FC5 unplanned reboots

[thedaver]

I'd like some guesses/insight/experience on FC5 and some expected/unplanned reboots I've incurred in the last few days.

A server I lease from a dedicated server firm had been stable and up for weeks on FC5.

"Fedora Core release 5 (Bordeaux)"

Then, on June 01 and again today June 04, the server has spontaneously been restarted.

Trouble is, I have not issued any commands nor are there any scripts - to my knowledge - that would reboot this thing.

Hosting provider has no knowledge of any reboots they explicitly performed. Of course, I cannot assure their total honesty or their total awareness of things like power strip malfunctions, etc.

I'm convinced that the machine has not been hacked, and I am the only administrator/user.

There's tons of available disk free space on all partitions and only 1% of inodes used. Machine reports using 10% of available RAM, Intel 2.8Ghz cpu.

I suppose there could be someone maliciously hitting the machine with some sort of inordinantly high load on httpd or sshd (my only exposed services. But I wouldn't expect that to restart a server spontaneously.

The logs clearly show dmesg processing at the times I'm being notified by an email monitor that the machine was rebooted cleanly. I just cannot figure out what's causing the reboot.

SOOOOOOOOOOOOOOOOOOOO

Any ideas about how I can introduce more monitoring or alerts/alarms and logging in order to get to the bottom of the next unexpected reboot?

The real trick is probably trying to capture unexpected power outages. But I can tell I don't have logging in place that would account for something like a "yum update" that might require a forced reboot or something stupid that I don't know about. I'm not even sure "yum" can do that... wouldn't be very linux-like to get all M$ reboot after a patch behavior like that.

I don't have MRTG enabled on this machine, but perhaps I should contemplate that?

I'll stop rambling and will be very appreciative for any suggestions on how to understand the situation better.

Thanks!
D.




D.E.R. Management - IT Project Management Consulting

http://www.dermanagement.com/

 

[QatQat]

hi thedaver,


yum update may require a reboot but only after user (root) confirmation.

Unless you find a line about shutdown being initiated, I would not blame the ISP either; sudden shutdown, when happens, leaves no log line for obvious reasons and when restarting you will still have "clean reboot" line.
Probably they are not aware if problems on their systems.
I would notify your ISP and tell them to check power or hardware.

QatQat


If I could have sex each time I reboot my server, I would definitely prefer Windoz over Linux!

 

[thedaver]

QatQat, thanks for the note.
ISP assures me all cables are in place and checked for solid connections. We'll see ;-)

I'm going to install snmp/mrtg and collect stats.

Yes, I agree that a kernel update from yum would require a reboot to load that kernel. As I say, I didn't issue or authorize the reboot so it's quite confusing to see it happen.

Q: Where does linux/FC5 log a request for shutdown? /var/log/messages? In what form?

D.E.R. Management - IT Project Management Consulting

http://www.dermanagement.com/

 

[QatQat]

hi thedaver,

cat /var/log/messages | grep "machinename shutdown"

where machine name is your server's short name (not FQDN)
Example of log line from a server of mine called linux1

Jun 4 17:02:46 linux1 shutdown[14328]: shutting down for system reboot

Cheers

QatQat

If I could have sex each time I reboot my server, I would definitely prefer Windoz over Linux!

 

[thedaver]

ISP fessed' up that they are having power plant problems... whee! I'm not insane!

D.E.R. Management - IT Project Management Consulting

http://www.dermanagement.com/