Mailing List Archive

rsyslog 3.11.5 released
Hi all,


Rsyslog 3.11.5 has been released today. It is a minor feature upgrade
release. Most importantly, the GSSAPI input module has become an
independent module. Formerly, the rsyslog core needed to be linked with
GSSAPI libraries, which now no longer is required. Also, there now is an
initial implementation of the pre-v3 compatibility layer and duplicated
$ModLoad's are now detected and prevented. Release 3.11.5 is a
recommended update for all v3 users.

Changelog:

http://www.rsyslog.com/Article181.phtml

Download:

http://www.rsyslog.com/Downloads-req-getit-lid-82.phtml


Best regards,

Florian Riedl
rsyslog 3.11.5 released [ In reply to ]
Hi

2008/2/25, Florian Riedl <friedl at hq.adiscon.com>:
> release. Most importantly, the GSSAPI input module has become an
> independent module. Formerly, the rsyslog core needed to be linked with
> GSSAPI libraries, which now no longer is required. Also, there now is an

Hm, rsyslogd_LDADD in Makefile.am still has $(gss_libs).
So when you enable the GSSAPI support, rsyslogd will be linked against
the GSSAPI libraries. Has it been forgotten to remove $(gss_libs)
while modularizing gssapi?

Cheers,
Michael
--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
rsyslog 3.11.5 released [ In reply to ]
2008/2/25, Michael Biebl <mbiebl at gmail.com>:
> Hi
>
> 2008/2/25, Florian Riedl <friedl at hq.adiscon.com>:
>
> > release. Most importantly, the GSSAPI input module has become an
> > independent module. Formerly, the rsyslog core needed to be linked with
> > GSSAPI libraries, which now no longer is required. Also, there now is an
>
>
> Hm, rsyslogd_LDADD in Makefile.am still has $(gss_libs).
> So when you enable the GSSAPI support, rsyslogd will be linked against
> the GSSAPI libraries. Has it been forgotten to remove $(gss_libs)
> while modularizing gssapi?

Some more comments:
plugins/imtcp/Makefile.am
builds imgssapi uncoditionally. Shouldn't imgssapi not be surrounded by
if ENABLE_GSSAPI
imgssapi_*
...
endif

Also, rsyslogd_SOURCES still contains gss_misc.* and syslogd.c and
net.c also have #ifdef USE_GSSAPI code. So it's not yet really
possible to build the GSSAPI support in a separate module without
linking the rsyslogd core against the GSSAPI libs.
The changelog is thus a bit misleading.

Cheers,
Michael

--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
rsyslog 3.11.5 released [ In reply to ]
Hi Michael,

thanks for the feedback. I have probably misunderstood the change, I
will talk with the contributors. For the time being, please all consider
this to be a bug.

Rainer

> -----Original Message-----
> From: rsyslog-bounces at lists.adiscon.com [mailto:rsyslog-
> bounces at lists.adiscon.com] On Behalf Of Michael Biebl
> Sent: Monday, February 25, 2008 3:51 PM
> To: rsyslog-users
> Subject: Re: [rsyslog] rsyslog 3.11.5 released
>
> 2008/2/25, Michael Biebl <mbiebl at gmail.com>:
> > Hi
> >
> > 2008/2/25, Florian Riedl <friedl at hq.adiscon.com>:
> >
> > > release. Most importantly, the GSSAPI input module has become an
> > > independent module. Formerly, the rsyslog core needed to be
> linked with
> > > GSSAPI libraries, which now no longer is required. Also, there
> now is an
> >
> >
> > Hm, rsyslogd_LDADD in Makefile.am still has $(gss_libs).
> > So when you enable the GSSAPI support, rsyslogd will be linked
> against
> > the GSSAPI libraries. Has it been forgotten to remove $(gss_libs)
> > while modularizing gssapi?
>
> Some more comments:
> plugins/imtcp/Makefile.am
> builds imgssapi uncoditionally. Shouldn't imgssapi not be surrounded
by
> if ENABLE_GSSAPI
> imgssapi_*
> ...
> endif
>
> Also, rsyslogd_SOURCES still contains gss_misc.* and syslogd.c and
> net.c also have #ifdef USE_GSSAPI code. So it's not yet really
> possible to build the GSSAPI support in a separate module without
> linking the rsyslogd core against the GSSAPI libs.
> The changelog is thus a bit misleading.
>
> Cheers,
> Michael
>
> --
> Why is it that all of the instruments seeking intelligent life in the
> universe are pointed away from Earth?
> _______________________________________________
> rsyslog mailing list
> http://lists.adiscon.net/mailman/listinfo/rsyslog
rsyslog 3.11.5 released [ In reply to ]
2008/2/25, Rainer Gerhards <rgerhards at hq.adiscon.com>:
> Hi Michael,
>
> thanks for the feedback. I have probably misunderstood the change, I
> will talk with the contributors. For the time being, please all consider
> this to be a bug.

One more question:
If you enable gssapi support (which is built from imtcp.c), can you
actually load both imtcp *and* imgssapi at the same time. Imo it will
lead to conflicts.

Cheers,
Michael

--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
rsyslog 3.11.5 released [ In reply to ]
> One more question:
> If you enable gssapi support (which is built from imtcp.c), can you
> actually load both imtcp *and* imgssapi at the same time. Imo it will
> lead to conflicts.

Yes, you are right. This is documented (at least I hope one can
understand it ;)) and is currently not supported. To fix it, we need the
rsyslog loader (just like with the generic database action). So it is a
temporary quirk.

Rainer
rsyslog 3.11.5 released [ In reply to ]
On 02/25/2008 03:51 PM, Michael Biebl wrote:
> 2008/2/25, Michael Biebl <mbiebl at gmail.com>:
>> Hi
>>
>> 2008/2/25, Florian Riedl <friedl at hq.adiscon.com>:
>>
>>> release. Most importantly, the GSSAPI input module has become an
>> > independent module. Formerly, the rsyslog core needed to be linked with
>> > GSSAPI libraries, which now no longer is required. Also, there now is an
>>
>>
>> Hm, rsyslogd_LDADD in Makefile.am still has $(gss_libs).
>> So when you enable the GSSAPI support, rsyslogd will be linked against
>> the GSSAPI libraries. Has it been forgotten to remove $(gss_libs)
>> while modularizing gssapi?
>
> Some more comments:
> plugins/imtcp/Makefile.am
> builds imgssapi uncoditionally. Shouldn't imgssapi not be surrounded by
> if ENABLE_GSSAPI
> imgssapi_*
> ...
> endif
>
> Also, rsyslogd_SOURCES still contains gss_misc.* and syslogd.c and
> net.c also have #ifdef USE_GSSAPI code. So it's not yet really
> possible to build the GSSAPI support in a separate module without
> linking the rsyslogd core against the GSSAPI libs.
> The changelog is thus a bit misleading.
>
> Cheers,
> Michael

Hi Michael,

thanks for the comments, attached patch should resolve these problems.
http://pastebin.com/m638295ac

Regarding the #ifdefs in syslogd.c and net.c, they are used to
conditionally compile some options, allowed senders list and such, no
external GSSAPI code is needed so really only the module needs to be
linked with the GSSAPI libs.
rsyslog 3.11.5 released [ In reply to ]
2008/2/25, theinric at redhat.com <theinric at redhat.com>:
> On 02/25/2008 03:51 PM, Michael Biebl wrote:
> > 2008/2/25, Michael Biebl <mbiebl at gmail.com>:
> >> Hi
> >>
> >> 2008/2/25, Florian Riedl <friedl at hq.adiscon.com>:
> >>
> >>> release. Most importantly, the GSSAPI input module has become an
> >> > independent module. Formerly, the rsyslog core needed to be linked with
> >> > GSSAPI libraries, which now no longer is required. Also, there now is an
> >>
> >>
> >> Hm, rsyslogd_LDADD in Makefile.am still has $(gss_libs).
> >> So when you enable the GSSAPI support, rsyslogd will be linked against
> >> the GSSAPI libraries. Has it been forgotten to remove $(gss_libs)
> >> while modularizing gssapi?
> >
> > Some more comments:
> > plugins/imtcp/Makefile.am
> > builds imgssapi uncoditionally. Shouldn't imgssapi not be surrounded by
> > if ENABLE_GSSAPI
> > imgssapi_*
> > ...
> > endif
> >
> > Also, rsyslogd_SOURCES still contains gss_misc.* and syslogd.c and
> > net.c also have #ifdef USE_GSSAPI code. So it's not yet really
> > possible to build the GSSAPI support in a separate module without
> > linking the rsyslogd core against the GSSAPI libs.
> > The changelog is thus a bit misleading.
> >
> > Cheers,
> > Michael
>
>
> Hi Michael,
>
> thanks for the comments, attached patch should resolve these problems.
> http://pastebin.com/m638295ac
>
> Regarding the #ifdefs in syslogd.c and net.c, they are used to
> conditionally compile some options, allowed senders list and such, no
> external GSSAPI code is needed so really only the module needs to be
> linked with the GSSAPI libs.
>

Funny. I basically ended up with the same patch, which I intended to
post just a few moments after your email.
I initially thought of compiling gss-misc into a convenience libtool
library and link imgssapi and omgssapi against it (to avoid the double
compilation of gss-misc.c). But then I decided against it, because it
was not really worth the effort.

One minor issue with your proposed patch: gss-misc.h is not added to
the dist tarball.
You have 3 options here:
1.) Include it in EXTRA_DIST in Makefile.am
2.) Include it in _SOURCES in plugins/(imtcp,omgssapi)/Makefile.am
3.) Keep it in rsyslogd_SOURCES (ugly)
4.) Add noinst_HEADERS = gss-misc.h to Makefile.am

I'd prefer 4.

Cheers,
Michael

--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
rsyslog 3.11.5 released [ In reply to ]
2008/2/25, Michael Biebl <mbiebl at gmail.com>:
> You have 3 options here:
>
> I'd prefer 4.

Argh, it would help if I could actually count ;-)

Cheers,
Michael
--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
rsyslog 3.11.5 released [ In reply to ]
2008/2/25, Michael Biebl <mbiebl at gmail.com>:
> I initially thought of compiling gss-misc into a convenience libtool
> library and link imgssapi and omgssapi against it (to avoid the double
> compilation of gss-misc.c).

FWIW, the patch for the covenience library is at

http://pastebin.com/m638295ac

As I had already done the work, I thought I'd share it anyways.

Rainer, take the one which you like better.

Cheers,
Michael
--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
rsyslog 3.11.5 released [ In reply to ]
2008/2/26, Michael Biebl <mbiebl at gmail.com>:
> 2008/2/25, Michael Biebl <mbiebl at gmail.com>:
>
> > I initially thought of compiling gss-misc into a convenience libtool
> > library and link imgssapi and omgssapi against it (to avoid the double
> > compilation of gss-misc.c).
>
>
> FWIW, the patch for the covenience library is at
>
> http://pastebin.com/m638295ac

Oops, cut&paste error. The correct link is

http://pastebin.com/mcdd2846

Michael
--
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?
rsyslog 3.11.5 released [ In reply to ]
That was a tough decision ;) I finally took Michael's patch, as it is
pointing into the same direction that the rsyslog loader is heading...
Now in CVS. I think I'll release a bugfix version later today, but I
will also work a bit on the doc, which also is sub-optimal in 3.11.5...

Rainer

> -----Original Message-----
> From: rsyslog-bounces at lists.adiscon.com [mailto:rsyslog-
> bounces at lists.adiscon.com] On Behalf Of Michael Biebl
> Sent: Tuesday, February 26, 2008 2:37 AM
> To: rsyslog-users
> Subject: Re: [rsyslog] rsyslog 3.11.5 released
>
> 2008/2/26, Michael Biebl <mbiebl at gmail.com>:
> > 2008/2/25, Michael Biebl <mbiebl at gmail.com>:
> >
> > > I initially thought of compiling gss-misc into a convenience
> libtool
> > > library and link imgssapi and omgssapi against it (to avoid the
> double
> > > compilation of gss-misc.c).
> >
> >
> > FWIW, the patch for the covenience library is at
> >
> > http://pastebin.com/m638295ac
>
> Oops, cut&paste error. The correct link is
>
> http://pastebin.com/mcdd2846
>
> Michael
> --
> Why is it that all of the instruments seeking intelligent life in the
> universe are pointed away from Earth?
> _______________________________________________
> rsyslog mailing list
> http://lists.adiscon.net/mailman/listinfo/rsyslog