cause of reboot

Michael Powell nightrecon at hotmail.com
Mon Sep 30 18:31:41 UTC 2013


Patrick Lamaiziere wrote:
[snip]
>> 
>> I looked "last" command,
>> reboot ~ ~   AM 03.15  ~
> 
> The last time It happened (one month ago) to me it was the hard disk
> (periodic scripts read a large part of the disk).
> 
> If the disk is smart capable try a full test with smartctl
> (sysutils/smartmontools)

My gateway/firewall/mail/ids router box at home has 2 GB RAM in it, so 
normally it has enough extra room that nothing ever pushes over into swap 
with one exception: the periodic run at 0300. It is generally never more 
than just a few kilobytes, but I find it slightly surprising nonetheless.

If a sector (or more) on the drive that is backing the swap partition has 
gone bad it might not even be noticeable until something pages out to swap 
(like my 0300 periodic run). 

If the drive is a WD the 'Quick' test using the manufacturers' wddiags 
utility should spot it, and is non-destructive. I have occasionally seen the 
full test not destroy data - but I wouldn't count on it being non-
destructive. However, as long as the remap area isn't full the long test 
will repair the drive by relocating and mapping out the bad spot. When this 
silent fading away of magnetic media occurs wrt to higher-end RAID 
controllers the scrub function in the controller BIOS is where you would 
want to go.

The other problem relative to this that I've run into is the apple before 
the cart syndrome around backups. I have seen dump fail to allow for backing 
up data prior to using the full wddiags to repair a drive so you kinda get 
stuck. If the full test is going to wipe the drive and you can't generate a 
fresh current backup you're stuck only being able to restore whatever is the 
last good backup you have on hand.

Wouldn't surpise me at all if this were to turn out to be the drive just 
recently grew one or more bad spots. A bad spot or few on an old drive that 
gets repaired I might continue to use the drive for a while, maybe even for 
like a year time-frame wise. If 2 months later it starts growing more bad 
spots the drive goes in the rubbish bin.

-Mike






More information about the freebsd-questions mailing list