[ANNOUNCE] ulogd 2.0.0beta3

[ANNOUNCE] ulogd 2.0.0beta3

[Pablo Neira Ayuso]

[-- Attachment #1: Type: text/plain, Size: 736 bytes --]

Hi!

The netfilter project proudly presents:

  * ulogd 2.0.0beta3

ulogd is a userspace logging daemon for netfilter/iptables related
logging. This includes per-packet logging of security violations,
per-packet logging for accounting purpose as well as per-flow logging.

ulogd comes in two flavours: ulogd-1.x has been around since 2000 and is
the stable series. All production systems should use the stable series
at this time. The 1.x series are frozen, do not expect new features into
this line. ulogd-2.x is currently in beta stage.

You can download them from:

http://www.netfilter.org/projects/ulogd/downloads.html
ftp://ftp.netfilter.org/pub/ulogd/

Regards,
Pablo

-- 
"Los honestos son inadaptados sociales" -- Les Luthiers

[-- Attachment #2: ChangeLog-ulogd-2.0.0beta3 --]
[-- Type: text/plain, Size: 2555 bytes --]

Eric Leblond (32):
      NFLOG: remove unused option
      doc: update ulogd SGML documentation
      doc: update author credits and fix reference to closed ML
      config: remove obsolete global variables
      Modify usage of nflog_bind_pf function.
      Get rid of addressfamily variable in NFLOG input plugin
      Document group 0 usage and suppress address_family
      Add SCTP support to BASE plugin.
      SCTP support for PRINTPKT.
      Add SCTP support to MySQL and PGSQL output.
      Treat nice function return.
      Fix stop function of NFCT plugin.
      Don't free pluginstance when leaving
      Fix minor memory leak in NFLOG plugin.
      Call pluginstance stop function when exiting
      Add SIGINT to list of terminal signal.
      Unload plugins when quitting.
      Introduce config_stop() function
      Free stacks when exiting.
      Fix memory leak in destructor_nfct().
      Add valgrind compilation option.
      Add variable to force binding of nfnetlink_log.
      fix config file: MAC2STR has been renamed to HWHDR.
      Display logfile to check in case of error.
      Fix crash on IPv6 packet.
      HWHDR: Fix size of allocated string.
      Replace INCLUDES by AM_CPPFLAGS in Makefile.am.
      Display which keys are optional in info mode.
      Explicitely type PCAP input key.
      Add threshold and timeout option to NFLOG plugin.
      pgsql: fix type of mac_protocol and oob_protocol.
      Delete timer in destructor function.

Jan Engelhardt (5):
      build: compile fix
      Update .gitignore and remove install-sh
      build: use -avoid-version for modules
      build: resolve autotools suggestions to use AC_CONFIG_MACRO_DIR
      build: update configure.ac

Pablo Neira Ayuso (4):
      doc: update ulogd man-page
      NFCT: remove unused constant NFLOG_RMEM_DEFAULT
      add ukey_* function for key assignation
      build: bump version to 2.00beta3

Pierre Chifflier (13):
      hwhdr: fix segfault when RAW_MAC is NULL
      hwhdr: finish missing renaming
      pgsql: remove useless checkings
      Fix light memory error in parse_mac2str
      Link ulogd2 with libpthread
      Search for libdbi includes during configure
      Add new output plugin DBI
      PostgreSQL schema: drop useless constraints
      Allow plain INSERT instead of procedure
      DBI: lower column name before comparing to key
      PostgreSQL: allow local connections
      Flat SQL schema for PostgreSQL
      Flat SQL schema for MySQL

Thomas Jacob (1):
      Return true/false instead of ULOGD_IRET_OK/STOP

Re: [ANNOUNCE] ulogd 2.0.0beta3

[Harald Welte]

Hi guys,

congratulations on the new beta release.

On Fri, Mar 06, 2009 at 07:09:36PM +0100, Pablo Neira Ayuso wrote:
>       Link ulogd2 with libpthread

I was just reading this and all alarm sirens went on.  I don't know
why that is neccessary, and I hope it is just due to some library dependency.

When I created ulogd and ulogd2, I never intended them to become some kind of
multi-threaded monster with a complex locking and synchronization nightmare.

Sure, I'm no longer maintaining it, and I respect the decision of current
maintainers, but I just wanted to point that out.

-- 
- Harald Welte <laforge@netfilter.org>                 http://netfilter.org/
============================================================================
  "Fragmentation is like classful addressing -- an interesting early
   architectural error that shows how much experimentation was going
   on while IP was being designed."                    -- Paul Vixie

Re: [ANNOUNCE] ulogd 2.0.0beta3

[Jan Engelhardt]

On Saturday 2009-03-07 11:34, Harald Welte wrote:
>On Fri, Mar 06, 2009 at 07:09:36PM +0100, Pablo Neira Ayuso wrote:
>>       Link ulogd2 with libpthread
>
>I was just reading this and all alarm sirens went on.  I don't know
>why that is neccessary, and I hope it is just due to some library dependency.
>
>When I created ulogd and ulogd2, I never intended them to become some kind of
>multi-threaded monster with a complex locking and synchronization nightmare.

Yeah there seems to be not a single pthread_ call or type in use,
so I wonder why it was done.

Re: [ANNOUNCE] ulogd 2.0.0beta3

[Pierre Chifflier]

On Sat, Mar 07, 2009 at 03:43:47PM +0100, Jan Engelhardt wrote:
> 
> On Saturday 2009-03-07 11:34, Harald Welte wrote:
> >On Fri, Mar 06, 2009 at 07:09:36PM +0100, Pablo Neira Ayuso wrote:
> >>       Link ulogd2 with libpthread
> >
> >I was just reading this and all alarm sirens went on.  I don't know
> >why that is neccessary, and I hope it is just due to some library dependency.

Hi,

I plead guilty, I'm the one who did the commit ;)

The commit message explains why:
Explicitly link with libpthread. This allows to run ulogd within gdb,
else it fails with message: Cannot find new threads: generic error

I don't know why, but gdb fails on some configs if not linked with
libpthread. Maybe this is related to some other parameters (32/64 bits ?
libc version ?), yet adding -lpthread solves the problem.
One could argue that gdb should be fixed, though.

> >
> >When I created ulogd and ulogd2, I never intended them to become some kind of
> >multi-threaded monster with a complex locking and synchronization nightmare.

Yep, we did not change anything regarding to that (and share the same
view for the multi-thread beast).

> 
> Yeah there seems to be not a single pthread_ call or type in use,
> so I wonder why it was done.

Threads are not used, the -lpthread could even be dropped without
any other effect than causing problems again with gdb.

Pierre

Re: [ANNOUNCE] ulogd 2.0.0beta3

[Pablo Neira Ayuso]

Pierre Chifflier wrote:
> On Sat, Mar 07, 2009 at 03:43:47PM +0100, Jan Engelhardt wrote:
>> Yeah there seems to be not a single pthread_ call or type in use,
>> so I wonder why it was done.
> 
> Threads are not used, the -lpthread could even be dropped without
> any other effect than causing problems again with gdb.

Then, please investigate the reason that causes this problem and remove
the libpthread dependency. This also seem hackish to me (it's a patch
that you applied because your setup complained about missing pthread).

-- 
"Los honestos son inadaptados sociales" -- Les Luthiers

Re: [ANNOUNCE] ulogd 2.0.0beta3

[Harald Welte]

Ok, thanks for the explanation. seems fine to me.

Now the only question is whether this wastes (significant) resources (i.e.
resident set size or anything) in case ulogd2 runs on a small embedded box.

If it is debugging-only, one could also enable it only for a debug-build (when
-g / -pg is set)

regards,
-- 
- Harald Welte <laforge@netfilter.org>                 http://netfilter.org/
============================================================================
  "Fragmentation is like classful addressing -- an interesting early
   architectural error that shows how much experimentation was going
   on while IP was being designed."                    -- Paul Vixie

Re: [ANNOUNCE] ulogd 2.0.0beta3

[Pablo Neira Ayuso]

Harald Welte wrote:
> Hi guys,
> 
> congratulations on the new beta release.
> 
> On Fri, Mar 06, 2009 at 07:09:36PM +0100, Pablo Neira Ayuso wrote:
>>       Link ulogd2 with libpthread
> 
> I was just reading this and all alarm sirens went on.  I don't know
> why that is neccessary, and I hope it is just due to some library dependency.
> 
> When I created ulogd and ulogd2, I never intended them to become some kind of
> multi-threaded monster with a complex locking and synchronization nightmare.

Please, don't worry, I will not allow such a crappy design change :)

The conntrack-tools follows the sample principle. I don't know if you
had the time to look at the code, but for example, we removed another
tricky issue from the synchronization point of view time ago which was
the use of SIGALARM to implement the timer framework. So this whole
thing now follows an event-driven design with a synchrnous alarm
framework to avoid complexity as expected.

> Sure, I'm no longer maintaining it, and I respect the decision of current
> maintainers, but I just wanted to point that out.

As said, it seems that the patch was applied by Pierre (via Eric) due to
some very kiddish reason, please, set alarms off :).

-- 
"Los honestos son inadaptados sociales" -- Les Luthiers