Bugzilla Bug 138495

Bacula fails to build the director, but the emerge continues on and seems to
finish properly with no exit errors etc.   After discovering that my
/usr/sbin/bacula-dir file was missing, I found that the compilation of the dir
was failing in the middle of the merge. (see log snip below)


# emerge -va bacula

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild  R    ] app-backup/bacula-1.38.9  USE="X bacula-console
bacula-split-init logrotate postgres python readline ssl tcpd wxwindows
-bacula-clientonly -bacula-nodir -bacula-nosd -doc -gnome -logwatch -mysql
-sqlite -sqlite3 -static" 0 kB


# emerge --info
Portage 2.1-r1 (default-linux/amd64/2006.0, gcc-3.4.6, glibc-2.3.6-r4,
2.6.16.16 x86_64)
=================================================================
System uname: 2.6.16.16 x86_64 AMD Athlon(tm) 64 X2 Dual Core Processor 3800+
Gentoo Base System version 1.6.15
dev-lang/python:     2.3.5-r2, 2.4.2
dev-python/pycrypto: 2.0.1-r5
dev-util/ccache:     [Not Present]
dev-util/confcache:  [Not Present]
sys-apps/sandbox:    1.2.17
sys-devel/autoconf:  2.13, 2.59-r7
sys-devel/automake:  1.4_p6, 1.5, 1.6.3, 1.7.9-r1, 1.8.5-r3, 1.9.6-r2
sys-devel/binutils:  2.16.1-r2
sys-devel/gcc-config: 1.3.13-r2
sys-devel/libtool:   1.5.22
virtual/os-headers:  2.6.11-r2
ACCEPT_KEYWORDS="amd64"
AUTOCLEAN="yes"
CBUILD="x86_64-pc-linux-gnu"
CFLAGS="-march=athlon64 -O2 -pipe"
CHOST="x86_64-pc-linux-gnu"
CONFIG_PROTECT="/etc /usr/kde/3.5/env /usr/kde/3.5/share/config
/usr/kde/3.5/shutdown /usr/lib/X11/xkb /usr/lib/fax /usr/share/config
/var/spool/fax/etc"
CONFIG_PROTECT_MASK="/etc/env.d /etc/gconf /etc/revdep-rebuild /etc/terminfo"
CXXFLAGS="-march=athlon64 -O2 -pipe"
DISTDIR="/usr/portage/distfiles"
FEATURES="autoconfig distlocks metadata-transfer sandbox sfperms strict"
GENTOO_MIRRORS="http://distfiles.gentoo.org
http://distro.ibiblio.org/pub/linux/distributions/gentoo"
MAKEOPTS="-j3"
PKGDIR="/usr/portage/packages"
PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --compress
--force --whole-file --delete --delete-after --stats --timeout=180
--exclude='/distfiles' --exclude='/local' --exclude='/packages'"
PORTAGE_TMPDIR="/var/tmp"
PORTDIR="/usr/portage"
SYNC="rsync://rsync.gentoo.org/gentoo-portage"
USE="amd64 X a52 aac alsa arts avi berkdb bidi bitmap-fonts cli crypt cups dbus
divx4linux dri dvd dvdread eds emboss encode faad ffmpeg flac foomaticdb
fortran gif gpm gstreamer gtk gtk2 hal ieee1394 imlib ipv6 isdnlog jpeg kde
kdeenablefinal ldap lm_sensors logrotate lzw lzw-tiff matroska mozilla mp3 mpeg
mplayer ncurses nls nptl nvidia ogg opengl pam pcre pdflib perl png postgres
ppds pppd python qt qt3 qt4 quicktime readline reflection sdl session spell spl
ssl tcpd tiff truetype truetype-fonts type1-fonts usb vorbis wxwindows xine
xorg xpm xv xvid zeroconf zlib elibc_glibc kernel_linux userland_GNU"
Unset:  CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, LDFLAGS,
LINGUAS, PORTAGE_RSYNC_EXTRA_OPTS, PORTDIR_OVERLAY




emerge snippet:

/usr/bin/x86_64-pc-linux-gnu-g++   -L../lib -L../cats -L../findlib -o
bacula-dir
dird.o admin.o authenticate.o autoprune.o backup.o bsr.o catreq.o dird_conf.o
exp
and.o fd_cmds.o getmsg.o inc_conf.o job.o jobq.o mac.o mountreq.o msgchan.o
next_
vol.o newvol.o pythondir.o recycle.o restore.o run_conf.o scheduler.o
sql_cmds.o
ua_acl.o ua_cmds.o ua_dotcmds.o ua_query.o ua_input.o ua_label.o ua_output.o
ua_p
rune.o ua_purge.o ua_restore.o ua_run.o ua_select.o ua_server.o ua_status.o
ua_tr
ee.o ua_update.o verify.o \
      -lsql -lfind -lbac -lm -L/usr/lib/python2.3/config -lpython2.3 -lutil
-lrt
  -L/usr/lib -lpq -lcrypt -lpthread -ldl  \
      -lwrap  -lssl -lcrypto
/usr/lib/python2.3/config/libpython2.3.a(posixmodule.o): In function
`posix_tmpna
m':
: warning: the use of `tmpnam_r' is dangerous, better use `mkstemp'
/usr/lib/python2.3/config/libpython2.3.a(posixmodule.o): In function
`posix_tempn
am':
: warning: the use of `tempnam' is dangerous, better use `mkstemp'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_compr
ess':
: undefined reference to `deflateInit_'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_compr
ess':
: undefined reference to `deflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_compr
ess':
: undefined reference to `deflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_compr
ess':
: undefined reference to `deflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_compr
ess':
: undefined reference to `deflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_decom
press':
: undefined reference to `inflateInit2_'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_decom
press':
: undefined reference to `inflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_decom
press':
: undefined reference to `inflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_decom
press':
: undefined reference to `inflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_decom
press':
: undefined reference to `inflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_decom
press':
: undefined reference to `inflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_compr
essobj':
: undefined reference to `deflateInit2_'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_decom
pressobj':
: undefined reference to `inflateInit2_'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`Comp_dealloc
':
: undefined reference to `deflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`Decomp_deall
oc':
: undefined reference to `inflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_objco
mpress':
: undefined reference to `deflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_objco
mpress':
: undefined reference to `deflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_objde
compress':
: undefined reference to `inflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_objde
compress':
: undefined reference to `inflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_flush
':
: undefined reference to `deflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_flush
':
: undefined reference to `deflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_flush
':
: undefined reference to `deflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_unflu
sh':
: undefined reference to `inflate'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_unflu
sh':
: undefined reference to `inflateEnd'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_adler
32':
: undefined reference to `adler32'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_adler
32':
: undefined reference to `adler32'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_crc32
':
: undefined reference to `crc32'
/usr/lib/python2.3/config/libpython2.3.a(zlibmodule.o): In function
`PyZlib_crc32
':
: undefined reference to `crc32'
collect2: ld returned 1 exit status
make[1]: *** [bacula-dir] Error 1
make[1]: Leaving directory
`/var/tmp/portage/bacula-1.38.9/work/bacula-1.38.9/src
/dird'


====== Error in /var/tmp/portage/bacula-1.38.9/work/bacula-1.38.9/src/dird
======

Any update/progress on this?

Is there anyhting else I can submit to help?

Re-assigning to herd wrt Bug 98505. Remove labmonkey from metadata.xml please.

I cannot reproduce this bug, app-backup/bacula-1.38.9 compiles and merges fine.

Hi  Evil Compile Person. 

Are you sure that during the bacula-dir phase of the compile there was no
error? The error I see fails to compile bacula-dir, BUT the emerge continues
on. 

Do you have abacula-dir binary in /usr/sbin when the compile is complete?

Thanks.

I have the same issue.  Whether or not Evil Compile Person has the same
problem, I think the main issue is that failures are silently ignored.  This is
dangerous, especially for critical packages like backup tools!

I don't see anything strange (e.g. something which would tell make or the
configure script to ignore errors) in the gentoo ebuild for bacula 1.38.9, but
I'm not an ebuild expert.  If Gentoo's ebuild file cannot find a way to detect
this problem (e.g. create a list of expected binaries, then check that they
exist after install), perhaps we should complain to the bacula maintainers.

Besides the reported errors, I'm also seeing errors in the console.  Consider
this snippet:

/usr/bin/x86_64-pc-linux-gnu-g++   -c    -I. -I..  -O2 -march=athlon64 -pipe 
authenticate.c
console.c:78: error: `bool tee' redeclared as different kind of symbol
/usr/include/gentoo-multilib/amd64/bits/fcntl.h:247: error: previous
declaration of `int tee(int, int, size_t, unsigned int)'
console.c:78: error: declaration of `bool tee'
/usr/include/gentoo-multilib/amd64/bits/fcntl.h:247: error: conflicts with
previous declaration `int tee(int, int, size_t, unsigned int)'
console.c: In function `int get_cmd(FILE*, const char*, BSOCK*, int)':
console.c:705: warning: the address of `int tee(int, int, size_t, unsigned
int)', will always evaluate as `true'
console.c: In function `int teecmd(FILE*, BSOCK*)':
console.c:784: error: assignment of function `int tee(int, int, size_t,
unsigned int)'
console.c:784: error: cannot convert `bool' to `int ()(int, int, size_t,
unsigned int) throw ()' in assignment
console.c: In function `int outputcmd(FILE*, BSOCK*)':
console.c:791: error: assignment of function `int tee(int, int, size_t,
unsigned int)'
console.c:791: error: cannot convert `bool' to `int ()(int, int, size_t,
unsigned int) throw ()' in assignment
console.c: In function `int do_outputcmd(FILE*, BSOCK*)':
console.c:809: error: assignment of function `int tee(int, int, size_t,
unsigned int)'
console.c:809: error: cannot convert `bool' to `int ()(int, int, size_t,
unsigned int) throw ()' in assignment
console.c: In function `void sendit(const char*)':
console.c:893: warning: the address of `int tee(int, int, size_t, unsigned
int)', will always evaluate as `true'
console.c:896: warning: the address of `int tee(int, int, size_t, unsigned
int)', will always evaluate as `true'
make[1]: *** [console.o] Error 1
make[1]: *** Waiting for unfinished jobs....
make[1]: Leaving directory
`/var/tmp/portage/app-backup/bacula-1.38.9/work/bacula-1.38.9/src/console'


  ====== Error in
/var/tmp/portage/app-backup/bacula-1.38.9/work/bacula-1.38.9/src/console ======


Still, the build finishes "successfully", in the sense that ebuild returns 0
after ebuild bacula-1.38.9 compile.

Running make from my portage tmp dir, make returns 0 despite the failures.  So
the problem is definitely higher-up the food chain.  Unless we grepped for
Error messages, we're not going to notice bacula's build is failing.

One more thing -- my current guess is that the bacula-dir link is failing
because it doesn't add -lz

1) Confirmed that -lz on the director link command (run make in src/dird) fixes
the dird compile problem.  bacula appears to count on the python config script
to include all necessary link flags for python modules, but python isn't
including -lz.  I think bacula should fix this, because it would be weird for
python to include every dependency of every module installed.

2) Running make in src/dird returns error 2 when -lz is not specified.

3) Running make in the root compile directory returns 0, even on errors.  I
think this is because they use a bash for-loop in their "all" target, and the
errors aren't propogated back to make:

all: Makefile
        @for I in ${all_subdirs}; \
          do (cd $$I; echo "==>Entering directory `pwd`"; \
              $(MAKE) $@ || (echo ""; echo ""; echo "  ====== Error in `pwd`
======"; \
                            echo ""; echo "";)); \
        done

I didn't see any "dird" or "bacula-dir" targets in the main Makefile.  I tried
running "make dird" and "make bacula-dir" in case they were implicit somewhere,
but those didn't work either.  I had hoped we might be able to have the bacula
ebuild script build the main components separately, so we could catch the
errors they miss.

At any rate, this should probably be fixed in bacula, so everyone can reap the
benefits.  I'm not sure what the bacula folks would like done with their
for-loop and fancy messages, and I'm tired and want to go to bed.  So I'm
leaving this for now, hoping someone else will follow-up with a real patch.

One more thing -- you can fix the console compile by changing bool tee on line
78 to something like "bool mytee".  Then change all references to that boolean
to "mytee", and you are set.

After that, and linking manually with -lz in src/dird, all tools build
correctly.  Finishing with "ebuild bacula-1.38.9 merge" worked fine.

Note, I am getting the console problems with the "tee" variable because my
build is using

  /usr/include/gentoo-multilib/amd64/bits/fcntl.h

which already declares "tee".  So maybe only amd64 people with multilib
installations will hit this.  Again, I think the bacula folks should probably
change their variable name, since they are the higher-level consumer.

As to why Evil Compile Person didn't get an error with bacula-dir, maybe he or
she is using a different version of python, or has a somewhat different python
config script, or somehow doesn't have zlib enabled for bacula builds.

-Paul

I just hopped over to the bugs.bacula.org to report these problems (by
referencing this thread), and noticed that they don't even allow reporting for
version 1.38.9 in their version drop-down.  1.38.11 is the oldest they mention.

Maybe these are all known problems, fixed in later builds.  In that case,
perhaps the right solution is to bring gentoo's bacula package up-to-date.

I got a response back from the Bacula folks, on their bug 0000779 (see
http://bugs.bacula.org/, login as anonymous/anonymous to view bugs).  In short,

* The tee problem was corrected a while ago, certainly by 1.38.11.  It is also
fixed in later versions of fcntl.h
* The master-makefile not-returning-error problem still exists
* The link problem wasn't mentioned

In that bug, I asked whether what versions we (gentoo) should be using.  The
response was that we certainly should be at 1.38.11 or later, and 2.0 as soon
as possible.

If gentoo's bacula maintainer(s) decide against 1.38.11 and 2.x, then we should
at least patch the "bool tee" problem in console.c, and the "-lz" problem in
src/dird.  Personally, I'd rather see our time spent catching up with bacula,
than maintaining old versions by ourselves.

I had the same 'tee' problem and editing console.c fixed it. About updating the
ebuild, there's an open bug 130674 with an ebuild for 1.38.11.

Only 1.36.3-r2 and 2.0.3 left in the tree. If you can reproduce w/ 2.0.3, feel
free to reopen this bug.