Improving Samba 3 automated testing

Sandeep Cashyap sandy4samba at
Tue Mar 14 08:40:55 GMT 2006


In realtion to improving Samba3 automated testing, I downloaded and
installed Samba versions 3.0.21c and 4.0.0tp1 and modified the
script under samba-3.0.21c/source/script/tests/ , so that 'make test' picks
up smbtorture4 executable and ran the test twice.  However, on both
occasions it failed to create the TDB password and the smbtorture failed to
pass with every test getting the NT_STATUS_HOST_UNREACHABLE error.

/samba/samba-3.0.21c/source # make test
Using FLAGS =  -I/usr/include/heimdal -O -D_SAMBA_BUILD_  -Iinclude
-I/root/samba/samba- 3.0.21c/source/tdb  -I. -I/usr/include/heimdal
-I/root/samba/samba-3.0.21c/source -D_SAMBA_BUILD_
      LIBS = -lcrypt -lresolv -lnsl -ldl
      LDSHFLAGS = -shared -Wl,-Bsymbolic  -Wl,-rpath,/usr/lib
      LDFLAGS = -Wl,-rpath,/usr/lib
      PIE_CFLAGS =
Compiling torture/torture.c
Compiling torture/nbio.c
Compiling torture/scanner.c
Compiling torture/utable.c
Compiling torture/denytest.c
Compiling torture/mangle_test.c
Linking bin/smbtorture
Compiling torture/msgtest.c
Linking bin/msgtest
Compiling torture/masktest.c
Linking bin/masktest
Compiling torture/locktest.c
Linking bin/locktest
Compiling torture/locktest2.c
Linking bin/locktest2
Compiling torture/nsstest.c
Linking bin/nsstest
Compiling torture/cmd_vfs.c
Compiling torture/vfstest.c
Linking bin/vfstest
Running Samba 3 Test suite
Unable to open/create TDB passwd
Unable to open/create TDB passwd
pdb_getsampwnam: TDB passwd (/root/samba/samba-3.0.21c
/source/t_dir/private/passdb.tdb) did not exist. File successfully created.
./script/tests/ line 99: 22937 Done                    ( echo
     22938 Segmentation fault      | smbpasswd -c $LIBDIR/smb.conf -L -s -a

>>>>>> Starting test driver <<<<<
Starting smbd....
session setup failed: Call timed out: server did not respond after 20000
Shutting down smbd (pid 22949)...
make: *** [test] Interrupt

Also, when I see the list of smb processes running,  it shows all smbtorture
processes to be 'defunct'.
# ps -ae| grep smb
23015 ?        00:00:00 smbd
23017 ?        00:00:00 smbd
23020 pts/0    00:00:20 smbtorture
23021 pts/0    00:00:00 smbtorture <defunct>
23022 pts/0    00:00:00 smbtorture <defunct>
23023 pts/0    00:00:00 smbtorture <defunct>
23024 pts/0    00:00:00 smbtorture <defunct>
23025 pts/0    00:00:00 smbtorture <defunct>
23026 pts/0    00:00:00 smbtorture <defunct>
23027 pts/0    00:00:00 smbtorture <defunct>
23028 pts/0    00:00:00 smbtorture <defunct>
23029 pts/0    00:00:00 smbtorture <defunct>
23030 pts/0    00:00:00 smbtorture <defunct>
23031 pts/0    00:00:00 smbtorture <defunct>
23032 pts/0    00:00:00 smbtorture <defunct>

Is this an expected behaviour?

Thanks and regards
Sandeep Cashyap.R

IBM India Ltd.
Subramanya Arcade - 2, Level 6,
Banerghatta Main Road,
Bangalore - 560029.
Ph:  91-80-51956573
Email:  sandeep.cashyap at

On 3/3/06, Volker Lendecke <Volker.Lendecke at> wrote:
> On Fri, Mar 03, 2006 at 05:43:05PM +0300, Alexander Bokovoy wrote:
> > Should we rewrite smbtorture tests in ejs? This would allow arbitrary
> > grouping (through include statements) and simpler integration.
> The main argument against rewriting in ejs is the amount of
> coding that has gone into the torture tests so far. If
> someone went through and rewrote all of that in ejs, fine
> :-)
> Volker

More information about the samba-technical mailing list