumps core on Solaris :-)
The director and fd both segfault in un
Just had a look at ua_status.c, because of something which
has been annoying me for a while.
If there are any scheduled jobs, the "st dir" command causes
num_jobs_run to be incremented when the console disconnects.
[ I noticed this while I was testing a script, and could not
understand why the "j
Kern Sibbald wrote:
> I suspect that awk would be much more portable on Unix systems.
Yes, but not necessarily the same version of awk, unfortunately.
GNU/Linux tends to have new awk, whereas other, particularly
commercial versions of UNIX (e.g. Solaris) retain the old awk,
because replacing it wo
Summary/reminder: a patch to update the state file whenever a job terminates,
rather than when the dir/sd/fd exits.
Kern Sibbald wrote:
On Wednesday 22 August 2007 22:07, Allan Black wrote:
Kern Sibbald wrote:
As it stands, the current patch that you have submitted will not work
correctly
Kern Sibbald wrote:
> What would be useful would be the full output produced when you set:
>
> export REGRESS_DEBUG=1
>
> and it blocks.
I actually have a typescript of that (although, unfortunately, I have
lost log1.out, because I ran the tests again without saving the tmp
directory).
The fil
Dan Langille wrote:
> Ahh, I think I gave up on those particular problems. And they went
> away. I've not run regression tests on trunk lately, mostly just the
Yup, if you run the tests often enough, the problem will indeed go away.
What a shame GCC cannot do that for you - if you compile the c
Dan Langille wrote:
> 2.2 branch was fine. A problem on trunk:
>
> 30-Jan 15:26 localhost-sd JobId 3: Job NightlySave.2008-01-30_15.26.05
> waiting. Cannot find any appendable volumes.
> Please use the "label" command to create a new Volume for:
> Storage: "Drive-0"
> (/usr/home/dan/s
Dan Langille wrote:
> Allan Black wrote:
>> Did anyone ever get to the bottom of Dan's problem back in September?
>> http://sourceforge.net/mailarchive/forum.php?thread_name=46E861B4.16066.6354631D%40dan.langille.org&forum_name=bacula-devel
>>
>> I have been h
Did anyone ever get to the bottom of Dan's problem back in September?
http://sourceforge.net/mailarchive/forum.php?thread_name=46E861B4.16066.6354631D%40dan.langille.org&forum_name=bacula-devel
I have been hit by the same problem (at least, it seems to be the
same one), running the regression tes
I've noticed something which I think, if not a bug, is at least "undesirable"
behaviour. Have a look at the following query (I used sqlquery because
"list media" generates long lines which is difficult to read in an email):
Enter SQL query: select MediaId,VolumeName,VolStatus,LastWritten from Medi
o
have their own locks.
In addition, as you note, you need to do
> something so that the state file is not updated more frequently than really
> necessary.
Yes, now I think about it, I can see how to do that quite easily.
Allan
> On Wednesday 22 August 2007 14:17, Allan Black wrote:
&
- Can we write the state file after every job terminates? On Win32
the system crashes and the state file is not updated.
Not just on Win32 :-) I have a similar problem sometimes, because I
don't live in a large town and I don't have a UPS.
I've changed the lib code and the three daemons, so tha
I had to add another #undef in the 'Super kludge for GNOME 2.0'
section :-) of console.h to get it to compile under Solaris
Nevada build 66, which has Gnome 2.18.0.
I have compiled this on earlier builds without problems, but I
haven't compiled it for a while, so Gnome 2.16 may have the same
prob
> On 14 Aug 2007 at 9:35, Kern Sibbald wrote:
>> On Monday 13 August 2007 23:50, Dan Langille wrote:
>>> Does this error mean anything to anyone?
>>>
>>> ==>Entering directory /usr/home/dan/src/bacula-bat/work/bacula-
>>> 2.2.0/src/qt-console
>>> make: don't know how to make depend. Stop
>>> *** Er
The C/C++ compiler selected by configure isn't being propagated
to qt-console/Makefile. I'm building Bacula 2.2.0 on Solaris.
Configure (correctly) determines that the C compiler is
/usr/sfw/bin/gcc and C++ is /usr/sfw/bin/g++.
However, BAT tries to use cc and CC respectively for the C and
C++ com
Kern Sibbald wrote:
On Wednesday 08 August 2007 15:09, Allan Black wrote:
Found what I think is a (minor) bug in the find_next_volume code;
patch attached.
Could you resend your email with the diff in unified format (diff -u). I
don't read the format you sent.
OK, here it is (and I&#x
Found what I think is a (minor) bug in the find_next_volume code;
patch attached.
I had been wondering why tapes with 'Recycle' status seemed to be
used in reverse order to what I expected
Allan
*** sql_find.c.orig Mon Nov 27 10:03:06 2006
--- sql_find.c Wed Aug 8 13:29:11 2007
***
35 matches
Mail list logo
