The Samba-Bugzilla – Bug 5594
"make test" before "make install" fails
Last modified: 2009-05-17 17:53:12 UTC
With Samba 3.0.x, I would:
3. make test
4. make install (if all went well)
When I do a 'make test' with Samba 3.2.0, though, I eventually run
into problems like this:
RUNNING TESTS testparm_s3
Testing netbios name (0)
Load smb config files from
params.c:Parameter() - Ignoring badly formed line in configuration file: netbios
map_file: Failed to load /usr/local/samba/samba-3.2.0/lib/valid.dat -
No such file or directory
creating default valid table
Loaded services file OK.
ERROR: lock directory /usr/local/samba/samba-3.2.0/var/locks does not exist
ERROR: pid directory /usr/local/samba/samba-3.2.0/var/locks does not exist
So it's no longer possible to do a make test in-place before installing the binary.
Michael Adam posted this to the samba mailing list in response to my report:
Oh, thanks for reporting.
Metze has also pointed this one out to me.
As you noted, it will work without complaints when you
do "make install" before "make test".
It is a bug and I am going to fix it.
The problem is that these new testparm_s3 tests call the testparm
binary, which uses the compiled in defaults for lock dir and so
on to set up locking and logging and so on. I did not notice
these problems in time since I only ran make test from a build
dir where I had previously done make install.
If you care, you might open a bugzilla entry (assigned to
email@example.com) or/and of course provide a patch.
The plan would be to extend testparm so that one can completely
bypass the lockdir etc from compiled in defaults.
Or rewrite the testparm_s3 tests in C to not call testparm...
I need to think about that....
Cheers - Michael
thanks for reporting.
Created attachment 3407 [details]
fix from v3-2-test
I have fixed this in v3-2-test (and v3-3-test)
by adding a --skip-global-checks parameter to testparm
and using it in the test scripts.
Please try the attached patch against the 3.2.0 tarball.
The first part of the patch (for test_testparm_s3.sh) applied correctly, but it wasn't happy with the testparm.c patch (this against the 3.2.0 tarball). I was able to apply it manually, though, and on first blush it seems to have fixed the problem. Now I need to figure out why smbd is crashing ...
This is fixed.
Closing the bug report...
Please reopen if problem reoccurs.