[syslog-ng] Solaris 10 info

Roberto Nibali ratz at tac.ch
Mon Jun 6 08:56:16 CEST 2005


Hello Dave,

I've just, for the sake of completeness and reduced redundancy, browsed through
the http://www.sunfreeware.com/ site and noticed that Steve Christensen seems to
be packaging syslog-ng for Solaris already. Steve, would you care to read
through the two previous messages of this thread in the syslog-ng archive below,
please?

https://lists.balabit.hu/pipermail/syslog-ng/2005-June/007525.html

Is there anything you could use to create a Solaris 10 package, including
application handling through the SMF facility?

I think if Bazsi does not mind, the additional files for Solaris could be added
to the syslog-ng package.

> A couple notes--
> * If one changes the --cfgfile option, then the svc.method
> file should be updated as well (it checks for the config file)...

True.

> * We use the convention /etc/syslog-ng/ -> /usr/local/etc here;
> If my memory serves correctly, I ran into a small issue between
> the builds (maybe it was sunfreeware and a compiled build?).

I think I know what you mean. The configure switch "--sysconfdir=" does not work
as expected. We have the following line in our top level Makefile above
syslog-ng (to switch from /etc/syslog-ng to /etc):

        echo "#undef PATH_SYSCONFDIR"                   >> src/src/config.h
        echo "#define PATH_SYSCONFDIR \"/etc/\""        >> src/src/config.h

Since configure is platform-independent I reckon this issue remains for a
Solaris compilation. I remember that recently someone sent a patch which however
was rejected by the master himself.

> In any event, the link was an easy workaround to make sure
> syslog-ng starts up (with a config file).

Agreed.

> * The other thing is, logrotation may need to be updated (logadm.conf) :
> /var/log/syslog -C 8 -P 'Sat May 28 08:10:00 2005' -a 'kill -HUP `cat
> /var/run/syslog.pid`'
> /var/adm/messages -C 4 -P 'Sat May 28 08:10:00 2005' -a 'kill -HUP `cat
> /var/run/syslog.pid`'
> to
> /var/log/syslog -C 8 -P 'Sat May 28 08:10:00 2005' -a '/usr/sbin/svcadm
> refresh syslog-ng`'
> /var/adm/messages -C 4 -P 'Sat May 28 08:10:00 2005' -a
> '/usr/sbin/svcadm refresh syslog-ng`'

Right.

> -----
> I'm not sure why Sun doesn't use svcadm to -HUP syslog...

<irony>
Maybe they don't trust svcadm ;) Or the change request is still in the testing
phase and will be ready for Solaris 11 with a proper change management mechanism
for all Solaris customers.
</sarcasm>

> It should be noted, that feature with smf (as Sun touts as automatic
> healing) is that if you
> `pkill -9 syslog-ng` smf will :respawn: a new syslog-ng process.

Yes, such a watchdog feature (the word I prefer) should be part of any
Unix/Linux OS. Debian has a similar thing, IIRC. It would technically also be
dead easy to implement. All you have to do in 99% of the cases is to have a
crontab entry which restarts the daemon every minute (yes, I'm very well aware
of the implications of this approach, but for 99% of the cases this works
wonders). A well written start script has proper application control and
basically exits again if the daemon is running or, in case of application
failure, respawns the daemon.

I hope we get someone to update the package to include our findings in this
thread. Personally I don't have the time to do this.

Best regards,
Roberto Nibali, ratz
-- 
-------------------------------------------------------------
addr://Rathausgasse 31, CH-5001 Aarau  tel://++41 62 823 9355
http://www.terreactive.com             fax://++41 62 823 9356
-------------------------------------------------------------
terreActive AG                       Wir sichern Ihren Erfolg
-------------------------------------------------------------


More information about the syslog-ng mailing list