[syslog-ng] syslog-ng maintenance changes

[Evan Rempel]

Can you please also fix bug 159

https://bugzilla.balabit.com/show_bug.cgi?id=159

It causes me pain almost weekly, and soon to be hourly as we
automate the production of the patterndb from a SQL database engine
which will result in a HUP very often.

Evan.

Gergely Nagy wrote:
> Hi!
> 
> Balazs Scheidler <bazsi at balabit.hu> writes:
> 
>> After talking to him in real life, we've decided to make a shift, he
>> is going to maintain the 3.3 tree and making releases instead of
>> myself. This will make 3.3 maintenance smoother, and free some
>> precious time for me, which I'd use to work on the 3.4 features.
> 
> I'd like to take this opportunity to outline my plans with regards to
> 3.3, and provide a short roadmap. There won't be many changes, as 3.3 is
> a stable release afterall, and I want to keep it that way. The latest
> 3.3.5 release was already almost perfect, and by far the best 3.3
> release so far, I'd like to make 3.3.6 top that.
> 
> As of now, the following items are on the 3.3.6 roadmap:
> 
>  [X] systemd enhancements:
>      [X] Correctly alias syslog-ng.service to syslog.service
>      [X] Teach the system() source about systemd, so configs using
>      system() will work on both systemd-enabled systems, and traditional
>      ones as well, without change.
>      [X] Smooth transition from /dev/log to journal: if systemd is in
>      use, and the config uses /dev/log, translate that to
>      /run/systemd/journal/syslog. This will go away post-3.4!
>      [ ] Include man page and admin guide references in the
>      Documentation= section of the unit file.
>  [X] Ignore SIGHUP in the supervisor [#143]
>  [X] Don't use deprecated glib symbols: a purely cosmetic change, to
>      reduce the number of compile-time warnings on modern systems.
>  [X] Restore fsync(yes) functionality [#175]
>  [X] README in spec, but not in source [#180]
>  [ ] Fix unix-dgram's handling of charset decoding errors [#181]
>  [ ] Fix spoofed source on Solaris [#178]
>  [ ] Fix ivykis-related run-time error on mips64 [#176]
> 
>  [#143]: https://bugzilla.balabit.com/show_bug.cgi?id=143
>  [#175]: https://bugzilla.balabit.com/show_bug.cgi?id=175
>  [#181]: https://bugzilla.balabit.com/show_bug.cgi?id=181
>  [#178]: https://bugzilla.balabit.com/show_bug.cgi?id=178
>  [#176]: https://bugzilla.balabit.com/show_bug.cgi?id=176
>  [#180]: https://bugzilla.balabit.com/show_bug.cgi?id=180
> 
> Apart from these, there's one major thing I want to change, which will -
> by default - not affect anyone, but I have hopes that it will prove to
> be beneficial in the future (and as a side effect, also fixes #176):
> starting with syslog-ng 3.3.6, along with the tarball, a patch file will
> be provided too, that makes it possible to use upstream ivykis with
> syslog-ng, instead of the embedded, forked copy it ships with now.
> 
> This means that distributions will be able to compile syslog-ng against
> a packaged ivykis, and any bugs found therein, will be easier to fix,
> and any improvements made to ivykis will also benefit syslog-ng
> immediately.
> 
> This also paves the way to switching to upstream ivykis by default at a
> later point in time - but that switch is out of the scope of a stable
> series, so it can only be done in 3.4 soonest. Nevertheless, an optional
> patch to make this easy is a worth the effort, I believe.
> 
> And that's all there is, the only remaining question is: when will 3.3.6
> land? I'm a firm believer of the Debianesque "when it is ready" school,
> so as soon as the list above has all boxes ticked, and there are no new
> issues to solve, I'll hit the release button.
> 
> I expect that to happen in a couple of weeks, mid-June or so, if all
> goes well. That way we might even have 3.3.6 in Debian Wheezy, using
> upstream ivykis, which would make me *very* happy.
> 


-- 
Evan Rempel                               erempel at uvic.ca
Senior Systems Administrator                 250.721.7691
Unix Services, University Systems, University of Victoria