tux0r, thanks! I can't remain alwoys online, but I read the transcript in log.omnios.org/omnios (I'm not an avid IRC user)

If I understood right, OmniOS has bot syslog and rsyslog, but only the latter is active, so I added a file in /etc/rsyslog.d/ with this line: "auth.notice /var/log/authlog" and restarted the rsyslog service, but the log file is still empty. What I'm missing?

^bot^both :)

warden: What messages are you expecting to see at the "auth.notice" level?

It looks like sshd emits publickey/disconnect messages at auth.info

Which is a lower severity than notice, so you'd have to specify "auth.info" instead

It looks like the default for for "LogLevel" in sshd_config is INFO, which matches up

I tried also matching auth.*, but nothing. But, hey! ... I just noticed that rsyslog was in "mainteinance mode since yesterday, because of a typo I entered in one of my tries! :S

so each time I launched "svcadm restart rsyslog" I was doing nothing actually... but the command didn't output any error

I just noticed this by looking in /var/svc/log/system-system-log\:rsyslog.log (and then with fmdump(8))

I think that fault management is a wonderful feature, but it didn't actively notified me of the problem, neither the output of svcadm(8).

In FreeBSD, if I try to restart a service that is down it immediately tells me "[servicename] not running? (check /var/run/[servicename].pid)"

I'm not saying that it's wrong in OmniOS, but I'd wish to know what I was missing... for example: maybe I should have configured the fault management system to notify the admin via email such type of problems?

One very useful command is `svcs -x` which shows any services which are in the degraded or maintenance states.

`svcs -xv` can be your friend

You can configure email notifications too, or snmp if I remember correctly.

ok, I'm reading fmd(8)... any switch to add to svcadm to sending something on stderr or interactive output when a stupid admin try to restart a service which is inactive or faulty?

well, maybe the response is in my own question.... I;m stupid because I'm still accustomized to the idea that a service has to be started/stopped, and not enabled/disabled! ;)

rc(8) is definitely a far less complex/featured than SMF(7)... :)

I don't know of any switch, but some kind of warning output is not a bad idea when retarting a maintenance server!

*service

andyf: it would be surely helpful for novices... thanks! :)

warden: If you do "svcadm enable -s system-log" it will emit a warning like: svcadm: Instance "svc:/system/system-log:default" is in maintenance state.

-s means "synchronous", or "wait for it to reach the online state"

Probably "svcadm refresh" and "svcadm restart" should print a similar warning if something is in the maintenance state. I'll file a bug