Unknown reasons for shutdown

[sstarer]

I have an Axil (Sun Clone) running Solaris 6 with a Sybase SQL server on it and an external RAID. It periodically shuts down to the boot prompt for no apparent reason. Nothing in /var/adm/messages at all that indicates any issues. It does not have a frame buffer, keyboard, mouse or monitor attached. No window system is running. I use ttya attached to a com port on my PC to act as a console, but that is not always connected. I have a pinger script hitting the box every 5 seconds for about 20 minutes without issue, but the pinger itself may perturb what I'm checking. I did have an issue with powerd not working and resolved that, courtesy of a hint provided me by this forum, btw. Now my power.conf file is identical to a power.conf file on a machine with no shutdown issues. My experience with HP-UX is that it logs all shutdowns in /etc/shutdownlog. Besides figuring out what is going on, my questions are, is there any place that Solaris logs reasons for shutting down, or is it possible to implement one. Many thanks to all for taking the time.
Regards,
Steve Starer

 

[cfsalmeida]

Do you have "autoshutdown" enable ? ...

Regards,

Carlos Almeida,

 

[sstarer]

Carlos,
Thanks so much for getting back to me. I checked /etc/power.conf, and the last row is:
autoshutdown 30 9:00 9:00 noshutdown

It's on one line, with no continuation. My background is HP-UX, and we had no autoshutdown on the servers I managed. So I'm kind of stumped as to where to look to see if autoshutdown is enabled, if this data doesn't provide the answer.

I can tell you I've been pinging the box every 30 minutes for 4 days, and it's stayed up. I advanced the ping time to 60 minutes today. And I pulled the serial connection sitting on ttya at 2:40 PM. If you can tell me where else to check to see if autoshutdown is enabled, I'll check it out.

Thank you again for your help.
Regards,
Steve Starer

 

[lfoata]

Look for something that is similar to the following line in
/etc/inittab :
inittab3:s1234owerfail:/usr/sbin/shutdown -y -i5 -g0 >/dev/msglog 2<>/dev/msglog

 

[sstarer]

I checked /etc/inittab and found:
p3:s1234owerfail:/usr/sbin/shutdown -y -i5 -g0 >/dev/console 2<>/dev/console
again, on one line.
So there is a shutdown enabled if a powerfail event is detected. Do you think that's the cause of the shutdown? a Powerfailure?

 

[cfsalmeida]

For the line you show us of your &quot;power.conf&quot; you have &quot;autoshutdown&quot; disable, so that's not it ... sorry !
You can also check for autoshutdown configuration with &quot;dtpower&quot; on CDE gui, autoshutdown will shutdown the machine (if configured that way), if idle for a configured number of minutes, and beeteween a interval of time (also configured), i.e:

# Idle(min) Start/Finish(hh:mm) Behavior
autoshutdown 45 00:00 0:00 shutdown

This means your system will power off at any time of day after a 45-minute period of inactivity, dtpower runs for you the &quot;pmconfig&quot;, if you change &quot;/etc/power.conf&quot; by hand, you need to run also &quot;/usr/sbin/pmconfig&quot; to inform the power management framework of the new setting.
I think the &quot;/etc/inittab&quot; line is used by UPS control programs and similar, also when you pull the plug on a sun4d machine, the kernel will get an interrupt 5ms before the power goes away, If OBP had an NVRAM parameter, say &quot;powerfail-time&quot;, then the kernel could set it to the string with the time of powerfail in place of the 5ms.

sorry ... it seems to have another problem ! I only say for you to check &quot;autoshutdown&quot; because from your reports it seems if the box is busy, shutdown don't ocurr, i don't know Axil machines ... the OBP has the feature to shutdown &quot;if&quot; cpu or motherboard temperature is high (reaching a threshold), you can have a problem with fan ... just guessing here, also a crash or core data is created at non wanted shutdowns ?

Regards,

Carlos Almeida,

 

[sstarer]

First off, thanks again for all the good ideas.
I agree that the inittab is for a UPS. the box could also get an SNMP shutdown event one way or another to trigger this.

I've disconnected the serial connection, and I'm pinging the box once an hour. It's now up for 7 days, so I'm beginning to think it was shutdown by someone, possibly even myself!
I used dtpower to set the noshutdown option in power.conf after I couldn't guess the syntax
The machine seems to be stable. I'm going to see what happens over the weekend without any external connections.

Thanks again for the great suggestions.
Regards,
Steve Starer


Thanks to all contributers.