anchor0057@yahoo.com wrote:
> Does anybody know why the default HP-UX install starts up the syslogd
> using the /sbin/init.d/syslogd script rather than as a respawning
> process from the /etc/inittab file? It would seem like this is the one
> process you want to have stay up when your system is getting whacked
> with a problem.

The respawn thingy in inittab is a feature which pre-dates the System
V Release 4 startup scripts. It was/is only used for spawning getty's,
etc.. Respawning things (like syslogd) which shouldn't terminate, is a
hack, not a fix.

I see. But, unfortunately, if it ever does fail it will probably do so
right at the moment when I most need to see the log messages. I'd be
happier if there were some other process making sure it was respawned.
(Similar to the way cimserverd will respawn cimserver.

--
Bob

In article <1138981374.415160.284680@g47g2000cwa.googlegroups. com>,
anchor0057@yahoo.com wrote:
> I see. But, unfortunately, if it ever does fail it will probably do so
> right at the moment when I most need to see the log messages. I'd be
> happier if there were some other process making sure it was respawned.
> (Similar to the way cimserverd will respawn cimserver.

But what happens if cimserverd dies? Also, why would expect syslogd to die when
you have a problem? If that were the case, why would you expect it to start
normally when re-spawned?

Kevin

--
Unix Guy Consulting, LLC
Unix and Linux Automation, Shell, Perl and CGI scripting
http://www.unix-guy.com

> But what happens if cimserverd dies?

Then I'd expect that cimserver wouldn't be respawned if it died. But
that would take two unlikely events to occur, rather than one.
Hopefully the event will get logged and eventually noticed by the
SysAdmin, who could restart cimserved. Meanwhile cimserver would
probably stay up...

> Also, why would expect syslogd to die when you have a problem?

Past experience and karma. Processes sometimes die hard; make due to a
bug, or whatever, and usually right when you least need it. *shrug*
Just a tendency toward caution, I guess.

> If that were the case, why would you expect it to start normally when re-spawned?

Why wouldn't I? Whatever killed the daemon could be a solitary event.

--
Bob

> Also, why would expect syslogd to die when you have a problem?

Or, in another instance, the syslogd could go down on a system and
nobody would notice for a while. Then, when a real problem hits,
nothing gets logged.

--
Bob

In article <1139001700.475628.263710@g14g2000cwa.googlegroups. com>,
anchor0057@yahoo.com wrote:
>> Also, why would expect syslogd to die when you have a problem?
>
> Or, in another instance, the syslogd could go down on a system and
> nobody would notice for a while. Then, when a real problem hits,
> nothing gets logged.

Then get a monitoring tool like ITO, Tivoli, Nagios, etc and monitor your
systems for that. That's what most people do...

Kevin
--
Unix Guy Consulting, LLC
Unix and Linux Automation, Shell, Perl and CGI scripting
http://www.unix-guy.com